Authors: Mohsen Gholami, Mohammad Akbari, Kevin Cannons, Yong Zhang
Abstract: In this work, we propose an extreme compression technique for Large Multimodal Models (LMMs). While previous studies have explored quantization as an efficient post-training compression method for Large Language Models (LLMs), low-bit compression for multimodal models remains under-explored. The redundant nature of inputs in multimodal models results in a highly sparse attention matrix. We theoretically and experimentally demonstrate that the attention matrix's sparsity bounds the compression error of the Query and Key weight matrices. Based on this, we introduce CASP, a model compression technique for LMMs. Our approach performs a data-aware low-rank decomposition on the Query and Key weight matrix, followed by quantization across all layers based on an optimal bit allocation process. CASP is compatible with any quantization technique and enhances state-of-the-art 2-bit quantization methods (AQLM and QuIP#) by an average of 21% on image- and video-language benchmarks.
Authors: Dominic Maggio, Luca Carlone
Abstract: Open-set semantic mapping requires (i) determining the correct granularity to represent the scene (e.g., how should objects be defined), and (ii) fusing semantic knowledge across multiple 2D observations into an overall 3D reconstruction -ideally with a high-fidelity yet low-memory footprint. While most related works bypass the first issue by grouping together primitives with similar semantics (according to some manually tuned threshold), we recognize that the object granularity is task-dependent, and develop a task-driven semantic mapping approach. To address the second issue, current practice is to average visual embedding vectors over multiple views. Instead, we show the benefits of using a probabilistic approach based on the properties of the underlying visual-language foundation model, and leveraging Bayesian updating to aggregate multiple observations of the scene. The result is Bayesian Fields, a task-driven and probabilistic approach for open-set semantic mapping. To enable high-fidelity objects and a dense scene representation, Bayesian Fields uses 3D Gaussians which we cluster into task-relevant objects, allowing for both easy 3D object extraction and reduced memory usage. We release Bayesian Fields open-source at https: //github.com/MIT-SPARK/Bayesian-Fields.
Authors: Ming Liu, Wensheng Zhang
Abstract: As video language models (VLMs) gain more applications in various scenarios, the need for robust and scalable evaluation of their performance becomes increasingly critical. The traditional human expert-based evaluation of VLMs has limitations in consistency and scalability, which sparked interest in automatic methods such as employing VLMs to evaluate VLMs. However, the reliability of VLMs as judges remains underexplored. Existing methods often rely on a single VLM as the evaluator. However, this approach can be unreliable or biased because such a model may lack the ability to fully understand the content and may have inherent biases, ultimately compromising evaluation reliability. A remedy is to apply the principle of collective thoughts, aggregating evaluations from multiple VLMs to enhance reliability. This study investigates the efficacy of such approaches, particularly when the pool of judges includes both reliable and unreliable models. Our findings reveal that incorporating collective judgments from such a mixed pool does not necessarily improve the accuracy of the final evaluation. The inclusion of less reliable judges can introduce noise, undermining the overall reliability of the outcomes. To explore the factors that impact evaluation reliability, we fine-tune an underperforming VLM judge, Video-LLaVA, and observe that improved understanding ability alone is insufficient to make VLM judges more reliable. These findings stress the limitations of collective thought approaches and highlight the need for more advanced methods that can account for the reliability of individual models. Our study promotes the development of more reliable evaluation methods for VLMs
Authors: Hongwei Yi, Tian Ye, Shitong Shao, Xuancheng Yang, Jiantong Zhao, Hanzhong Guo, Terrance Wang, Qingyu Yin, Zeke Xie, Lei Zhu, Wei Li, Michael Lingelbach, Daquan Zhou
Abstract: We present MagicInfinite, a novel diffusion Transformer (DiT) framework that overcomes traditional portrait animation limitations, delivering high-fidelity results across diverse character types-realistic humans, full-body figures, and stylized anime characters. It supports varied facial poses, including back-facing views, and animates single or multiple characters with input masks for precise speaker designation in multi-character scenes. Our approach tackles key challenges with three innovations: (1) 3D full-attention mechanisms with a sliding window denoising strategy, enabling infinite video generation with temporal coherence and visual quality across diverse character styles; (2) a two-stage curriculum learning scheme, integrating audio for lip sync, text for expressive dynamics, and reference images for identity preservation, enabling flexible multi-modal control over long sequences; and (3) region-specific masks with adaptive loss functions to balance global textual control and local audio guidance, supporting speaker-specific animations. Efficiency is enhanced via our innovative unified step and cfg distillation techniques, achieving a 20x inference speed boost over the basemodel: generating a 10 second 540x540p video in 10 seconds or 720x720p in 30 seconds on 8 H100 GPUs, without quality loss. Evaluations on our new benchmark demonstrate MagicInfinite's superiority in audio-lip synchronization, identity preservation, and motion naturalness across diverse scenarios. It is publicly available at https://www.hedra.com/, with examples at https://magicinfinite.github.io/.
URLs: https://www.hedra.com/,, https://magicinfinite.github.io/.
Authors: Md Azim Khan, Aryya Gangopadhyay, Jianwu Wang, Robert F. Erbacher
Abstract: Situational awareness applications rely heavily on real-time processing of visual and textual data to provide actionable insights. Vision language models (VLMs) have become essential tools for interpreting complex environments by connecting visual inputs with natural language descriptions. However, these models often face computational challenges, especially when required to perform efficiently in real environments. This research presents a novel vision language model (VLM) framework that leverages frequency domain transformations and low-rank adaptation (LoRA) to enhance feature extraction, scalability, and efficiency. Unlike traditional VLMs, which rely solely on spatial-domain representations, our approach incorporates Discrete Fourier Transform (DFT) based low-rank features while retaining pretrained spatial weights, enabling robust performance in noisy or low visibility scenarios. We evaluated the proposed model on caption generation and Visual Question Answering (VQA) tasks using benchmark datasets with varying levels of Gaussian noise. Quantitative results demonstrate that our model achieves evaluation metrics comparable to state-of-the-art VLMs, such as CLIP ViT-L/14 and SigLIP. Qualitative analysis further reveals that our model provides more detailed and contextually relevant responses, particularly for real-world images captured by a RealSense camera mounted on an Unmanned Ground Vehicle (UGV).
Authors: Zhenrong Wang, Qi Zheng, Sihan Ma, Maosheng Ye, Yibing Zhan, Dongjiang Li
Abstract: With the diversification of human-object interaction (HOI) applications and the success of capturing human meshes, HOI reconstruction has gained widespread attention. Existing mainstream HOI reconstruction methods often rely on explicitly modeling interactions between humans and objects. However, such a way leads to a natural conflict between 3D mesh reconstruction, which emphasizes global structure, and fine-grained contact reconstruction, which focuses on local details. To address the limitations of explicit modeling, we propose the End-to-End HOI Reconstruction Transformer with Graph-based Encoding (HOI-TG). It implicitly learns the interaction between humans and objects by leveraging self-attention mechanisms. Within the transformer architecture, we devise graph residual blocks to aggregate the topology among vertices of different spatial structures. This dual focus effectively balances global and local representations. Without bells and whistles, HOI-TG achieves state-of-the-art performance on BEHAVE and InterCap datasets. Particularly on the challenging InterCap dataset, our method improves the reconstruction results for human and object meshes by 8.9% and 8.6%, respectively.
Authors: Xiaohao Xu, Feng Xue, Xiang Li, Haowei Li, Shusheng Yang, Tianyi Zhang, Matthew Johnson-Roberson, Xiaonan Huang
Abstract: Depth ambiguity is a fundamental challenge in spatial scene understanding, especially in transparent scenes where single-depth estimates fail to capture full 3D structure. Existing models, limited to deterministic predictions, overlook real-world multi-layer depth. To address this, we introduce a paradigm shift from single-prediction to multi-hypothesis spatial foundation models. We first present \texttt{MD-3k}, a benchmark exposing depth biases in expert and foundational models through multi-layer spatial relationship labels and new metrics. To resolve depth ambiguity, we propose Laplacian Visual Prompting (LVP), a training-free spectral prompting technique that extracts hidden depth from pre-trained models via Laplacian-transformed RGB inputs. By integrating LVP-inferred depth with standard RGB-based estimates, our approach elicits multi-layer depth without model retraining. Extensive experiments validate the effectiveness of LVP in zero-shot multi-layer depth estimation, unlocking more robust and comprehensive geometry-conditioned visual generation, 3D-grounded spatial reasoning, and temporally consistent video-level depth inference. Our benchmark and code will be available at https://github.com/Xiaohao-Xu/Ambiguity-in-Space.
Authors: Yuheng Li, Yuxiang Lai, Maria Thor, Deborah Marshall, Zachary Buchwald, David S. Yu, Xiaofeng Yang
Abstract: Computed tomography (CT) is extensively used for accurate visualization and segmentation of organs and lesions. While deep learning models such as convolutional neural networks (CNNs) and vision transformers (ViTs) have significantly improved CT image analysis, their performance often declines when applied to diverse, real-world clinical data. Although foundation models offer a broader and more adaptable solution, their potential is limited due to the challenge of obtaining large-scale, voxel-level annotations for medical images. In response to these challenges, prompting-based models using visual or text prompts have emerged. Visual-prompting methods, such as the Segment Anything Model (SAM), still require significant manual input and can introduce ambiguity when applied to clinical scenarios. Instead, foundation models that use text prompts offer a more versatile and clinically relevant approach. Notably, current text-prompt models, such as the CLIP-Driven Universal Model, are limited to text prompts already encountered during training and struggle to process the complex and diverse scenarios of real-world clinical applications. Instead of fine-tuning models trained from natural imaging, we propose OpenVocabCT, a vision-language model pretrained on large-scale 3D CT images for universal text-driven segmentation. Using the large-scale CT-RATE dataset, we decompose the diagnostic reports into fine-grained, organ-level descriptions using large language models for multi-granular contrastive learning. We evaluate our OpenVocabCT on downstream segmentation tasks across nine public datasets for organ and tumor segmentation, demonstrating the superior performance of our model compared to existing methods. All code, datasets, and models will be publicly released at https://github.com/ricklisz/OpenVocabCT.
Authors: Hongtao Wang, Jiandong Liang, Lei Wang, Shuaizhe Liang, Jinping Zhu, Chunxia Zhang, Jiangshe Zhang
Abstract: Residual moveout (RMO) provides critical information for travel time tomography. The current industry-standard method for fitting RMO involves scanning high-order polynomial equations. However, this analytical approach does not accurately capture local saltation, leading to low iteration efficiency in tomographic inversion. Supervised learning-based image segmentation methods for picking can effectively capture local variations; however, they encounter challenges such as a scarcity of reliable training samples and the high complexity of post-processing. To address these issues, this study proposes a deep learning-based cascade picking method. It distinguishes accurate and robust RMOs using a segmentation network and a post-processing technique based on trend regression. Additionally, a data synthesis method is introduced, enabling the segmentation network to be trained on synthetic datasets for effective picking in field data. Furthermore, a set of metrics is proposed to quantify the quality of automatically picked RMOs. Experimental results based on both model and real data demonstrate that, compared to semblance-based methods, our approach achieves greater picking density and accuracy.
Authors: Jiaming Liu, Linghe Kong, Guihai Chen
Abstract: Segment anything model (SAM) has shown impressive general-purpose segmentation performance on natural images, but its performance on camouflaged object detection (COD) is unsatisfactory. In this paper, we propose SAM-COD that performs camouflaged object detection for RGB-D inputs. While keeping the SAM architecture intact, dual stream adapters are expanded on the image encoder to learn potential complementary information from RGB images and depth images, and fine-tune the mask decoder and its depth replica to perform dual-stream mask prediction. In practice, the dual stream adapters are embedded into the attention block of the image encoder in a parallel manner to facilitate the refinement and correction of the two types of image embeddings. To mitigate channel discrepancies arising from dual stream embeddings that do not directly interact with each other, we augment the association of dual stream embeddings using bidirectional knowledge distillation including a model distiller and a modal distiller. In addition, to predict the masks for RGB and depth attention maps, we hybridize the two types of image embeddings which are jointly learned with the prompt embeddings to update the initial prompt, and then feed them into the mask decoders to synchronize the consistency of image embeddings and prompt embeddings. Experimental results on four COD benchmarks show that our SAM-COD achieves excellent detection performance gains over SAM and achieves state-of-the-art results with a given fine-tuning paradigm.
Authors: Runze Zhang, Guoguang Du, Xiaochuan Li, Qi Jia, Liang Jin, Lu Liu, Jingjing Wang, Cong Xu, Zhenhua Guo, Yaqian Zhao, Xiaoli Gong, Rengang Li, Baoyu Fan
Abstract: Spatio-temporal consistency is a critical research topic in video generation. A qualified generated video segment must ensure plot plausibility and coherence while maintaining visual consistency of objects and scenes across varying viewpoints. Prior research, especially in open-source projects, primarily focuses on either temporal or spatial consistency, or their basic combination, such as appending a description of a camera movement after a prompt without constraining the outcomes of this movement. However, camera movement may introduce new objects to the scene or eliminate existing ones, thereby overlaying and affecting the preceding narrative. Especially in videos with numerous camera movements, the interplay between multiple plots becomes increasingly complex. This paper introduces and examines integral spatio-temporal consistency, considering the synergy between plot progression and camera techniques, and the long-term impact of prior content on subsequent generation. Our research encompasses dataset construction through to the development of the model. Initially, we constructed a DropletVideo-10M dataset, which comprises 10 million videos featuring dynamic camera motion and object actions. Each video is annotated with an average caption of 206 words, detailing various camera movements and plot developments. Following this, we developed and trained the DropletVideo model, which excels in preserving spatio-temporal coherence during video generation. The DropletVideo dataset and model are accessible at https://dropletx.github.io.
Authors: Weixi Zheng, Aoling Huang. Jingping Yuan, Haoyu Zhao, Zhou Zhao, Yongchao Xu, Thierry G\'eraud
Abstract: In histopathology, intelligent diagnosis of Whole Slide Images (WSIs) is essential for automating and objectifying diagnoses, reducing the workload of pathologists. However, diagnostic models often face the challenge of forgetting previously learned data during incremental training on datasets from different sources. To address this issue, we propose a new framework PaGMIL to mitigate catastrophic forgetting in breast cancer WSI classification. Our framework introduces two key components into the common MIL model architecture. First, it leverages microscopic pathological prior to select more accurate and diverse representative patches for MIL. Secondly, it trains separate classification heads for each task and uses macroscopic pathological prior knowledge, treating the thumbnail as a prompt guide (PG) to select the appropriate classification head. We evaluate the continual learning performance of PaGMIL across several public breast cancer datasets. PaGMIL achieves a better balance between the performance of the current task and the retention of previous tasks, outperforming other continual learning methods. Our code will be open-sourced upon acceptance.
Authors: Junyan Lin, Haoran Chen, Yue Fan, Yingqi Fan, Xin Jin, Hui Su, Jinlan Fu, Xiaoyu Shen
Abstract: Multimodal Large Language Models (MLLMs) have made significant advancements in recent years, with visual features playing an increasingly critical role in enhancing model performance. However, the integration of multi-layer visual features in MLLMs remains underexplored, particularly with regard to optimal layer selection and fusion strategies. Existing methods often rely on arbitrary design choices, leading to suboptimal outcomes. In this paper, we systematically investigate two core aspects of multi-layer visual feature fusion: (1) selecting the most effective visual layers and (2) identifying the best fusion approach with the language model. Our experiments reveal that while combining visual features from multiple stages improves generalization, incorporating additional features from the same stage typically leads to diminished performance. Furthermore, we find that direct fusion of multi-layer visual features at the input stage consistently yields superior and more stable performance across various configurations. We make all our code publicly available: https://github.com/EIT-NLP/Layer_Select_Fuse_for_MLLM.
URLs: https://github.com/EIT-NLP/Layer_Select_Fuse_for_MLLM.
Authors: Wenzhuo Du, Gerun Wang, Guancheng Chen, Hang Zhao, Xin Li, Jian Gao
Abstract: With the exponential growth of user-generated content on video-sharing platforms, the challenge of facilitating efficient searching and browsing of videos has garnered significant attention. To enhance users' ability to swiftly locate and review pertinent videos, the creation of concise and informative video summaries has become increasingly important. Video-llama is an effective tool for generating video summarization, but it cannot effectively unify and optimize the modeling of temporal and spatial features and requires a lot of computational resources and time. Therefore, we propose MiLoRA-ViSum to more efficiently capture complex temporal dynamics and spatial relationships inherent in video data and to control the number of parameters for training. By extending traditional Low-Rank Adaptation (LoRA) into a sophisticated mixture-of-experts paradigm, MiLoRA-ViSum incorporates a dual temporal-spatial adaptation mechanism tailored specifically for video summarization tasks. This approach dynamically integrates specialized LoRA experts, each fine-tuned to address distinct temporal or spatial dimensions. Extensive evaluations of the VideoXum and ActivityNet datasets demonstrate that MiLoRA-ViSum achieves the best summarization performance compared to state-of-the-art models, while maintaining significantly lower computational costs. The proposed mixture-of-experts strategy, combined with the dual adaptation mechanism, highlights the model's potential to enhance video summarization capabilities, particularly in large-scale applications requiring both efficiency and precision.
Authors: Hangyu Du, Chee-Meng Chew
Abstract: In recent years, fully differentiable end-to-end autonomous driving systems have become a research hotspot in the field of intelligent transportation. Among various research directions, automatic parking is particularly critical as it aims to enable precise vehicle parking in complex environments. In this paper, we present a purely vision-based transformer model for end-to-end automatic parking, trained using expert trajectories. Given camera-captured data as input, the proposed model directly outputs future trajectory coordinates. Experimental results demonstrate that the various errors of our model have decreased by approximately 50% in comparison with the current state-of-the-art end-to-end trajectory prediction algorithm of the same type. Our approach thus provides an effective solution for fully differentiable automatic parking.
Authors: Yubin Wang, Xinyang Jiang, De Cheng, Xiangqian Zhao, Zilong Wang, Dongsheng Li, Cairong Zhao
Abstract: Visual prompt tuning offers significant advantages for adapting pre-trained visual foundation models to specific tasks. However, current research provides limited insight into the interpretability of this approach, which is essential for enhancing AI reliability and enabling AI-driven knowledge discovery. In this paper, rather than learning abstract prompt embeddings, we propose the first framework, named Interpretable Visual Prompt Tuning (IVPT), to explore interpretability for visual prompts, by introducing hierarchical concept prototypes. Specifically, visual prompts are linked to human-understandable semantic concepts, represented as a set of category-agnostic prototypes, each corresponding to a specific region of the image. Then, IVPT aggregates features from these regions to generate interpretable prompts, which are structured hierarchically to explain visual prompts at different granularities. Comprehensive qualitative and quantitative evaluations on fine-grained classification benchmarks show its superior interpretability and performance over conventional visual prompt tuning methods and existing interpretable methods.
Authors: David C. Jeong, Aditya Puranik, James Vong, Vrushabh Abhijit Deogirikar, Ryan Fell, Julianna Dietrich, Maria Kyrarini, Christopher Kitts
Abstract: Egocentric human body estimation allows for the inference of user body pose and shape from a wearable camera's first-person perspective. Although research has used pose estimation techniques to overcome self-occlusions and image distortions caused by head-mounted fisheye images, similar advances in 3D human mesh recovery (HMR) techniques have been limited. We introduce Fish2Mesh, a fisheye-aware transformer-based model designed for 3D egocentric human mesh recovery. We propose an egocentric position embedding block to generate an ego-specific position table for the Swin Transformer to reduce fisheye image distortion. Our model utilizes multi-task heads for SMPL parametric regression and camera translations, estimating 3D and 2D joints as auxiliary loss to support model training. To address the scarcity of egocentric camera data, we create a training dataset by employing the pre-trained 4D-Human model and third-person cameras for weak supervision. Our experiments demonstrate that Fish2Mesh outperforms previous state-of-the-art 3D HMR models.
Authors: Lunchen Xie, Eugenio Lomurno, Matteo Gambella, Danilo Ardagna, Manual Roveri, Matteo Matteucci, Qingjiang Shi
Abstract: Differentiable Neural Architecture Search (NAS) provides a promising avenue for automating the complex design of deep learning (DL) models. However, current differentiable NAS methods often face constraints in efficiency, operation selection, and adaptability under varying resource limitations. We introduce ZO-DARTS++, a novel NAS method that effectively balances performance and resource constraints. By integrating a zeroth-order approximation for efficient gradient handling, employing a sparsemax function with temperature annealing for clearer and more interpretable architecture distributions, and adopting a size-variable search scheme for generating compact yet accurate architectures, ZO-DARTS++ establishes a new balance between model complexity and performance. In extensive tests on medical imaging datasets, ZO-DARTS++ improves the average accuracy by up to 1.8\% over standard DARTS-based methods and shortens search time by approximately 38.6\%. Additionally, its resource-constrained variants can reduce the number of parameters by more than 35\% while maintaining competitive accuracy levels. Thus, ZO-DARTS++ offers a versatile and efficient framework for generating high-quality, resource-aware DL models suitable for real-world medical applications.
Authors: Yong He, Hongshan Yu, Mingtao Feng, Tongjia Chen, Zechuan Li, Anwaar Ulhaq, Saeed Anwar, Ajmal Saeed Mian
Abstract: Diffusion probabilistic models are traditionally used to generate colors at fixed pixel positions in 2D images. Building on this, we extend diffusion models to point cloud semantic segmentation, where point positions also remain fixed, and the diffusion model generates point labels instead of colors. To accelerate the denoising process in reverse diffusion, we introduce a noisy label embedding mechanism. This approach integrates semantic information into the noisy label, providing an initial semantic reference that improves the reverse diffusion efficiency. Additionally, we propose a point frequency transformer that enhances the adjustment of high-level context in point clouds. To reduce computational complexity, we introduce the position condition into MLP and propose denoising PointNet to process the high-resolution point cloud without sacrificing geometric details. Finally, we integrate the proposed noisy label embedding, point frequency transformer and denoising PointNet in our proposed dual conditional diffusion model-based network (PointDiffuse) to perform large-scale point cloud semantic segmentation. Extensive experiments on five benchmarks demonstrate the superiority of PointDiffuse, achieving the state-of-the-art mIoU of 74.2\% on S3DIS Area 5, 81.2\% on S3DIS 6-fold and 64.8\% on SWAN dataset.
Authors: Xianjie Liu, Keren Fu, Qijun Zhao
Abstract: Dichotomous Image Segmentation (DIS) is a high-precision object segmentation task for high-resolution natural images. The current mainstream methods focus on the optimization of local details but overlook the fundamental challenge of modeling the integrity of objects. We have found that the depth integrity-prior implicit in the the pseudo-depth maps generated by Depth Anything Model v2 and the local detail features of image patches can jointly address the above dilemmas. Based on the above findings, we have designed a novel Patch-Depth Fusion Network (PDFNet) for high-precision dichotomous image segmentation. The core of PDFNet consists of three aspects. Firstly, the object perception is enhanced through multi-modal input fusion. By utilizing the patch fine-grained strategy, coupled with patch selection and enhancement, the sensitivity to details is improved. Secondly, by leveraging the depth integrity-prior distributed in the depth maps, we propose an integrity-prior loss to enhance the uniformity of the segmentation results in the depth maps. Finally, we utilize the features of the shared encoder and, through a simple depth refinement decoder, improve the ability of the shared encoder to capture subtle depth-related information in the images. Experiments on the DIS-5K dataset show that PDFNet significantly outperforms state-of-the-art non-diffusion methods. Due to the incorporation of the depth integrity-prior, PDFNet achieves or even surpassing the performance of the latest diffusion-based methods while using less than 11% of the parameters of diffusion-based methods. The source code at https://github.com/Tennine2077/PDFNet.
Authors: Syed Sajid Ullah, Li Gang, Mudassir Riaz, Ahsan Ashfaq, Salman Khan, Sajawal Khan
Abstract: Handwritten digit recognition remains a fundamental challenge in computer vision, with applications ranging from postal code reading to document digitization. This paper presents an ensemble-based approach that combines Convolutional Neural Networks (CNNs) with traditional machine learning techniques to improve recognition accuracy and robustness. We evaluate our method on the MNIST dataset, comprising 70,000 handwritten digit images. Our hybrid model, which uses CNNs for feature extraction and Support Vector Machines (SVMs) for classification, achieves an accuracy of 99.30%. We also explore the effectiveness of data augmentation and various ensemble techniques in enhancing model performance. Our results demonstrate that this approach not only achieves high accuracy but also shows improved generalization across diverse handwriting styles. The findings contribute to the development of more reliable handwritten digit recognition systems and highlight the potential of combining deep learning with traditional machine learning methods in pattern recognition tasks.
Authors: Kuanghong Liu, Jin Wang, Kangjian He, Dan Xu, Xuejie Zhang
Abstract: Conventional multi-source domain few-shot adaptation (MFDA) faces the challenge of further reducing the load on edge-side devices in low-resource scenarios. Considering the native language-supervised advantage of CLIP and the plug-and-play nature of prompt to transfer CLIP efficiently, this paper introduces an uploadable multi-source few-shot domain adaptation (UMFDA) schema. It belongs to a decentralized edge collaborative learning in the edge-side models that must maintain a low computational load. And only a limited amount of annotations in source domain data is provided, with most of the data being unannotated. Further, this paper proposes a vision-aware multimodal prompt tuning framework (VAMP) under the decentralized schema, where the vision-aware prompt guides the text domain-specific prompt to maintain semantic discriminability and perceive the domain information. The cross-modal semantic and domain distribution alignment losses optimize each edge-side model, while text classifier consistency and semantic diversity losses promote collaborative learning among edge-side models. Extensive experiments were conducted on OfficeHome and DomainNet datasets to demonstrate the effectiveness of the proposed VAMP in the UMFDA, which outperformed the previous prompt tuning methods.
Authors: Akshat Jain
Abstract: This paper presents a novel approach to image dehazing by combining Feature Fusion Attention (FFA) networks with CycleGAN architecture. Our method leverages both supervised and unsupervised learning techniques to effectively remove haze from images while preserving crucial image details. The proposed hybrid architecture demonstrates significant improvements in image quality metrics, achieving superior PSNR and SSIM scores compared to traditional dehazing methods. Through extensive experimentation on the RESIDE and DenseHaze CVPR 2019 dataset, we show that our approach effectively handles both synthetic and real-world hazy images. CycleGAN handles the unpaired nature of hazy and clean images effectively, enabling the model to learn mappings even without paired data.
Authors: Hongjia Zhai, Boming Zhao, Hai Li, Xiaokun Pan, Yijia He, Zhaopeng Cui, Hujun Bao, Guofeng Zhang
Abstract: Recently, neural radiance fields (NeRF) have gained significant attention in the field of visual localization. However, existing NeRF-based approaches either lack geometric constraints or require extensive storage for feature matching, limiting their practical applications. To address these challenges, we propose an efficient and novel visual localization approach based on the neural implicit map with complementary features. Specifically, to enforce geometric constraints and reduce storage requirements, we implicitly learn a 3D keypoint descriptor field, avoiding the need to explicitly store point-wise features. To further address the semantic ambiguity of descriptors, we introduce additional semantic contextual feature fields, which enhance the quality and reliability of 2D-3D correspondences. Besides, we propose descriptor similarity distribution alignment to minimize the domain gap between 2D and 3D feature spaces during matching. Finally, we construct the matching graph using both complementary descriptors and contextual features to establish accurate 2D-3D correspondences for 6-DoF pose estimation. Compared with the recent NeRF-based approaches, our method achieves a 3$\times$ faster training speed and a 45$\times$ reduction in model storage. Extensive experiments on two widely used datasets demonstrate that our approach outperforms or is highly competitive with other state-of-the-art NeRF-based visual localization methods. Project page: \href{https://zju3dv.github.io/neuraloc}{https://zju3dv.github.io/neuraloc}
URLs: https://zju3dv.github.io/neuraloc, https://zju3dv.github.io/neuraloc
Authors: Xuanyu Zhang, Jiarui Meng, Zhipei Xu, Shuzhou Yang, Yanmin Wu, Ronggang Wang, Jian Zhang
Abstract: 3D Gaussian Splatting (3DGS) has emerged as a premier method for 3D representation due to its real-time rendering and high-quality outputs, underscoring the critical need to protect the privacy of 3D assets. Traditional NeRF steganography methods fail to address the explicit nature of 3DGS since its point cloud files are publicly accessible. Existing GS steganography solutions mitigate some issues but still struggle with reduced rendering fidelity, increased computational demands, and security flaws, especially in the security of the geometric structure of the visualized point cloud. To address these demands, we propose a SecureGS, a secure and efficient 3DGS steganography framework inspired by Scaffold-GS's anchor point design and neural decoding. SecureGS uses a hybrid decoupled Gaussian encryption mechanism to embed offsets, scales, rotations, and RGB attributes of the hidden 3D Gaussian points in anchor point features, retrievable only by authorized users through privacy-preserving neural networks. To further enhance security, we propose a density region-aware anchor growing and pruning strategy that adaptively locates optimal hiding regions without exposing hidden information. Extensive experiments show that SecureGS significantly surpasses existing GS steganography methods in rendering fidelity, speed, and security.
Authors: Jiebin Yan, Kangcheng Wu, Junjie Chen, Ziwen Tan, Yuming Fang
Abstract: Most of existing blind omnidirectional image quality assessment (BOIQA) models rely on viewport generation by modeling user viewing behavior or transforming omnidirectional images (OIs) into varying formats; however, these methods are either computationally expensive or less scalable. To solve these issues, in this paper, we present a flexible and effective paradigm, which is viewport-unaware and can be easily adapted to 2D plane image quality assessment (2D-IQA). Specifically, the proposed BOIQA model includes an adaptive prior-equator sampling module for extracting a patch sequence from the equirectangular projection (ERP) image in a resolution-agnostic manner, a progressive deformation-unaware feature fusion module which is able to capture patch-wise quality degradation in a deformation-immune way, and a local-to-global quality aggregation module to adaptively map local perception to global quality. Extensive experiments across four OIQA databases (including uniformly distorted OIs and non-uniformly distorted OIs) demonstrate that the proposed model achieves competitive performance with low complexity against other state-of-the-art models, and we also verify its adaptive capacity to 2D-IQA.
Authors: Xiangxiang Chu, Renda Li, Yong Wang
Abstract: Recent studies have highlighted the interplay between diffusion models and representation learning. Intermediate representations from diffusion models can be leveraged for downstream visual tasks, while self-supervised vision models can enhance the convergence and generation quality of diffusion models. However, transferring pretrained weights from vision models to diffusion models is challenging due to input mismatches and the use of latent spaces. To address these challenges, we propose Unified Self-supervised Pretraining (USP), a framework that initializes diffusion models via masked latent modeling in a Variational Autoencoder (VAE) latent space. USP achieves comparable performance in understanding tasks while significantly improving the convergence speed and generation quality of diffusion models. Our code will be publicly available at https://github.com/cxxgtxy/USP.
Authors: Jian Ma, Qirong Peng, Xu Guo, Chen Chen, Haonan Lu, Zhenyu Yang
Abstract: Text-to-image (T2I) models are well known for their ability to produce highly realistic images, while multimodal large language models (MLLMs) are renowned for their proficiency in understanding and integrating multiple modalities. However, currently there is no straightforward and efficient framework to transfer the multimodal comprehension abilities of MLLMs to T2I models to enable them to understand multimodal inputs. In this paper, we propose the X2I framework, which endows Diffusion Transformer (DiT) models with the capability to comprehend various modalities, including multilingual text, screenshot documents, images, videos, and audio. X2I is trained using merely 100K English corpus with 160 GPU hours. Building on the DiT teacher model, we adopt an innovative distillation method to extract the inference capabilities of the teacher model and design a lightweight AlignNet structure to serve as an intermediate bridge. Compared to the teacher model, X2I shows a decrease in performance degradation of less than 1\% while gaining various multimodal understanding abilities, including multilingual to image, image to image, image-text to image, video to image, audio to image, and utilizing creative fusion to enhance imagery. Furthermore, it is applicable for LoRA training in the context of image-text to image generation, filling a void in the industry in this area. We further design a simple LightControl to enhance the fidelity of instructional image editing. Finally, extensive experiments demonstrate the effectiveness, efficiency, multifunctionality, and transferability of our X2I. The open-source code and checkpoints for X2I can be found at the following link: https://github.com/OPPO-Mente-Lab/X2I.
Authors: Ye Tao, Jiawei Zhang, Yahao Shi, Dongqing Zou, Bin Zhou
Abstract: Image-based 3D generation has vast applications in robotics and gaming, where high-quality, diverse outputs and consistent 3D representations are crucial. However, existing methods have limitations: 3D diffusion models are limited by dataset scarcity and the absence of strong pre-trained priors, while 2D diffusion-based approaches struggle with geometric consistency. We propose a method that leverages 2D diffusion models' implicit 3D reasoning ability while ensuring 3D consistency via Gaussian-splatting-based geometric distillation. Specifically, the proposed Gaussian Splatting Decoder enforces 3D consistency by transforming SV3D latent outputs into an explicit 3D representation. Unlike SV3D, which only relies on implicit 2D representations for video generation, Gaussian Splatting explicitly encodes spatial and appearance attributes, enabling multi-view consistency through geometric constraints. These constraints correct view inconsistencies, ensuring robust geometric consistency. As a result, our approach simultaneously generates high-quality, multi-view-consistent images and accurate 3D models, providing a scalable solution for single-image-based 3D generation and bridging the gap between 2D Diffusion diversity and 3D structural coherence. Experimental results demonstrate state-of-the-art multi-view consistency and strong generalization across diverse datasets. The code will be made publicly available upon acceptance.
Authors: Bohan Liu, Xiaosen Wang
Abstract: Transfer-based attacks pose a significant threat to real-world applications by directly targeting victim models with adversarial examples generated on surrogate models. While numerous approaches have been proposed to enhance adversarial transferability, existing works often overlook the intrinsic relationship between adversarial perturbations and input images. In this work, we find that adversarial perturbation often exhibits poor translation invariance for a given clean image and model, which is attributed to local invariance. Through empirical analysis, we demonstrate that there is a positive correlation between the local invariance of adversarial perturbations w.r.t. the input image and their transferability across different models. Based on this finding, we propose a general adversarial transferability boosting technique called Local Invariance Boosting approach (LI-Boost). Extensive experiments on the standard ImageNet dataset demonstrate that LI-Boost could significantly boost various types of transfer-based attacks (e.g., gradient-based, input transformation-based, model-related, advanced objective function, ensemble, etc.) on CNNs, ViTs, and defense mechanisms. Our approach presents a promising direction for future research in improving adversarial transferability across different models.
Authors: Mingxing Li, Rui Wang, Lei Sun, Yancheng Bai, Xiangxiang Chu
Abstract: The rapid expansion of mobile internet has resulted in a substantial increase in user-generated content (UGC) images, thereby making the thorough assessment of UGC images both urgent and essential. Recently, multimodal large language models (MLLMs) have shown great potential in image quality assessment (IQA) and image aesthetic assessment (IAA). Despite this progress, effectively scoring the quality and aesthetics of UGC images still faces two main challenges: 1) A single score is inadequate to capture the hierarchical human perception. 2) How to use MLLMs to output numerical scores, such as mean opinion scores (MOS), remains an open question. To address these challenges, we introduce a novel dataset, named Realistic image Quality and Aesthetic (RealQA), including 14,715 UGC images, each of which is annoted with 10 fine-grained attributes. These attributes span three levels: low level (e.g., image clarity), middle level (e.g., subject integrity) and high level (e.g., composition). Besides, we conduct a series of in-depth and comprehensive investigations into how to effectively predict numerical scores using MLLMs. Surprisingly, by predicting just two extra significant digits, the next token paradigm can achieve SOTA performance. Furthermore, with the help of chain of thought (CoT) combined with the learnt fine-grained attributes, the proposed method can outperform SOTA methods on five public datasets for IQA and IAA with superior interpretability and show strong zero-shot generalization for video quality assessment (VQA). The code and dataset will be released.
Authors: Xinan He, Yue Zhou, Bing Fan, Bin Li, Guopu Zhu, Feng Ding
Abstract: Faces synthesized by diffusion models (DMs) with high-quality and controllable attributes pose a significant challenge for Deepfake detection. Most state-of-the-art detectors only yield a binary decision, incapable of forgery localization, attribution of forgery methods, and providing analysis on the cause of forgeries. In this work, we integrate Multimodal Large Language Models (MLLMs) within DM-based face forensics, and propose a fine-grained analysis triad framework called VLForgery, that can 1) predict falsified facial images; 2) locate the falsified face regions subjected to partial synthesis; and 3) attribute the synthesis with specific generators. To achieve the above goals, we introduce VLF (Visual Language Forensics), a novel and diverse synthesis face dataset designed to facilitate rich interactions between Visual and Language modalities in MLLMs. Additionally, we propose an extrinsic knowledge-guided description method, termed EkCot, which leverages knowledge from the image generation pipeline to enable MLLMs to quickly capture image content. Furthermore, we introduce a low-level vision comparison pipeline designed to identify differential features between real and fake that MLLMs can inherently understand. These features are then incorporated into EkCot, enhancing its ability to analyze forgeries in a structured manner, following the sequence of detection, localization, and attribution. Extensive experiments demonstrate that VLForgery outperforms other state-of-the-art forensic approaches in detection accuracy, with additional potential for falsified region localization and attribution analysis.
Authors: Ziyue Huang, Yongchao Feng, Shuai Yang, Ziqi Liu, Qingjie Liu, Yunhong Wang
Abstract: Remote sensing object detection has made significant progress, but most studies still focus on closed-set detection, limiting generalization across diverse datasets. Open-vocabulary object detection (OVD) provides a solution by leveraging multimodal associations between text prompts and visual features. However, existing OVD methods for remote sensing (RS) images are constrained by small-scale datasets and fail to address the unique challenges of remote sensing interpretation, include oriented object detection and the need for both high precision and real-time performance in diverse scenarios. To tackle these challenges, we propose OpenRSD, a universal open-prompt RS object detection framework. OpenRSD supports multimodal prompts and integrates multi-task detection heads to balance accuracy and real-time requirements. Additionally, we design a multi-stage training pipeline to enhance the generalization of model. Evaluated on seven public datasets, OpenRSD demonstrates superior performance in oriented and horizontal bounding box detection, with real-time inference capabilities suitable for large-scale RS image analysis. Compared to YOLO-World, OpenRSD exhibits an 8.7\% higher average precision and achieves an inference speed of 20.8 FPS. Codes and models will be released.
Authors: Zixi Kang, Xinghan Wang, Yadong Mu
Abstract: Human motion generation holds significant promise in fields such as animation, film production, and robotics. However, existing methods often fail to produce physically plausible movements that adhere to biomechanical principles. While recent autoregressive and diffusion models have improved visual quality, they frequently overlook essential biodynamic features, such as muscle activation patterns and joint coordination, leading to motions that either violate physical laws or lack controllability. This paper introduces BioMoDiffuse, a novel biomechanics-aware diffusion framework that addresses these limitations. It features three key innovations: (1) A lightweight biodynamic network that integrates muscle electromyography (EMG) signals and kinematic features with acceleration constraints, (2) A physics-guided diffusion process that incorporates real-time biomechanical verification via modified Euler-Lagrange equations, and (3) A decoupled control mechanism that allows independent regulation of motion speed and semantic context. We also propose a set of comprehensive evaluation protocols that combines traditional metrics (FID, R-precision, etc.) with new biomechanical criteria (smoothness, foot sliding, floating, etc.). Our approach bridges the gap between data-driven motion synthesis and biomechanical authenticity, establishing new benchmarks for physically accurate motion generation.
Authors: Zidu Wang, Jiankuo Zhao, Miao Xu, Xiangyu Zhu, Zhen Lei
Abstract: 3D Morphable Models (3DMMs) have played a pivotal role as a fundamental representation or initialization for 3D avatar animation and reconstruction. However, extending 3DMMs to hair remains challenging due to the difficulty of enforcing vertex-level consistent semantic meaning across hair shapes. This paper introduces a novel method, Semantic-consistent Ray Modeling of Hair (SRM-Hair), for making 3D hair morphable and controlled by coefficients. The key contribution lies in semantic-consistent ray modeling, which extracts ordered hair surface vertices and exhibits notable properties such as additivity for hairstyle fusion, adaptability, flipping, and thickness modification. We collect a dataset of over 250 high-fidelity real hair scans paired with 3D face data to serve as a prior for the 3D morphable hair. Based on this, SRM-Hair can reconstruct a hair mesh combined with a 3D head from a single image. Note that SRM-Hair produces an independent hair mesh, facilitating applications in virtual avatar creation, realistic animation, and high-fidelity hair rendering. Both quantitative and qualitative experiments demonstrate that SRM-Hair achieves state-of-the-art performance in 3D mesh reconstruction. Our project is available at https://github.com/wang-zidu/SRM-Hair
Authors: Baining Zhao, Jianjie Fang, Zichao Dai, Ziyou Wang, Jirong Zha, Weichen Zhang, Chen Gao, Yue Wang, Jinqiang Cui, Xinlei Chen, Yong Li
Abstract: Large multimodal models exhibit remarkable intelligence, yet their embodied cognitive abilities during motion in open-ended urban 3D space remain to be explored. We introduce a benchmark to evaluate whether video-large language models (Video-LLMs) can naturally process continuous first-person visual observations like humans, enabling recall, perception, reasoning, and navigation. We have manually control drones to collect 3D embodied motion video data from real-world cities and simulated environments, resulting in 1.5k video clips. Then we design a pipeline to generate 5.2k multiple-choice questions. Evaluations of 17 widely-used Video-LLMs reveal current limitations in urban embodied cognition. Correlation analysis provides insight into the relationships between different tasks, showing that causal reasoning has a strong correlation with recall, perception, and navigation, while the abilities for counterfactual and associative reasoning exhibit lower correlation with other tasks. We also validate the potential for Sim-to-Real transfer in urban embodiment through fine-tuning.
Authors: Kai Li, Junhao Wang, William Han, Ding Zhao
Abstract: Minimally invasive surgery (MIS) has transformed clinical practice by reducing recovery times, minimizing complications, and enhancing precision. Nonetheless, MIS inherently relies on indirect visualization and precise instrument control, posing unique challenges. Recent advances in artificial intelligence have enabled real-time surgical scene understanding through techniques such as image classification, object detection, and segmentation, with scene reconstruction emerging as a key element for enhanced intraoperative guidance. Although neural radiance fields (NeRFs) have been explored for this purpose, their substantial data requirements and slow rendering inhibit real-time performance. In contrast, 3D Gaussian Splatting (3DGS) offers a more efficient alternative, achieving state-of-the-art performance in dynamic surgical scene reconstruction. In this work, we introduce Feature-EndoGaussian (FEG), an extension of 3DGS that integrates 2D segmentation cues into 3D rendering to enable real-time semantic and scene reconstruction. By leveraging pretrained segmentation foundation models, FEG incorporates semantic feature distillation within the Gaussian deformation framework, thereby enhancing both reconstruction fidelity and segmentation accuracy. On the EndoNeRF dataset, FEG achieves superior performance (SSIM of 0.97, PSNR of 39.08, and LPIPS of 0.03) compared to leading methods. Additionally, on the EndoVis18 dataset, FEG demonstrates competitive class-wise segmentation metrics while balancing model size and real-time performance.
Authors: Li Li, Jiashu Qu, Yuxiao Zhou, Yuehan Qin, Tiankai Yang, Yue Zhao
Abstract: Vision-Language Models (VLMs) have advanced multi-modal tasks like image captioning, visual question answering, and reasoning. However, they often generate hallucinated outputs inconsistent with the visual context or prompt, limiting reliability in critical applications like autonomous driving and medical imaging. Existing studies link hallucination to statistical biases, language priors, and biased feature learning but lack a structured causal understanding. In this work, we introduce a causal perspective to analyze and mitigate hallucination in VLMs. We hypothesize that hallucination arises from unintended direct influences of either the vision or text modality, bypassing proper multi-modal fusion. To address this, we construct a causal graph for VLMs and employ counterfactual analysis to estimate the Natural Direct Effect (NDE) of vision, text, and their cross-modal interaction on the output. We systematically identify and mitigate these unintended direct effects to ensure that responses are primarily driven by genuine multi-modal fusion. Our approach consists of three steps: (1) designing structural causal graphs to distinguish correct fusion pathways from spurious modality shortcuts, (2) estimating modality-specific and cross-modal NDE using perturbed image representations, hallucinated text embeddings, and degraded visual inputs, and (3) implementing a test-time intervention module to dynamically adjust the model's dependence on each modality. Experimental results demonstrate that our method significantly reduces hallucination while preserving task performance, providing a robust and interpretable framework for improving VLM reliability. To enhance accessibility and reproducibility, our code is publicly available at https://github.com/TREE985/Treble-Counterfactual-VLMs.
URLs: https://github.com/TREE985/Treble-Counterfactual-VLMs.
Authors: Wongi Park, Myeongseok Nam, Siwon Kim, Sangwoo Jo, Soomok Lee
Abstract: Recently, 3D Gaussian Splatting (3D-GS) has emerged, showing real-time rendering speeds and high-quality results in static scenes. Although 3D-GS shows effectiveness in static scenes, their performance significantly degrades in real-world environments due to transient objects, lighting variations, and diverse levels of occlusion. To tackle this, existing methods estimate occluders or transient elements by leveraging pre-trained models or integrating additional transient field pipelines. However, these methods still suffer from two defects: 1) Using semantic features from the Vision Foundation model (VFM) causes additional computational costs. 2) The transient field requires significant memory to handle transient elements with per-view Gaussians and struggles to define clear boundaries for occluders, solely relying on photometric errors. To address these problems, we propose ForestSplats, a novel approach that leverages the deformable transient field and a superpixel-aware mask to efficiently represent transient elements in the 2D scene across unconstrained image collections and effectively decompose static scenes from transient distractors without VFM. We designed the transient field to be deformable, capturing per-view transient elements. Furthermore, we introduce a superpixel-aware mask that clearly defines the boundaries of occluders by considering photometric errors and superpixels. Additionally, we propose uncertainty-aware densification to avoid generating Gaussians within the boundaries of occluders during densification. Through extensive experiments across several benchmark datasets, we demonstrate that ForestSplats outperforms existing methods without VFM and shows significant memory efficiency in representing transient elements.
Authors: Antonio Alliegro, Francesca Pistilli, Tatiana Tommasi, Giuseppe Averta
Abstract: Forecasting human-environment interactions in daily activities is challenging due to the high variability of human behavior. While predicting directly from videos is possible, it is limited by confounding factors like irrelevant objects or background noise that do not contribute to the interaction. A promising alternative is using Scene Graphs (SGs) to track only the relevant elements. However, current methods for forecasting future SGs face significant challenges and often rely on unrealistic assumptions, such as fixed objects over time, limiting their applicability to long-term activities where interacted objects may appear or disappear. In this paper, we introduce FORESCENE, a novel framework for Scene Graph Anticipation (SGA) that predicts both object and relationship evolution over time. FORESCENE encodes observed video segments into a latent representation using a tailored Graph Auto-Encoder and forecasts future SGs using a Latent Diffusion Model (LDM). Our approach enables continuous prediction of interaction dynamics without making assumptions on the graph's content or structure. We evaluate FORESCENE on the Action Genome dataset, where it outperforms existing SGA methods while solving a significantly more complex task.
Authors: Xiang Gao, Shuai Yang, Jiaying Liu
Abstract: Optical illusion hidden picture is an interesting visual perceptual phenomenon where an image is cleverly integrated into another picture in a way that is not immediately obvious to the viewer. Established on the off-the-shelf text-to-image (T2I) diffusion model, we propose a novel training-free text-guided image-to-image (I2I) translation framework dubbed as \textbf{P}hase-\textbf{T}ransferred \textbf{Diffusion} Model (PTDiffusion) for hidden art syntheses. PTDiffusion embeds an input reference image into arbitrary scenes as described by the text prompts, while exhibiting hidden visual cues of the reference image. At the heart of our method is a plug-and-play phase transfer mechanism that dynamically and progressively transplants diffusion features' phase spectrum from the denoising process to reconstruct the reference image into the one to sample the generated illusion image, realizing harmonious fusion of the reference structural information and the textual semantic information. Furthermore, we propose asynchronous phase transfer to enable flexible control to the degree of hidden content discernability. Our method bypasses any model training and fine-tuning, all while substantially outperforming related methods in image quality, text fidelity, visual discernibility, and contextual naturalness for illusion picture synthesis, as demonstrated by extensive qualitative and quantitative experiments.
Authors: Si Zhou, Yain-Whar Si, Xiaochen Yuan, Xiaofan Li, Xiaoxiang Liu, Xinyuan Zhang, Cong Lin, Xueyuan Gong
Abstract: In Neural Networks, there are various methods of feature fusion. Different strategies can significantly affect the effectiveness of feature representation, consequently influencing the ability of model to extract representative and discriminative features. In the field of face recognition, traditional feature fusion methods include feature concatenation and feature addition. Recently, various attention mechanism-based fusion strategies have emerged. However, we found that these methods primarily focus on the important features in the image, referred to as salient features in this paper, while neglecting another equally important set of features for image recognition tasks, which we term differential features. This may cause the model to overlook critical local differences when dealing with complex facial samples. Therefore, in this paper, we propose an efficient convolution module called MSConv (Multiplicative and Subtractive Convolution), designed to balance the learning of model about salient and differential features. Specifically, we employ multi-scale mixed convolution to capture both local and broader contextual information from face images, and then utilize Multiplication Operation (MO) and Subtraction Operation (SO) to extract salient and differential features, respectively. Experimental results demonstrate that by integrating both salient and differential features, MSConv outperforms models that only focus on salient features.
Authors: Shashata Sawmya, Thomas L. Athey, Gwyneth Liu, Nir Shavit
Abstract: Training segmentation models from scratch has been the standard approach for new electron microscopy connectomics datasets. However, leveraging pretrained models from existing datasets could improve efficiency and performance in constrained annotation budget. In this study, we investigate domain adaptation in connectomics by analyzing six major datasets spanning different organisms. We show that, Maximum Mean Discrepancy (MMD) between neuron image distributions serves as a reliable indicator of transferability, and identifies the optimal source domain for transfer learning. Building on this, we introduce NeuroADDA, a method that combines optimal domain selection with source-free active learning to effectively adapt pretrained backbones to a new dataset. NeuroADDA consistently outperforms training from scratch across diverse datasets and fine-tuning sample sizes, with the largest gain observed at $n=4$ samples with a 25-67\% reduction in Variation of Information. Finally, we show that our analysis of distributional differences among neuron images from multiple species in a learned feature space reveals that these domain "distances" correlate with phylogenetic distance among those species.
Authors: Yecong Wan, Mingwen Shao, Yuanshuo Cheng, Jun Shu, Shuigen Wang
Abstract: Videos captured under real-world adverse weather conditions typically suffer from uncertain hybrid weather artifacts with heterogeneous degradation distributions. However, existing algorithms only excel at specific single degradation distributions due to limited adaption capacity and have to deal with different weather degradations with separately trained models, thus may fail to handle real-world stochastic weather scenarios. Besides, the model training is also infeasible due to the lack of paired video data to characterize the coexistence of multiple weather. To ameliorate the aforementioned issue, we propose a novel unified model, dubbed UniWRV, to remove multiple heterogeneous video weather degradations in an all-in-one fashion. Specifically, to tackle degenerate spatial feature heterogeneity, we propose a tailored weather prior guided module that queries exclusive priors for different instances as prompts to steer spatial feature characterization. To tackle degenerate temporal feature heterogeneity, we propose a dynamic routing aggregation module that can automatically select optimal fusion paths for different instances to dynamically integrate temporal features. Additionally, we managed to construct a new synthetic video dataset, termed HWVideo, for learning and benchmarking multiple hybrid adverse weather removal, which contains 15 hybrid weather conditions with a total of 1500 adverse-weather/clean paired video clips. Real-world hybrid weather videos are also collected for evaluating model generalizability. Comprehensive experiments demonstrate that our UniWRV exhibits robust and superior adaptation capability in multiple heterogeneous degradations learning scenarios, including various generic video restoration tasks beyond weather removal.
Authors: Yixin Wu, Feiran Zhang, Tianyuan Shi, Ruicheng Yin, Zhenghua Wang, Zhenliang Gan, Xiaohua Wang, Changze Lv, Xiaoqing Zheng, Xuanjing Huang
Abstract: Recent advances in diffusion models have enabled the creation of deceptively real images, posing significant security risks when misused. In this study, we reveal that natural and synthetic images exhibit distinct differences in the high-frequency domains of their Fourier power spectra after undergoing iterative noise perturbations through an inverse multi-step denoising process, suggesting that such noise can provide additional discriminative information for identifying synthetic images. Based on this observation, we propose a novel detection method that amplifies these differences by progressively adding noise to the original images across multiple timesteps, and train an ensemble of classifiers on these noised images. To enhance human comprehension, we introduce an explanation generation and refinement module to identify flaws located in AI-generated images. Additionally, we construct two new datasets, GenHard and GenExplain, derived from the GenImage benchmark, providing detection samples of greater difficulty and high-quality rationales for fake images. Extensive experiments show that our method achieves state-of-the-art performance with 98.91% and 95.89% detection accuracy on regular and harder samples, increasing a minimal of 2.51% and 3.46% compared to baselines. Furthermore, our method also generalizes effectively to images generated by other diffusion models. Our code and datasets will be made publicly available.
Authors: Meng Wang, Huilong Pi, Ruihui Li, Yunchuan Qin, Zhuo Tang, Kenli Li
Abstract: Camera-based 3D semantic scene completion (SSC) provides dense geometric and semantic perception for autonomous driving. However, images provide limited information making the model susceptible to geometric ambiguity caused by occlusion and perspective distortion. Existing methods often lack explicit semantic modeling between objects, limiting their perception of 3D semantic context. To address these challenges, we propose a novel method VLScene: Vision-Language Guidance Distillation for Camera-based 3D Semantic Scene Completion. The key insight is to use the vision-language model to introduce high-level semantic priors to provide the object spatial context required for 3D scene understanding. Specifically, we design a vision-language guidance distillation process to enhance image features, which can effectively capture semantic knowledge from the surrounding environment and improve spatial context reasoning. In addition, we introduce a geometric-semantic sparse awareness mechanism to propagate geometric structures in the neighborhood and enhance semantic information through contextual sparse interactions. Experimental results demonstrate that VLScene achieves rank-1st performance on challenging benchmarks--SemanticKITTI and SSCBench-KITTI-360, yielding remarkably mIoU scores of 17.52 and 19.10, respectively.
Authors: Xin Ding, Hao Wu, Yifan Yang, Shiqi Jiang, Donglin Bai, Zhibo Chen, Ting Cao
Abstract: With the rise of real-world human-AI interaction applications, such as AI assistants, the need for Streaming Video Dialogue is critical. To address this need, we introduce \sys, a video LLM framework that achieves ultra-FPS streaming video processing (100 fps on a single A100) and enables proactive, always-on responses in real time, without explicit user intervention. To solve the key challenge of the contradiction between linear video streaming speed and quadratic transformer computation cost, we propose a novel perception-cognition interleaving paradigm named ''event-gated LLM invocation'', in contrast to the existing per-time-step LLM invocation. By introducing a Cognition Gate network between the video encoder and the LLM, LLM is only invoked when relevant events occur. To realize the event feature extraction with constant cost, we propose Event-Preserving Feature Extractor (EPFE) based on state-space method, generating a single perception token for spatiotemporal features. These techniques enable the video LLM with full-FPS perception and real-time cognition response. Experiments on Ego4D and SoccerNet streaming tasks, as well as standard offline benchmarks, demonstrate state-of-the-art performance in both model capability and real-time efficiency, paving the way for ultra-high-FPS applications, such as Game AI Copilot and interactive media.
Authors: Meng Wang, Fan Wu, Yunchuan Qin, Ruihui Li, Zhuo Tang, Kenli Li
Abstract: The vision-based semantic scene completion task aims to predict dense geometric and semantic 3D scene representations from 2D images. However, the presence of dynamic objects in the scene seriously affects the accuracy of the model inferring 3D structures from 2D images. Existing methods simply stack multiple frames of image input to increase dense scene semantic information, but ignore the fact that dynamic objects and non-texture areas violate multi-view consistency and matching reliability. To address these issues, we propose a novel method, CDScene: Vision-based Robust Semantic Scene Completion via Capturing Dynamic Representations. First, we leverage a multimodal large-scale model to extract 2D explicit semantics and align them into 3D space. Second, we exploit the characteristics of monocular and stereo depth to decouple scene information into dynamic and static features. The dynamic features contain structural relationships around dynamic objects, and the static features contain dense contextual spatial information. Finally, we design a dynamic-static adaptive fusion module to effectively extract and aggregate complementary features, achieving robust and accurate semantic scene completion in autonomous driving scenarios. Extensive experimental results on the SemanticKITTI, SSCBench-KITTI360, and SemanticKITTI-C datasets demonstrate the superiority and robustness of CDScene over existing state-of-the-art methods.
Authors: Ruofan Wang, Xiang Zheng, Xiaosen Wang, Cong Wang, Xingjun Ma
Abstract: The rapid advancement of large Vision-Language Models (VLMs) has raised significant safety concerns, particularly regarding their vulnerability to jailbreak attacks. While existing research primarily focuses on VLMs' susceptibility to harmful instructions, this work identifies a critical yet overlooked vulnerability: current alignment mechanisms often fail to address the risks posed by toxic text continuation tasks. To investigate this issue, we propose a novel Red Team Diffuser (RTD) framework, which leverages reinforcement learning to generate red team images that effectively induce highly toxic continuations from target black-box VLMs. The RTD pipeline begins with a greedy search for high-quality image prompts that maximize the toxicity of VLM-generated sentence continuations, guided by a Large Language Model (LLM). These prompts are then used as input for the reinforcement fine-tuning of a diffusion model, which employs toxicity and alignment rewards to further amplify harmful outputs. Experimental results demonstrate the effectiveness of RTD, increasing the toxicity rate of LLaVA outputs by 10.69% on the original attack set and 8.91% on a hold-out set. Moreover, RTD exhibits strong cross-model transferability, raising the toxicity rate by 5.1% on Gemini and 26.83% on LLaMA. These findings reveal significant deficiencies in existing alignment strategies, particularly their inability to prevent harmful continuations. Our work underscores the urgent need for more robust and adaptive alignment mechanisms to ensure the safe deployment of VLMs in real-world applications.
Authors: Yang LI, Jinglu Wang, Lei Chu, Xiao Li, Shiu-hong Kao, Ying-Cong Chen, Yan Lu
Abstract: The advent of 3D Gaussian Splatting (3DGS) has advanced 3D scene reconstruction and novel view synthesis. With the growing interest of interactive applications that need immediate feedback, online 3DGS reconstruction in real-time is in high demand. However, none of existing methods yet meet the demand due to three main challenges: the absence of predetermined camera parameters, the need for generalizable 3DGS optimization, and the necessity of reducing redundancy. We propose StreamGS, an online generalizable 3DGS reconstruction method for unposed image streams, which progressively transform image streams to 3D Gaussian streams by predicting and aggregating per-frame Gaussians. Our method overcomes the limitation of the initial point reconstruction \cite{dust3r} in tackling out-of-domain (OOD) issues by introducing a content adaptive refinement. The refinement enhances cross-frame consistency by establishing reliable pixel correspondences between adjacent frames. Such correspondences further aid in merging redundant Gaussians through cross-frame feature aggregation. The density of Gaussians is thereby reduced, empowering online reconstruction by significantly lowering computational and memory costs. Extensive experiments on diverse datasets have demonstrated that StreamGS achieves quality on par with optimization-based approaches but does so 150 times faster, and exhibits superior generalizability in handling OOD scenes.
Authors: Zhaori Liu, Mengyang Li, Hu Han, Enli Zhang, Shiguang Shan, Zhiming Zhao
Abstract: Medical image segmentation is essential for clinical diagnosis, surgical planning, and treatment monitoring. Traditional approaches typically strive to tackle all medical image segmentation scenarios via one-time learning. However, in practical applications, the diversity of scenarios and tasks in medical image segmentation continues to expand, necessitating models that can dynamically evolve to meet the demands of various segmentation tasks. Here, we introduce EvoSAM, a dynamically evolving medical image segmentation model that continuously accumulates new knowledge from an ever-expanding array of scenarios and tasks, enhancing its segmentation capabilities. Extensive evaluations on surgical image blood vessel segmentation and multi-site prostate MRI segmentation demonstrate that EvoSAM not only improves segmentation accuracy but also mitigates catastrophic forgetting. Further experiments conducted by surgical clinicians on blood vessel segmentation confirm that EvoSAM enhances segmentation efficiency based on user prompts, highlighting its potential as a promising tool for clinical applications.
Authors: Yifan Chang, Junjie Huang, Xiaofeng Wang, Yun Ye, Zhujin Liang, Yi Shan, Dalong Du, Xingang Wang
Abstract: Monocular 3D lane detection is a fundamental task in autonomous driving. Although sparse-point methods lower computational load and maintain high accuracy in complex lane geometries, current methods fail to fully leverage the geometric structure of lanes in both lane geometry representations and model design. In lane geometry representations, we present a theoretical analysis alongside experimental validation to verify that current sparse lane representation methods contain inherent flaws, resulting in potential errors of up to 20 m, which raise significant safety concerns for driving. To address this issue, we propose a novel patching strategy to completely represent the full lane structure. To enable existing models to match this strategy, we introduce the EndPoint head (EP-head), which adds a patching distance to endpoints. The EP-head enables the model to predict more complete lane representations even with fewer preset points, effectively addressing existing limitations and paving the way for models that are faster and require fewer parameters in the future. In model design, to enhance the model's perception of lane structures, we propose the PointLane attention (PL-attention), which incorporates prior geometric knowledge into the attention mechanism. Extensive experiments demonstrate the effectiveness of the proposed methods on various state-of-the-art models. For instance, in terms of the overall F1-score, our methods improve Persformer by 4.4 points, Anchor3DLane by 3.2 points, and LATR by 2.8 points. The code will be available soon.
Authors: Kun Xiang, Zhili Liu, Zihao Jiang, Yunshuang Nie, Kaixin Cai, Yiyang Yin, Runhui Huang, Haoxiang Fan, Hanhui Li, Weiran Huang, Yihan Zeng, Yu-Jie Yuan, Jianhua Han, Lanqing Hong, Hang Xu, Xiaodan Liang
Abstract: In this paper, we address the challenging task of multimodal mathematical reasoning by incorporating the ability of "slow thinking" into multimodal large language models (MLLMs). Our core idea is that different levels of reasoning abilities can be combined dynamically to tackle questions with different complexity. To this end, we propose a paradigm of Self-structured Chain of Thought (SCoT), which is composed of minimal semantic atomic steps. Different from existing methods that rely on structured templates or free-form paradigms, our method can not only generate cognitive CoT structures for various complex tasks but also mitigates the phenomenon of overthinking. To introduce structured reasoning capabilities into visual understanding models, we further design a novel AtomThink framework with four key modules, including (i) a data engine to generate high-quality multimodal reasoning paths; (ii) a supervised fine-tuning process with serialized inference data; (iii) a policy-guided multi-turn inference method; and (iv) an atomic capability metric to evaluate the single step utilization rate. We conduct extensive experiments to show that the proposed AtomThink significantly improves the performance of baseline MLLMs, achieving more than 10\% average accuracy gains on MathVista and MathVerse. Compared to state-of-the-art structured CoT approaches, our method not only achieves higher accuracy but also improves data utilization by 5 times and boosts inference efficiency by 85.3\%. Our code is now public available in https://github.com/Quinn777/AtomThink.
Authors: Muzhi Dai, Jiashuo Sun, Zhiyuan Zhao, Shixuan Liu, Rui Li, Junyu Gao, Xuelong Li
Abstract: Aligning large vision-language models (LVLMs) with human preferences is challenging due to the scarcity of fine-grained, high-quality, and multimodal preference data without human annotations. Existing methods relying on direct distillation often struggle with low-confidence data, leading to suboptimal performance. To address this, we propose CAREVL, a novel method for preference reward modeling by reliably using both high- and low-confidence data. First, a cluster of auxiliary expert models (textual reward models) innovatively leverages image captions as weak supervision signals to filter high-confidence data. The high-confidence data are then used to fine-tune the LVLM. Second, low-confidence data are used to generate diverse preference samples using the fine-tuned LVLM. These samples are then scored and selected to construct reliable chosen-rejected pairs for further training. CAREVL achieves performance improvements over traditional distillation-based methods on VL-RewardBench and MLLM-as-a-Judge benchmark, demonstrating its effectiveness. The code will be released soon.
Authors: Wei-En Tai, Yu-Lin Shih, Cheng Sun, Yu-Chiang Frank Wang, Hwann-Tzong Chen
Abstract: Amodal instance segmentation, which aims to detect and segment both visible and invisible parts of objects in images, plays a crucial role in various applications including autonomous driving, robotic manipulation, and scene understanding. While existing methods require training both front-end detectors and mask decoders jointly, this approach lacks flexibility and fails to leverage the strengths of pre-existing modal detectors. To address this limitation, we propose SAMEO, a novel framework that adapts the Segment Anything Model (SAM) as a versatile mask decoder capable of interfacing with various front-end detectors to enable mask prediction even for partially occluded objects. Acknowledging the constraints of limited amodal segmentation datasets, we introduce Amodal-LVIS, a large-scale synthetic dataset comprising 300K images derived from the modal LVIS and LVVIS datasets. This dataset significantly expands the training data available for amodal segmentation research. Our experimental results demonstrate that our approach, when trained on the newly extended dataset, including Amodal-LVIS, achieves remarkable zero-shot performance on both COCOA-cls and D2SA benchmarks, highlighting its potential for generalization to unseen scenarios.
Authors: Shaobin Zhuang, Zhipeng Huang, Binxin Yang, Ying Zhang, Fangyikang Wang, Canmiao Fu, Chong Sun, Zheng-Jun Zha, Chen Li, Yali Wang
Abstract: Video editing increasingly demands the ability to incorporate specific real-world instances into existing footage, yet current approaches fundamentally fail to capture the unique visual characteristics of particular subjects and ensure natural instance/scene interactions. We formalize this overlooked yet critical editing paradigm as "Get-In-Video Editing", where users provide reference images to precisely specify visual elements they wish to incorporate into videos. Addressing this task's dual challenges, severe training data scarcity and technical challenges in maintaining spatiotemporal coherence, we introduce three key contributions. First, we develop GetIn-1M dataset created through our automated Recognize-Track-Erase pipeline, which sequentially performs video captioning, salient instance identification, object detection, temporal tracking, and instance removal to generate high-quality video editing pairs with comprehensive annotations (reference image, tracking mask, instance prompt). Second, we present GetInVideo, a novel end-to-end framework that leverages a diffusion transformer architecture with 3D full attention to process reference images, condition videos, and masks simultaneously, maintaining temporal coherence, preserving visual identity, and ensuring natural scene interactions when integrating reference objects into videos. Finally, we establish GetInBench, the first comprehensive benchmark for Get-In-Video Editing scenario, demonstrating our approach's superior performance through extensive evaluations. Our work enables accessible, high-quality incorporation of specific real-world subjects into videos, significantly advancing personalized video editing capabilities.
Authors: Anh Thai, Songyou Peng, Kyle Genova, Leonidas Guibas, Thomas Funkhouser
Abstract: Language-guided 3D scene understanding is important for advancing applications in robotics, AR/VR, and human-computer interaction, enabling models to comprehend and interact with 3D environments through natural language. While 2D vision-language models (VLMs) have achieved remarkable success in 2D VQA tasks, progress in the 3D domain has been significantly slower due to the complexity of 3D data and the high cost of manual annotations. In this work, we introduce SplatTalk, a novel method that uses a generalizable 3D Gaussian Splatting (3DGS) framework to produce 3D tokens suitable for direct input into a pretrained LLM, enabling effective zero-shot 3D visual question answering (3D VQA) for scenes with only posed images. During experiments on multiple benchmarks, our approach outperforms both 3D models trained specifically for the task and previous 2D-LMM-based models utilizing only images (our setting), while achieving competitive performance with state-of-the-art 3D LMMs that additionally utilize 3D inputs.
Authors: Jeong Hun Yeo, Minsu Kim, Chae Won Kim, Stavros Petridis, Yong Man Ro
Abstract: We explore a novel zero-shot Audio-Visual Speech Recognition (AVSR) framework, dubbed Zero-AVSR, which enables speech recognition in target languages without requiring any audio-visual speech data in those languages. Specifically, we introduce the Audio-Visual Speech Romanizer (AV-Romanizer), which learns language-agnostic speech representations by predicting Roman text. Then, by leveraging the strong multilingual modeling capabilities of Large Language Models (LLMs), we propose converting the predicted Roman text into language-specific graphemes, forming the proposed Cascaded Zero-AVSR. Taking it a step further, we explore a unified Zero-AVSR approach by directly integrating the audio-visual speech representations encoded by the AV-Romanizer into the LLM. This is achieved through finetuning the adapter and the LLM using our proposed multi-task learning scheme. To capture the wide spectrum of phonetic and linguistic diversity, we also introduce a Multilingual Audio-Visual Romanized Corpus (MARC) consisting of 2,916 hours of audio-visual speech data across 82 languages, along with transcriptions in both language-specific graphemes and Roman text. Extensive analysis and experiments confirm that the proposed Zero-AVSR framework has the potential to expand language support beyond the languages seen during the training of the AV-Romanizer.
Authors: Songping Wang, Xinquan Yue, Yueming Lyu, Caifeng Shan
Abstract: Kolmogorov-Arnold Networks (KANs) have emerged as a transformative model paradigm, significantly impacting various fields. However, their adversarial robustness remains less underexplored, especially across different KAN architectures. To explore this critical safety issue, we conduct an analysis and find that due to overfitting to the specific basis functions of KANs, they possess poor adversarial transferability among different KANs. To tackle this challenge, we propose AdvKAN, the first transfer attack method for KANs. AdvKAN integrates two key components: 1) a Breakthrough-Defense Surrogate Model (BDSM), which employs a breakthrough-defense training strategy to mitigate overfitting to the specific structures of KANs. 2) a Global-Local Interaction (GLI) technique, which promotes sufficient interaction between adversarial gradients of hierarchical levels, further smoothing out loss surfaces of KANs. Both of them work together to enhance the strength of transfer attack among different KANs. Extensive experimental results on various KANs and datasets demonstrate the effectiveness of AdvKAN, which possesses notably superior attack capabilities and deeply reveals the vulnerabilities of KANs. Code will be released upon acceptance.
Authors: Siyi Du, Xinzhe Luo, Declan P. O'Regan, Chen Qin
Abstract: Multimodal image-tabular learning is gaining attention, yet it faces challenges due to limited labeled data. While earlier work has applied self-supervised learning (SSL) to unlabeled data, its task-agnostic nature often results in learning suboptimal features for downstream tasks. Semi-supervised learning (SemiSL), which combines labeled and unlabeled data, offers a promising solution. However, existing multimodal SemiSL methods typically focus on unimodal or modality-shared features, ignoring valuable task-relevant modality-specific information, leading to a Modality Information Gap. In this paper, we propose STiL, a novel SemiSL tabular-image framework that addresses this gap by comprehensively exploring task-relevant information. STiL features a new disentangled contrastive consistency module to learn cross-modal invariant representations of shared information while retaining modality-specific information via disentanglement. We also propose a novel consensus-guided pseudo-labeling strategy to generate reliable pseudo-labels based on classifier consensus, along with a new prototype-guided label smoothing technique to refine pseudo-label quality with prototype embeddings, thereby enhancing task-relevant information learning in unlabeled data. Experiments on natural and medical image datasets show that STiL outperforms the state-of-the-art supervised/SSL/SemiSL image/multimodal approaches. Our code is publicly available.
Authors: Shuangzhi Li, Junlong Shen, Lei Ma, Xingyu Li
Abstract: LiDAR-based 3D object detection datasets have been pivotal for autonomous driving, yet they cover a limited range of objects, restricting the model's generalization across diverse deployment environments. To address this, we introduce the first generalized cross-domain few-shot (GCFS) task in 3D object detection, which focuses on adapting a source-pretrained model for high performance on both common and novel classes in a target domain with few-shot samples. Our solution integrates multi-modal fusion and contrastive-enhanced prototype learning within one framework, holistically overcoming challenges related to data scarcity and domain adaptation in the GCFS setting. The multi-modal fusion module utilizes 2D vision-language models to extract rich, open-set semantic knowledge. To address biases in point distributions across varying structural complexities, we particularly introduce a physically-aware box searching strategy that leverages laser imaging principles to generate high-quality 3D box proposals from 2D insights, enhancing object recall. To effectively capture domain-specific representations for each class from limited target data, we further propose a contrastive-enhanced prototype learning, which strengthens the model's adaptability. We evaluate our approach with three GCFS benchmark settings, and extensive experiments demonstrate the effectiveness of our solution for GCFS tasks. The code will be publicly available.
Authors: Seil Kang, Jinyeong Kim, Junhyeok Kim, Seong Jae Hwang
Abstract: Visual grounding seeks to localize the image region corresponding to a free-form text description. Recently, the strong multimodal capabilities of Large Vision-Language Models (LVLMs) have driven substantial improvements in visual grounding, though they inevitably require fine-tuning and additional model components to explicitly generate bounding boxes or segmentation masks. However, we discover that a few attention heads in frozen LVLMs demonstrate strong visual grounding capabilities. We refer to these heads, which consistently capture object locations related to text semantics, as localization heads. Using localization heads, we introduce a straightforward and effective training-free visual grounding framework that utilizes text-to-image attention maps from localization heads to identify the target objects. Surprisingly, only three out of thousands of attention heads are sufficient to achieve competitive localization performance compared to existing LVLM-based visual grounding methods that require fine-tuning. Our findings suggest that LVLMs can innately ground objects based on a deep comprehension of the text-image relationship, as they implicitly focus on relevant image regions to generate informative text outputs. All the source codes will be made available to the public.
Authors: Qizhen Lan, Qing Tian
Abstract: Dense visual prediction tasks, such as detection and segmentation, are crucial for time-critical applications (e.g., autonomous driving and video surveillance). While deep models achieve strong performance, their efficiency remains a challenge. Knowledge distillation (KD) is an effective model compression technique, but existing feature-based KD methods rely on static, teacher-driven feature selection, failing to adapt to the student's evolving learning state or leverage dynamic student-teacher interactions. To address these limitations, we propose Adaptive student-teacher Cooperative Attention Masking for Knowledge Distillation (ACAM-KD), which introduces two key components: (1) Student-Teacher Cross-Attention Feature Fusion (STCA-FF), which adaptively integrates features from both models for a more interactive distillation process, and (2) Adaptive Spatial-Channel Masking (ASCM), which dynamically generates importance masks to enhance both spatial and channel-wise feature selection. Unlike conventional KD methods, ACAM-KD adapts to the student's evolving needs throughout the entire distillation process. Extensive experiments on multiple benchmarks validate its effectiveness. For instance, on COCO2017, ACAM-KD improves object detection performance by up to 1.4 mAP over the state-of-the-art when distilling a ResNet-50 student from a ResNet-101 teacher. For semantic segmentation on Cityscapes, it boosts mIoU by 3.09 over the baseline with DeepLabV3-MobileNetV2 as the student model.
Authors: Taewon Kang, Divya Kothandaraman, Ming C. Lin
Abstract: Generating coherent long-form video sequences from discrete input using only text prompts is a critical task in content creation. While diffusion-based models excel at short video synthesis, long-form storytelling from text remains largely unexplored and a challenge due to challenges pertaining to temporal coherency, preserving semantic meaning and action continuity across the video. We introduce a novel storytelling approach to enable seamless video generation with natural action transitions and structured narratives. We present a bidirectional time-weighted latent blending strategy to ensure temporal consistency between segments of the long-form video being generated. Further, our method extends the Black-Scholes algorithm from prompt mixing for image generation to video generation, enabling controlled motion evolution through structured text conditioning. To further enhance motion continuity, we propose a semantic action representation framework to encode high-level action semantics into the blending process, dynamically adjusting transitions based on action similarity, ensuring smooth yet adaptable motion changes. Latent space blending maintains spatial coherence between objects in a scene, while time-weighted blending enforces bidirectional constraints for temporal consistency. This integrative approach prevents abrupt transitions while ensuring fluid storytelling. Extensive experiments demonstrate significant improvements over baselines, achieving temporally consistent and visually compelling video narratives without any additional training. Our approach bridges the gap between short clips and extended video to establish a new paradigm in GenAI-driven video synthesis from text.
Authors: Zhitong Xiong, Yi Wang, Weikang Yu, Adam J Stewart, Jie Zhao, Nils Lehmann, Thomas Dujardin, Zhenghang Yuan, Pedram Ghamisi, Xiao Xiang Zhu
Abstract: Earth observation (EO) data, collected from diverse sensors with varying imaging principles, present significant challenges in creating unified analytical frameworks. We present GeoLangBind, a novel agglomerative vision--language foundation model that bridges the gap between heterogeneous EO data modalities using language as a unifying medium. Our approach aligns different EO data types into a shared language embedding space, enabling seamless integration and complementary feature learning from diverse sensor data. To achieve this, we construct a large-scale multimodal image--text dataset, GeoLangBind-2M, encompassing six data modalities. GeoLangBind leverages this dataset to develop a zero-shot foundation model capable of processing arbitrary numbers of EO data channels as input. Through our designed Modality-aware Knowledge Agglomeration (MaKA) module and progressive multimodal weight merging strategy, we create a powerful agglomerative foundation model that excels in both zero-shot vision--language comprehension and fine-grained visual understanding. Extensive evaluation across 23 datasets covering multiple tasks demonstrates GeoLangBind's superior performance and versatility in EO applications, offering a robust framework for various environmental monitoring and analysis tasks. The dataset and pretrained models will be publicly available.
Authors: Chandan Kumar Sah, Ankit Kumar Shaw, Xiaoli Lian, Arsalan Shahid Baig, Tuopu Wen, Kun Jiang, Mengmeng Yang, Diange Yang
Abstract: Autonomous vehicles (AVs) require reliable traffic sign recognition and robust lane detection capabilities to ensure safe navigation in complex and dynamic environments. This paper introduces an integrated approach combining advanced deep learning techniques and Multimodal Large Language Models (MLLMs) for comprehensive road perception. For traffic sign recognition, we systematically evaluate ResNet-50, YOLOv8, and RT-DETR, achieving state-of-the-art performance of 99.8% with ResNet-50, 98.0% accuracy with YOLOv8, and achieved 96.6% accuracy in RT-DETR despite its higher computational complexity. For lane detection, we propose a CNN-based segmentation method enhanced by polynomial curve fitting, which delivers high accuracy under favorable conditions. Furthermore, we introduce a lightweight, Multimodal, LLM-based framework that directly undergoes instruction tuning using small yet diverse datasets, eliminating the need for initial pretraining. This framework effectively handles various lane types, complex intersections, and merging zones, significantly enhancing lane detection reliability by reasoning under adverse conditions. Despite constraints in available training resources, our multimodal approach demonstrates advanced reasoning capabilities, achieving a Frame Overall Accuracy (FRM) of 53.87%, a Question Overall Accuracy (QNS) of 82.83%, lane detection accuracies of 99.6% in clear conditions and 93.0% at night, and robust performance in reasoning about lane invisibility due to rain (88.4%) or road degradation (95.6%). The proposed comprehensive framework markedly enhances AV perception reliability, thus contributing significantly to safer autonomous driving across diverse and challenging road scenarios.
Authors: Tieqiao Wang, Sinisa Todorovic
Abstract: Existing approaches to action segmentation use pre-computed frame features extracted by methods which have been trained on tasks that are different from action segmentation. Also, recent approaches typically use deep framewise representations that lack explicit modeling of action segments. To address these shortcomings, we introduce the first end-to-end solution to action segmentation -- End-to-End Action Segmentation Transformer (EAST). Our key contributions include: (1) a simple and efficient adapter design for effective backbone fine-tuning; (2) a segmentation-by-detection framework for leveraging action proposals initially predicted over a coarsely downsampled video toward labeling of all frames; and (3) a new action-proposal based data augmentation for robust training. EAST achieves state-of-the-art performance on standard benchmarks, including GTEA, 50Salads, Breakfast, and Assembly-101. The model and corresponding code will be released.
Authors: Badhan Chandra Das, M. Hadi Amini, Yanzhao Wu
Abstract: Object detection in videos plays a crucial role in advancing applications such as public safety and anomaly detection. Existing methods have explored different techniques, including CNN, deep learning, and Transformers, for object detection and video classification. However, detecting tiny objects, e.g., guns, in videos remains challenging due to their small scale and varying appearances in complex scenes. Moreover, existing video analysis models for classification or detection often perform poorly in real-world gun detection scenarios due to limited labeled video datasets for training. Thus, developing efficient methods for effectively capturing tiny object features and designing models capable of accurate gun detection in real-world videos is imperative. To address these challenges, we make three original contributions in this paper. First, we conduct an empirical study of several existing video classification and object detection methods to identify guns in videos. Our extensive analysis shows that these methods may not accurately detect guns in videos. Second, we propose a novel two-stage gun detection method. In stage 1, we train an image-augmented model to effectively classify ``Gun'' videos. To make the detection more precise and efficient, stage 2 employs an object detection model to locate the exact region of the gun within video frames for videos classified as ``Gun'' by stage 1. Third, our experimental results demonstrate that the proposed domain-specific method achieves significant performance improvements and enhances efficiency compared with existing techniques. We also discuss challenges and future research directions in gun detection tasks in computer vision.
Authors: \"Omer Veysel \c{C}a\u{g}atan, \"Omer Faruk Tal, M. Emre G\"ursoy
Abstract: Self-supervised learning (SSL) has advanced significantly in visual representation learning, yet comprehensive evaluations of its adversarial robustness remain limited. In this study, we evaluate the adversarial robustness of seven discriminative self-supervised models and one supervised model across diverse tasks, including ImageNet classification, transfer learning, segmentation, and detection. Our findings suggest that discriminative SSL models generally exhibit better robustness to adversarial attacks compared to their supervised counterpart on ImageNet, with this advantage extending to transfer learning when using linear evaluation. However, when fine-tuning is applied, the robustness gap between SSL and supervised models narrows considerably. Similarly, this robustness advantage diminishes in segmentation and detection tasks. We also investigate how various factors might influence adversarial robustness, including architectural choices, training duration, data augmentations, and batch sizes. Our analysis contributes to the ongoing exploration of adversarial robustness in visual self-supervised representation systems.
Authors: Umberto Cappellazzo, Minsu Kim, Stavros Petridis
Abstract: Audio-Visual Speech Recognition (AVSR) leverages both audio and visual modalities to enhance speech recognition robustness, particularly in noisy environments. Recent advancements in Large Language Models (LLMs) have demonstrated their effectiveness in speech recognition, including AVSR. However, due to the significant length of speech representations, direct integration with LLMs imposes substantial computational costs. Prior approaches address this by compressing speech representations before feeding them into LLMs. However, higher compression ratios often lead to performance degradation, necessitating a trade-off between computational efficiency and recognition accuracy. To address this challenge, we propose Llama-MTSK, the first Matryoshka-based Multimodal LLM for AVSR, which enables flexible adaptation of the audio-visual token allocation based on specific computational constraints while preserving high performance. Our approach, inspired by Matryoshka Representation Learning, encodes audio-visual representations at multiple granularities within a single model, eliminating the need to train separate models for different compression levels. Moreover, to efficiently fine-tune the LLM, we introduce three LoRA-based Matryoshka strategies using global and scale-specific LoRA modules. Extensive evaluations on the two largest AVSR datasets demonstrate that Llama-MTSK achieves state-of-the-art results, matching or surpassing models trained independently at fixed compression levels.
Authors: Chen Liu, Tobias Ritschel
Abstract: We propose a novel generative video model by robustly learning temporal change as a neural Ordinary Differential Equation (ODE) flow with a bilinear objective of combining two aspects: The first is to map from the past into future video frames directly. Previous work has mapped the noise to new frames, a more computationally expensive process. Unfortunately, starting from the previous frame, instead of noise, is more prone to drifting errors. Hence, second, we additionally learn how to remove the accumulated errors as the joint objective by adding noise during training. We demonstrate unconditional video generation in a streaming manner for various video datasets, all at competitive quality compared to a baseline conditional diffusion but with higher speed, i.e., fewer ODE solver steps.
Authors: Leonardo Scabini, Kallil M. Zielinski, Emir Konuk, Ricardo T. Fares, Lucas C. Ribas, Kevin Smith, Odemir M. Bruno
Abstract: Texture recognition has recently been dominated by ImageNet-pre-trained deep Convolutional Neural Networks (CNNs), with specialized modifications and feature engineering required to achieve state-of-the-art (SOTA) performance. However, although Vision Transformers (ViTs) were introduced a few years ago, little is known about their texture recognition ability. Therefore, in this work, we introduce VORTEX (ViTs with Orderless and Randomized Token Encodings for Texture Recognition), a novel method that enables the effective use of ViTs for texture analysis. VORTEX extracts multi-depth token embeddings from pre-trained ViT backbones and employs a lightweight module to aggregate hierarchical features and perform orderless encoding, obtaining a better image representation for texture recognition tasks. This approach allows seamless integration with any ViT with the common transformer architecture. Moreover, no fine-tuning of the backbone is performed, since they are used only as frozen feature extractors, and the features are fed to a linear SVM. We evaluate VORTEX on nine diverse texture datasets, demonstrating its ability to achieve or surpass SOTA performance in a variety of texture analysis scenarios. By bridging the gap between texture recognition with CNNs and transformer-based architectures, VORTEX paves the way for adopting emerging transformer foundation models. Furthermore, VORTEX demonstrates robust computational efficiency when coupled with ViT backbones compared to CNNs with similar costs. The method implementation and experimental scripts are publicly available in our online repository.
Authors: Sahar Dastani, Ali Bahri, Moslem Yazdanpanah, Mehrdad Noori, David Osowiechi, Gustavo Adolfo Vargas Hakim, Farzad Beizaee, Milad Cheraghalikhani, Arnab Kumar Mondal, Herve Lombaert, Christian Desrosiers
Abstract: State Space Models (SSMs) have recently emerged as an alternative to Vision Transformers (ViTs) due to their unique ability of modeling global relationships with linear complexity. SSMs are specifically designed to capture spatially proximate relationships of image patches. However, they fail to identify relationships between conceptually related yet not adjacent patches. This limitation arises from the non-causal nature of image data, which lacks inherent directional relationships. Additionally, current vision-based SSMs are highly sensitive to transformations such as rotation. Their predefined scanning directions depend on the original image orientation, which can cause the model to produce inconsistent patch-processing sequences after rotation. To address these limitations, we introduce Spectral VMamba, a novel approach that effectively captures the global structure within an image by leveraging spectral information derived from the graph Laplacian of image patches. Through spectral decomposition, our approach encodes patch relationships independently of image orientation, achieving rotation invariance with the aid of our Rotational Feature Normalizer (RFN) module. Our experiments on classification tasks show that Spectral VMamba outperforms the leading SSM models in vision, such as VMamba, while maintaining invariance to rotations and a providing a similar runtime efficiency.
Authors: Khang H. N. Vo, Duc P. T. Nguyen, Thong Nguyen, Tho T. Quan
Abstract: This paper focuses on multimodal alignment within the realm of Artificial Intelligence, particularly in text and image modalities. The semantic gap between the textual and visual modality poses a discrepancy problem towards the effectiveness of multi-modalities fusion. Therefore, we introduce Text-Image Joint Embedding Predictive Architecture (TI-JEPA), an innovative pre-training strategy that leverages energy-based model (EBM) framework to capture complex cross-modal relationships. TI-JEPA combines the flexibility of EBM in self-supervised learning to facilitate the compatibility between textual and visual elements. Through extensive experiments across multiple benchmarks, we demonstrate that TI-JEPA achieves state-of-the-art performance on multimodal sentiment analysis task (and potentially on a wide range of multimodal-based tasks, such as Visual Question Answering), outperforming existing pre-training methodologies. Our findings highlight the potential of using energy-based framework in advancing multimodal fusion and suggest significant improvements for downstream applications.
Authors: Yanyu Zhu, Licheng Bai, Jintao Xu, Jiwei Tang, Hai-tao Zheng
Abstract: Recent advances in diffusion-based lip-syncing generative models have demonstrated their ability to produce highly synchronized talking face videos for visual dubbing. Although these models excel at lip synchronization, they often struggle to maintain fine-grained control over facial details in generated images. In this work, we identify "lip averaging" phenomenon where the model fails to preserve subtle facial details when dubbing unseen in-the-wild videos. This issue arises because the commonly used UNet backbone primarily integrates audio features into visual representations in the latent space via cross-attention mechanisms and multi-scale fusion, but it struggles to retain fine-grained lip details in the generated faces. To address this issue, we propose UnAvgLip, which extracts identity embeddings from reference videos to generate highly faithful facial sequences while maintaining accurate lip synchronization. Specifically, our method comprises two primary components: (1) an Identity Perceiver module that encodes facial embeddings to align with conditioned audio features; and (2) an ID-CrossAttn module that injects facial embeddings into the generation process, enhancing model's capability of identity retention. Extensive experiments demonstrate that, at a modest training and inference cost, UnAvgLip effectively mitigates the "averaging" phenomenon in lip inpainting, significantly preserving unique facial characteristics while maintaining precise lip synchronization. Compared with the original approach, our method demonstrates significant improvements of 5% on the identity consistency metric and 2% on the SSIM metric across two benchmark datasets (HDTF and LRW).
Authors: Haisheng Fu, Jie Liang, Zhenman Fang, Jingning Han
Abstract: Learned image compression (LIC) methods have recently outperformed traditional codecs such as VVC in rate-distortion performance. However, their large models and high computational costs have limited their practical adoption. In this paper, we first construct a high-capacity teacher model by integrating Swin-Transformer V2-based attention modules, additional residual blocks, and expanded latent channels, thus achieving enhanced compression performance. Building on this foundation, we propose a \underline{F}eature and \underline{E}ntropy-based \underline{D}istillation \underline{S}trategy (\textbf{FEDS}) that transfers key knowledge from the teacher to a lightweight student model. Specifically, we align intermediate feature representations and emphasize the most informative latent channels through an entropy-based loss. A staged training scheme refines this transfer in three phases: feature alignment, channel-level distillation, and final fine-tuning. Our student model nearly matches the teacher across Kodak (1.24\% BD-Rate increase), Tecnick (1.17\%), and CLIC (0.55\%) while cutting parameters by about 63\% and accelerating encoding/decoding by around 73\%. Moreover, ablation studies indicate that FEDS generalizes effectively to transformer-based networks. The experimental results demonstrate our approach strikes a compelling balance among compression performance, speed, and model parameters, making it well-suited for real-time or resource-limited scenarios.
Authors: Alex Dolce, Ryan Lavelle, Bernard Scott, Ashlyn Urbanski, Joseph Klobusicky
Abstract: The turning distance is an efficient metric for measuring the similarity between two polygons. This metric is constructed by taking an $L^p$ distance between step functions which track each shape's tangent angle of a path tracing its boundary. For $p = 2$, computing the turning distance between two polygons with $n$ and $m$ sides takes $O(mn \log(m+n))$ time. In this study, we introduce turning disorders for polygonal planar networks, defined by averaging turning distances between network faces and "ordered" shapes (regular polygons or circles). The run time for computing turning distances reduces to $O(m+n) \log(m+n)$ when both shapes are ordered. We also derive closed-form expressions of turning distances for special classes of regular polygons, related to the divisibility of $m$ and $n$, and also between regular polygons and circles. We apply these formulas to several examples of network microstructure with varying disorder. For Archimedean lattices, a class of regular tilings, we can express turning disorders with exact expressions. We also consider turning disorders applied to two examples of stochastic processes on networks: spring networks evolving under T1 moves and polygonal rupture processes. We find that the two aspects of defining different turning disorders, the choice of ordered shape and whether to apply area-weighting, can capture different notions of network disorder.
Authors: Tao Xia, Yudi Zhang, Ting Liu Lei Zhang
Abstract: Despite the great success of large-scale text-to-image diffusion models in image generation and image editing, existing methods still struggle to edit the layout of real images. Although a few works have been proposed to tackle this problem, they either fail to adjust the layout of images, or have difficulty in preserving visual appearance of objects after the layout adjustment. To bridge this gap, this paper proposes a novel image layout editing method that can not only re-arrange a real image to a specified layout, but also can ensure the visual appearance of the objects consistent with their appearance before editing. Concretely, the proposed method consists of two key components. Firstly, a multi-concept learning scheme is used to learn the concepts of different objects from a single image, which is crucial for keeping visual consistency in the layout editing. Secondly, it leverages the semantic consistency within intermediate features of diffusion models to project the appearance information of objects to the desired regions directly. Besides, a novel initialization noise design is adopted to facilitate the process of re-arranging the layout. Extensive experiments demonstrate that the proposed method outperforms previous works in both layout alignment and visual consistency for the task of image layout editing
Authors: Adrian Chow, Evelien Riddell, Yimu Wang, Sean Sedwards, Krzysztof Czarnecki
Abstract: Open-vocabulary 3D object detection for autonomous driving aims to detect novel objects beyond the predefined training label sets in point cloud scenes. Existing approaches achieve this by connecting traditional 3D object detectors with vision-language models (VLMs) to regress 3D bounding boxes for novel objects and perform open-vocabulary classification through cross-modal alignment between 3D and 2D features. However, achieving robust cross-modal alignment remains a challenge due to semantic inconsistencies when generating corresponding 3D and 2D feature pairs. To overcome this challenge, we present OV-SCAN, an Open-Vocabulary 3D framework that enforces Semantically Consistent Alignment for Novel object discovery. OV-SCAN employs two core strategies: discovering precise 3D annotations and filtering out low-quality or corrupted alignment pairs (arising from 3D annotation, occlusion-induced, or resolution-induced noise). Extensive experiments on the nuScenes dataset demonstrate that OV-SCAN achieves state-of-the-art performance.
Authors: Juhyeon Park, Peter Yongho Kim, Jiook Cha, Shinjae Yoo, Taesup Moon
Abstract: We present SEED (\textbf{Se}mantic \textbf{E}valuation for Visual Brain \textbf{D}ecoding), a novel metric for evaluating the semantic decoding performance of visual brain decoding models. It integrates three complementary metrics, each capturing a different aspect of semantic similarity between images. Using carefully crowd-sourced human judgment data, we demonstrate that SEED achieves the highest alignment with human evaluations, outperforming other widely used metrics. Through the evaluation of existing visual brain decoding models, we further reveal that crucial information is often lost in translation, even in state-of-the-art models that achieve near-perfect scores on existing metrics. To facilitate further research, we open-source the human judgment data, encouraging the development of more advanced evaluation methods for brain decoding models. Additionally, we propose a novel loss function designed to enhance semantic decoding performance by leveraging the order of pairwise cosine similarity in CLIP image embeddings. This loss function is compatible with various existing methods and has been shown to consistently improve their semantic decoding performances when used for training, with respect to both existing metrics and SEED.
Authors: Yu Liu, Hao Tang, Haiqi Zhang, Jing Qin, Zechao Li
Abstract: Out-of-distribution (OOD) detection is crucial for ensuring the reliability and safety of machine learning models in real-world applications. While zero-shot OOD detection, which requires no training on in-distribution (ID) data, has become feasible with the emergence of vision-language models like CLIP, existing methods primarily focus on semantic matching and fail to fully capture distributional discrepancies. To address these limitations, we propose OT-DETECTOR, a novel framework that employs Optimal Transport (OT) to quantify both semantic and distributional discrepancies between test samples and ID labels. Specifically, we introduce cross-modal transport mass and transport cost as semantic-wise and distribution-wise OOD scores, respectively, enabling more robust detection of OOD samples. Additionally, we present a semantic-aware content refinement (SaCR) module, which utilizes semantic cues from ID labels to amplify the distributional discrepancy between ID and hard OOD samples. Extensive experiments on several benchmarks demonstrate that OT-DETECTOR achieves state-of-the-art performance across various OOD detection tasks, particularly in challenging hard-OOD scenarios.
Authors: Mingxiang Cao, Weiying Xie, Xin Zhang, Jiaqing Zhang, Kai Jiang, Jie Lei, Yunsong Li
Abstract: Multi-modal fusion holds great promise for integrating information from different modalities. However, due to a lack of consideration for modal consistency, existing multi-modal fusion methods in the field of remote sensing still face challenges of incomplete semantic information and low computational efficiency in their fusion designs. Inspired by the observation that the visual language pre-training model CLIP can effectively extract strong semantic information from visual features, we propose M$^3$amba, a novel end-to-end CLIP-driven Mamba model for multi-modal fusion to address these challenges. Specifically, we introduce CLIP-driven modality-specific adapters in the fusion architecture to avoid the bias of understanding specific domains caused by direct inference, making the original CLIP encoder modality-specific perception. This unified framework enables minimal training to achieve a comprehensive semantic understanding of different modalities, thereby guiding cross-modal feature fusion. To further enhance the consistent association between modality mappings, a multi-modal Mamba fusion architecture with linear complexity and a cross-attention module Cross-SS2D are designed, which fully considers effective and efficient information interaction to achieve complete fusion. Extensive experiments have shown that M$^3$amba has an average performance improvement of at least 5.98\% compared with the state-of-the-art methods in multi-modal hyperspectral image classification tasks in the remote sensing field, while also demonstrating excellent training efficiency, achieving a double improvement in accuracy and efficiency. The code is released at https://github.com/kaka-Cao/M3amba.
Authors: Basudha Pal (Cyan), Siyuan (Cyan), Huang, Rama Chellappa
Abstract: Person Re-identification (ReID) systems identify individuals across images or video frames and play a critical role in various real-world applications. However, many ReID methods are influenced by sensitive attributes such as gender, pose, and body mass index (BMI), which vary in uncontrolled environments, leading to biases and reduced generalization. To address this, we extend the concept of expressivity to the body recognition domain to better understand how ReID models encode these attributes. Expressivity, defined as the mutual information between feature vector representations and specific attributes, is computed using a secondary neural network that takes feature and attribute vectors as inputs. This provides a quantitative framework for analyzing the extent to which sensitive attributes are embedded in the model's representations. We apply expressivity analysis to SemReID, a state-of-the-art self-supervised ReID model, and find that BMI consistently exhibits the highest expressivity scores in the model's final layers, underscoring its dominant role in feature encoding. In the final attention layer of the trained network, the expressivity order for body attributes is BMI > Pitch > Yaw > Gender, highlighting their relative importance in learned representations. Additionally, expressivity values evolve progressively across network layers and training epochs, reflecting a dynamic encoding of attributes during feature extraction. These insights emphasize the influence of body-related attributes on ReID models and provide a systematic methodology for identifying and mitigating attribute-driven biases. By leveraging expressivity analysis, we offer valuable tools to enhance the fairness, robustness, and generalization of ReID systems in diverse real-world settings.
Authors: Chengxuan Qian, Kai Han, Jingchao Wang, Zhenlong Yuan, Rui Qian, Chongwen Lyu, Jun Chen, Zhe Liu
Abstract: Multimodal learning integrates complementary information from diverse modalities to enhance the decision-making process. However, the potential of multimodal collaboration remains under-exploited due to disparities in data quality and modality representation capabilities. To address this, we introduce DynCIM, a novel dynamic curriculum learning framework designed to quantify the inherent imbalances from both sample and modality perspectives. DynCIM employs a sample-level curriculum to dynamically assess each sample's difficulty according to prediction deviation, consistency, and stability, while a modality-level curriculum measures modality contributions from global and local. Furthermore, a gating-based dynamic fusion mechanism is introduced to adaptively adjust modality contributions, minimizing redundancy and optimizing fusion effectiveness. Extensive experiments on six multimodal benchmarking datasets, spanning both bimodal and trimodal scenarios, demonstrate that DynCIM consistently outperforms state-of-the-art methods. Our approach effectively mitigates modality and sample imbalances while enhancing adaptability and robustness in multimodal learning tasks. Our code is available at https://github.com/Raymond-Qiancx/DynCIM.
Authors: Yanbiao Ma, Wei Dai, Wenke Huang, Jiayi Chen
Abstract: Data heterogeneity in federated learning, characterized by a significant misalignment between local and global distributions, leads to divergent local optimization directions and hinders global model training. Existing studies mainly focus on optimizing local updates or global aggregation, but these indirect approaches demonstrate instability when handling highly heterogeneous data distributions, especially in scenarios where label skew and domain skew coexist. To address this, we propose a geometry-guided data generation method that centers on simulating the global embedding distribution locally. We first introduce the concept of the geometric shape of an embedding distribution and then address the challenge of obtaining global geometric shapes under privacy constraints. Subsequently, we propose GGEUR, which leverages global geometric shapes to guide the generation of new samples, enabling a closer approximation to the ideal global distribution. In single-domain scenarios, we augment samples based on global geometric shapes to enhance model generalization; in multi-domain scenarios, we further employ class prototypes to simulate the global distribution across domains. Extensive experimental results demonstrate that our method significantly enhances the performance of existing approaches in handling highly heterogeneous data, including scenarios with label skew, domain skew, and their coexistence. Code published at: https://github.com/WeiDai-David/2025CVPR_GGEUR
Authors: Guanyu Cao, Takuya Maekawa, Kazuya Ohara, Yasue Kishino
Abstract: This study proposes a new deep learning method for reconstructing depth images of moving objects within a specific area using Wi-Fi channel state information (CSI). The Wi-Fi-based depth imaging technique has novel applications in domains such as security and elder care. However, reconstructing depth images from CSI is challenging because learning the mapping function between CSI and depth images, both of which are high-dimensional data, is particularly difficult. To address the challenge, we propose a new approach called Wi-Depth. The main idea behind the design of Wi-Depth is that a depth image of a moving object can be decomposed into three core components: the shape, depth, and position of the target. Therefore, in the depth-image reconstruction task, Wi-Depth simultaneously estimates the three core pieces of information as auxiliary tasks in our proposed VAE-based teacher-student architecture, enabling it to output images with the consistency of a correct shape, depth, and position. In addition, the design of Wi-Depth is based on our idea that this decomposition efficiently takes advantage of the fact that shape, depth, and position relate to primitive information inferred from CSI such as angle-of-arrival, time-of-flight, and Doppler frequency shift.
Authors: Seungju Cho, Hongsin Lee, Changick Kim
Abstract: Adversarial training significantly enhances adversarial robustness, yet superior performance is predominantly achieved on balanced datasets. Addressing adversarial robustness in the context of unbalanced or long-tailed distributions is considerably more challenging, mainly due to the scarcity of tail data instances. Previous research on adversarial robustness within long-tailed distributions has primarily focused on combining traditional long-tailed natural training with existing adversarial robustness methods. In this study, we provide an in-depth analysis for the challenge that adversarial training struggles to achieve high performance on tail classes in long-tailed distributions. Furthermore, we propose a simple yet effective solution to advance adversarial robustness on long-tailed distributions through a novel self-distillation technique. Specifically, this approach leverages a balanced self-teacher model, which is trained using a balanced dataset sampled from the original long-tailed dataset. Our extensive experiments demonstrate state-of-the-art performance in both clean and robust accuracy for long-tailed adversarial robustness, with significant improvements in tail class performance on various datasets. We improve the accuracy against PGD attacks for tail classes by 20.3, 7.1, and 3.8 percentage points on CIFAR-10, CIFAR-100, and Tiny-ImageNet, respectively, while achieving the highest robust accuracy.
Authors: Zexu Huang, Min Xu, Stuart Perry
Abstract: Recent advancements in 3D reconstruction coupled with neural rendering techniques have greatly improved the creation of photo-realistic 3D scenes, influencing both academic research and industry applications. The technique of 3D Gaussian Splatting and its variants incorporate the strengths of both primitive-based and volumetric representations, achieving superior rendering quality. While 3D Geometric Scattering (3DGS) and its variants have advanced the field of 3D representation, they fall short in capturing the stochastic properties of non-local structural information during the training process. Additionally, the initialisation of spherical functions in 3DGS-based methods often fails to engage higher-order terms in early training rounds, leading to unnecessary computational overhead as training progresses. Furthermore, current 3DGS-based approaches require training on higher resolution images to render higher resolution outputs, significantly increasing memory demands and prolonging training durations. We introduce StructGS, a framework that enhances 3D Gaussian Splatting (3DGS) for improved novel-view synthesis in 3D reconstruction. StructGS innovatively incorporates a patch-based SSIM loss, dynamic spherical harmonics initialisation and a Multi-scale Residual Network (MSRN) to address the above-mentioned limitations, respectively. Our framework significantly reduces computational redundancy, enhances detail capture and supports high-resolution rendering from low-resolution inputs. Experimentally, StructGS demonstrates superior performance over state-of-the-art (SOTA) models, achieving higher quality and more detailed renderings with fewer artifacts.
Authors: Shijia Zhao, Qiming Xia, Xusheng Guo, Pufan Zou, Maoji Zheng, Hai Wu, Chenglu Wen, Cheng Wang
Abstract: Recently, sparsely-supervised 3D object detection has gained great attention, achieving performance close to fully-supervised 3D objectors while requiring only a few annotated instances. Nevertheless, these methods suffer challenges when accurate labels are extremely absent. In this paper, we propose a boosting strategy, termed SP3D, explicitly utilizing the cross-modal semantic prompts generated from Large Multimodal Models (LMMs) to boost the 3D detector with robust feature discrimination capability under sparse annotation settings. Specifically, we first develop a Confident Points Semantic Transfer (CPST) module that generates accurate cross-modal semantic prompts through boundary-constrained center cluster selection. Based on these accurate semantic prompts, which we treat as seed points, we introduce a Dynamic Cluster Pseudo-label Generation (DCPG) module to yield pseudo-supervision signals from the geometry shape of multi-scale neighbor points. Additionally, we design a Distribution Shape score (DS score) that chooses high-quality supervision signals for the initial training of the 3D detector. Experiments on the KITTI dataset and Waymo Open Dataset (WOD) have validated that SP3D can enhance the performance of sparsely supervised detectors by a large margin under meager labeling conditions. Moreover, we verified SP3D in the zero-shot setting, where its performance exceeded that of the state-of-the-art methods. The code is available at https://github.com/xmuqimingxia/SP3D.
Authors: George Tang, Aditya Agarwal, Weiqiao Han, Trevor Darrell, Yutong Bai
Abstract: We generalize lifting to semantic lifting by incorporating per-view masks that indicate relevant pixels for lifting tasks. These masks are determined by querying corresponding multiscale pixel-aligned feature maps, which are derived from scene representations such as distilled feature fields and feature point clouds. However, storing per-view feature maps rendered from distilled feature fields is impractical, and feature point clouds are expensive to store and query. To enable lightweight on-demand retrieval of pixel-aligned relevance masks, we introduce the Vector-Quantized Feature Field. We demonstrate the effectiveness of the Vector-Quantized Feature Field on complex indoor and outdoor scenes. Semantic lifting, when paired with a Vector-Quantized Feature Field, can unlock a myriad of applications in scene representation and embodied intelligence. Specifically, we showcase how our method enables text-driven localized scene editing and significantly improves the efficiency of embodied question answering.
Authors: Qiaole Dong, Yanwei Fu
Abstract: Dense point tracking is a challenging task requiring the continuous tracking of every point in the initial frame throughout a substantial portion of a video, even in the presence of occlusions. Traditional methods use optical flow models to directly estimate long-range motion, but they often suffer from appearance drifting without considering temporal consistency. Recent point tracking algorithms usually depend on sliding windows for indirect information propagation from the first frame to the current one, which is slow and less effective for long-range tracking. To account for temporal consistency and enable efficient information propagation, we present a lightweight and fast model with \textbf{S}treaming memory for dense \textbf{PO}int \textbf{T}racking and online video processing. The \textbf{SPOT} framework features three core components: a customized memory reading module for feature enhancement, a sensory memory for short-term motion dynamics modeling, and a visibility-guided splatting module for accurate information propagation. This combination enables SPOT to perform dense point tracking with state-of-the-art accuracy on the CVO benchmark, as well as comparable or superior performance to offline models on sparse tracking benchmarks such as TAP-Vid and RoboTAP. Notably, SPOT with 10$\times$ smaller parameter numbers operates at least 2$\times$ faster than previous state-of-the-art models while maintaining the best performance on CVO. We will release the models and codes at: https://github.com/DQiaole/SPOT.
Authors: Yuxuan Luo, Jiaqi Tang, Chenyi Huang, Feiyang Hao, Zhouhui Lian
Abstract: Chinese calligraphy, a UNESCO Heritage, remains computationally challenging due to visual ambiguity and cultural complexity. Existing AI systems fail to contextualize their intricate scripts, because of limited annotated data and poor visual-semantic alignment. We propose CalliReader, a vision-language model (VLM) that solves the Chinese Calligraphy Contextualization (CC$^2$) problem through three innovations: (1) character-wise slicing for precise character extraction and sorting, (2) CalliAlign for visual-text token compression and alignment, (3) embedding instruction tuning (e-IT) for improving alignment and addressing data scarcity. We also build CalliBench, the first benchmark for full-page calligraphic contextualization, addressing three critical issues in previous OCR and VQA approaches: fragmented context, shallow reasoning, and hallucination. Extensive experiments including user studies have been conducted to verify our CalliReader's \textbf{superiority to other state-of-the-art methods and even human professionals in page-level calligraphy recognition and interpretation}, achieving higher accuracy while reducing hallucination. Comparisons with reasoning models highlight the importance of accurate recognition as a prerequisite for reliable comprehension. Quantitative analyses validate CalliReader's efficiency; evaluations on document and real-world benchmarks confirm its robust generalization ability.
Authors: Hanze Li, Xiande Huang
Abstract: Growing evidence suggests that layer attention mechanisms, which enhance interaction among layers in deep neural networks, have significantly advanced network architectures. However, existing layer attention methods suffer from redundancy, as attention weights learned by adjacent layers often become highly similar. This redundancy causes multiple layers to extract nearly identical features, reducing the model's representational capacity and increasing training time. To address this issue, we propose a novel approach to quantify redundancy by leveraging the Kullback-Leibler (KL) divergence between adjacent layers. Additionally, we introduce an Enhanced Beta Quantile Mapping (EBQM) method that accurately identifies and skips redundant layers, thereby maintaining model stability. Our proposed Efficient Layer Attention (ELA) architecture, improves both training efficiency and overall performance, achieving a 30\% reduction in training time while enhancing performance in tasks such as image classification and object detection.
Authors: Chuheng Wei, Ziye Qin, Siyan Li, Ziyan Zhang, Xuanpeng Zhao, Amr Abdelraouf, Rohit Gupta, Kyungtae Han, Matthew J. Barth, Guoyuan Wu
Abstract: Driving behavior is inherently personal, influenced by individual habits, decision-making styles, and physiological states. However, most existing datasets treat all drivers as homogeneous, overlooking driver-specific variability. To address this gap, we introduce the Personalized Driving Behavior (PDB) dataset, a multi-modal dataset designed to capture personalization in driving behavior under naturalistic driving conditions. Unlike conventional datasets, PDB minimizes external influences by maintaining consistent routes, vehicles, and lighting conditions across sessions. It includes sources from 128-line LiDAR, front-facing camera video, GNSS, 9-axis IMU, CAN bus data (throttle, brake, steering angle), and driver-specific signals such as facial video and heart rate. The dataset features 12 participants, approximately 270,000 LiDAR frames, 1.6 million images, and 6.6 TB of raw sensor data. The processed trajectory dataset consists of 1,669 segments, each spanning 10 seconds with a 0.2-second interval. By explicitly capturing drivers' behavior, PDB serves as a unique resource for human factor analysis, driver identification, and personalized mobility applications, contributing to the development of human-centric intelligent transportation systems.
Authors: Honglin Li, Zhongyi Shui, Yunlong Zhang, Chenglu Zhu, Lin Yang
Abstract: Computational pathology and whole-slide image (WSI) analysis are pivotal in cancer diagnosis and prognosis. However, the ultra-high resolution of WSIs presents significant modeling challenges. Recent advancements in pathology foundation models have improved performance, yet most approaches rely on [CLS] token representation of tile ViT as slide-level inputs (16x16 pixels is refereed as patch and 224x224 pixels as tile). This discards critical spatial details from patch tokens, limiting downstream WSI analysis tasks. We find that leveraging all spatial patch tokens benefits WSI analysis but incurs nearly 200x higher storage and training costs (e.g., 196 tokens in ViT$_{224}$). To address this, we introduce vector quantized (VQ) distillation on patch feature, which efficiently compresses spatial patch tokens using discrete indices and a decoder. Our method reduces token dimensionality from 1024 to 16, achieving a 64x compression rate while preserving reconstruction fidelity. Furthermore, we employ a multi-scale VQ (MSVQ) strategy, which not only enhances VQ reconstruction performance but also serves as a Self-supervised Learning (SSL) supervision for a seamless slide-level pretraining objective. Built upon the quantized patch features and supervision targets of tile via MSVQ, we develop a progressive convolutional module and slide-level SSL to extract representations with rich spatial-information for downstream WSI tasks. Extensive evaluations on multiple datasets demonstrate the effectiveness of our approach, achieving state-of-the-art performance in WSI analysis. Code will be available soon.
Authors: Xiao Wang, Yuehang Li, Fuling Wang, Bo Jiang, Yaowei Wang, Yonghong Tian, Jin Tang, Bin Luo
Abstract: Accurate sign language understanding serves as a crucial communication channel for individuals with disabilities. Current sign language translation algorithms predominantly rely on RGB frames, which may be limited by fixed frame rates, variable lighting conditions, and motion blur caused by rapid hand movements. Inspired by the recent successful application of event cameras in other fields, we propose to leverage event streams to assist RGB cameras in capturing gesture data, addressing the various challenges mentioned above. Specifically, we first collect a large-scale RGB-Event sign language translation dataset using the DVS346 camera, termed VECSL, which contains 15,676 RGB-Event samples, 15,191 glosses, and covers 2,568 Chinese characters. These samples were gathered across a diverse range of indoor and outdoor environments, capturing multiple viewing angles, varying light intensities, and different camera motions. Due to the absence of benchmark algorithms for comparison in this new task, we retrained and evaluated multiple state-of-the-art SLT algorithms, and believe that this benchmark can effectively support subsequent related research. Additionally, we propose a novel RGB-Event sign language translation framework (i.e., M$^2$-SLT) that incorporates fine-grained micro-sign and coarse-grained macro-sign retrieval, achieving state-of-the-art results on the proposed dataset. Both the source code and dataset will be released on https://github.com/Event-AHU/OpenESL.
Authors: Jiajie Fan, Amal Trigui, Andrea Bonfanti, Felix Dietrich, Thomas B\"ack, Hao Wang
Abstract: Recent advancements in learning latent codes derived from high-dimensional shapes have demonstrated impressive outcomes in 3D generative modeling. Traditionally, these approaches employ a trained autoencoder to acquire a continuous implicit representation of source shapes, which can be computationally expensive. This paper introduces a novel framework, spectral-domain diffusion for high-quality shape generation SpoDify, that utilizes singular value decomposition (SVD) for shape encoding. The resulting eigenvectors can be stored for subsequent decoding, while generative modeling is performed on the eigenfeatures. This approach efficiently encodes complex meshes into continuous implicit representations, such as encoding a 15k-vertex mesh to a 512-dimensional latent code without learning. Our method exhibits significant advantages in scenarios with limited samples or GPU resources. In mesh generation tasks, our approach produces high-quality shapes that are comparable to state-of-the-art methods.
Authors: Cong Chen, Mingyu Liu, Chenchen Jing, Yizhou Zhou, Fengyun Rao, Hao Chen, Bo Zhang, Chunhua Shen
Abstract: This paper aims to address the challenge of hallucinations in Multimodal Large Language Models (MLLMs) particularly for dense image captioning tasks. To tackle the challenge, we identify the current lack of a metric that finely measures the caption quality in concept level. We hereby introduce HalFscore, a novel metric built upon the language graph and is designed to evaluate both the accuracy and completeness of dense captions at a granular level. Additionally, we identify the root cause of hallucination as the model's over-reliance on its language prior. To address this, we propose PerturboLLaVA, which reduces the model's reliance on the language prior by incorporating adversarially perturbed text during training. This method enhances the model's focus on visual inputs, effectively reducing hallucinations and producing accurate, image-grounded descriptions without incurring additional computational overhead. PerturboLLaVA significantly improves the fidelity of generated captions, outperforming existing approaches in handling multimodal hallucinations and achieving improved performance across general multimodal benchmarks.
Authors: Enming Zhang, Peizhe Gong, Xingyuan Dai, Yisheng Lv, Qinghai Miao
Abstract: Assessing the safety of vision-language models (VLMs) in autonomous driving is particularly important; however, existing work mainly focuses on traditional benchmark evaluations. As interactive components within autonomous driving systems, VLMs must maintain strong safety cognition during interactions. From this perspective, we propose a novel evaluation method: Safety Cognitive Driving Benchmark (SCD-Bench) . To address the large-scale annotation challenge for SCD-Bench, we develop the Autonomous Driving Image-Text Annotation System (ADA) . Additionally, to ensure data quality in SCD-Bench, our dataset undergoes manual refinement by experts with professional knowledge in autonomous driving. We further develop an automated evaluation method based on large language models (LLMs). To verify its effectiveness, we compare its evaluation results with those of expert human evaluations, achieving a consistency rate of 99.74%. Preliminary experimental results indicate that existing open-source models still lack sufficient safety cognition, showing a significant gap compared to GPT-4o. Notably, lightweight models (1B-4B) demonstrate minimal safety cognition. However, since lightweight models are crucial for autonomous driving systems, this presents a significant challenge for integrating VLMs into the field.
Authors: Xukun Zhou, Fengxin Li, Ming Chen, Yan Zhou, Pengfei Wan, Di Zhang, Hongyan Liu, Jun He, Zhaoxin Fan
Abstract: Audio-driven human gesture synthesis is a crucial task with broad applications in virtual avatars, human-computer interaction, and creative content generation. Despite notable progress, existing methods often produce gestures that are coarse, lack expressiveness, and fail to fully align with audio semantics. To address these challenges, we propose ExGes, a novel retrieval-enhanced diffusion framework with three key designs: (1) a Motion Base Construction, which builds a gesture library using training dataset; (2) a Motion Retrieval Module, employing constrative learning and momentum distillation for fine-grained reference poses retreiving; and (3) a Precision Control Module, integrating partial masking and stochastic masking to enable flexible and fine-grained control. Experimental evaluations on BEAT2 demonstrate that ExGes reduces Fr\'echet Gesture Distance by 6.2\% and improves motion diversity by 5.3\% over EMAGE, with user studies revealing a 71.3\% preference for its naturalness and semantic relevance. Code will be released upon acceptance.
Authors: Huaqi Tao, Bingxi Liu, Calvin Chen, Tingjun Huang, He Li, Jinqiang Cui, Hong Zhang
Abstract: Visual Place Recognition (VPR) is a crucial capability for long-term autonomous robots, enabling them to identify previously visited locations using visual information. However, existing methods remain limited in indoor settings due to the highly repetitive structures inherent in such environments. We observe that scene text typically appears in indoor spaces, serving to distinguish visually similar but different places. This inspires us to propose TextInPlace, a simple yet effective VPR framework that integrates Scene Text Spotting (STS) to mitigate visual perceptual ambiguity in repetitive indoor environments. Specifically, TextInPlace adopts a dual-branch architecture within a local parameter sharing network. The VPR branch employs attention-based aggregation to extract global descriptors for coarse-grained retrieval, while the STS branch utilizes a bridging text spotter to detect and recognize scene text. Finally, the discriminative text is filtered to compute text similarity and re-rank the top-K retrieved images. To bridge the gap between current text-based repetitive indoor scene datasets and the typical scenarios encountered in robot navigation, we establish an indoor VPR benchmark dataset, called Maze-with-Text. Extensive experiments on both custom and public datasets demonstrate that TextInPlace achieves superior performance over existing methods that rely solely on appearance information. The dataset, code, and trained models are publicly available at https://github.com/HqiTao/TextInPlace.
Authors: Xirui Hu, Jiahao Wang, Hao Chen, Weizhan Zhang, Benqi Wang, Yikun Li, Haishun Nan
Abstract: Recent advancements in text-to-image generation have spurred interest in personalized human image generation, which aims to create novel images featuring specific human identities as reference images indicate. Although existing methods achieve high-fidelity identity preservation, they often struggle with limited multi-ID usability and inadequate facial editability. We present DynamicID, a tuning-free framework supported by a dual-stage training paradigm that inherently facilitates both single-ID and multi-ID personalized generation with high fidelity and flexible facial editability. Our key innovations include: 1) Semantic-Activated Attention (SAA), which employs query-level activation gating to minimize disruption to the original model when injecting ID features and achieve multi-ID personalization without requiring multi-ID samples during training. 2) Identity-Motion Reconfigurator (IMR), which leverages contrastive learning to effectively disentangle and re-entangle facial motion and identity features, thereby enabling flexible facial editing. Additionally, we have developed a curated VariFace-10k facial dataset, comprising 10k unique individuals, each represented by 35 distinct facial images. Experimental results demonstrate that DynamicID outperforms state-of-the-art methods in identity fidelity, facial editability, and multi-ID personalization capability.
Authors: Amir Mohammad Izadi, Seyed Mohammad Hadi Hosseini, Soroush Vafaie Tabar, Ali Abdollahi, Armin Saghafian, Mahdieh Soleymani Baghshah
Abstract: Text-to-image generative models have made significant advancements in recent years; however, accurately capturing intricate details in textual prompts, such as entity missing, attribute binding errors, and incorrect relationships remains a formidable challenge. In response, we present an innovative, training-free method that directly addresses these challenges by incorporating tailored objectives to account for textual constraints. Unlike layout-based approaches that enforce rigid structures and limit diversity, our proposed approach offers a more flexible arrangement of the scene by imposing just the extracted constraints from the text, without any unnecessary additions. These constraints are formulated as losses-entity missing, entity mixing, attribute binding, and spatial relationships, integrated into a unified loss that is applied in the first generation stage. Furthermore, we introduce a feedback-driven system for fine-grained initial noise refinement. This system integrates a verifier that evaluates the generated image, identifies inconsistencies, and provides corrective feedback. Leveraging this feedback, our refinement method first targets the unmet constraints by refining the faulty attention maps caused by initial noise, through the optimization of selective losses associated with these constraints. Subsequently, our unified loss function is reapplied to proceed the second generation phase. Experimental results demonstrate that our method, relying solely on our proposed objective functions, significantly enhances compositionality, achieving a 24% improvement in human evaluation and a 25% gain in spatial relationships. Furthermore, our fine-grained noise refinement proves effective, boosting performance by up to 5%. Code is available at https://github.com/hadi-hosseini/noise-refinement.
Authors: Quanjian Song, Zhihang Lin, Zhanpeng Zeng, Ziyue Zhang, Liujuan Cao, Rongrong Ji
Abstract: Existing camera motion-controlled video generation methods face computational bottlenecks in fine-tuning and inference. This paper proposes LightMotion, a light and tuning-free method for simulating camera motion in video generation. Operating in the latent space, it eliminates additional fine-tuning, inpainting, and depth estimation, making it more streamlined than existing methods. The endeavors of this paper comprise: (i) The latent space permutation operation effectively simulates various camera motions like panning, zooming, and rotation. (ii) The latent space resampling strategy combines background-aware sampling and cross-frame alignment to accurately fill new perspectives while maintaining coherence across frames. (iii) Our in-depth analysis shows that the permutation and resampling cause an SNR shift in latent space, leading to poor-quality generation. To address this, we propose latent space correction, which reintroduces noise during denoising to mitigate SNR shift and enhance video generation quality. Exhaustive experiments show that our LightMotion outperforms existing methods, both quantitatively and qualitatively.
Authors: Jing Zhang, Zhikai Li, Qingyi Gu
Abstract: Segment Anything Model (SAM) exhibits remarkable zero-shot segmentation capability; however, its prohibitive computational costs make edge deployment challenging. Although post-training quantization (PTQ) offers a promising compression solution, existing methods yield unsatisfactory results when applied to SAM, owing to its specialized model components and promptable workflow: (i) The mask decoder's attention exhibits extreme outliers, and we find that aggressive clipping (ranging down to even 100$\times$), instead of smoothing or isolation, is effective in suppressing outliers while maintaining semantic capabilities. Unfortunately, traditional metrics (e.g., MSE) fail to provide such large-scale clipping. (ii) Existing reconstruction methods potentially neglect prompts' intention, resulting in distorted visual encodings during prompt interactions. To address the above issues, we propose SAQ-SAM in this paper, which boosts PTQ of SAM with semantic alignment. Specifically, we propose Perceptual-Consistency Clipping, which exploits attention focus overlap as clipping metric, to significantly suppress outliers. Furthermore, we propose Prompt-Aware Reconstruction, which incorporates visual-prompt interactions by leveraging cross-attention responses in mask decoder, thus facilitating alignment in both distribution and semantics. To ensure the interaction efficiency, we also introduce a layer-skipping strategy for visual tokens. Extensive experiments are conducted on different segmentation tasks and SAMs of various sizes, and the results show that the proposed SAQ-SAM consistently outperforms baselines. For example, when quantizing SAM-B to 4-bit, our method achieves 11.7% higher mAP than the baseline in instance segmentation task.
Authors: Shinnosuke Matsuo, Riku Togashi, Ryoma Bise, Seiichi Uchida, Masahiro Nomura
Abstract: Active learning (AL) is a label-efficient machine learning paradigm that focuses on selectively annotating high-value instances to maximize learning efficiency. Its effectiveness can be further enhanced by incorporating weak supervision, which uses rough yet cost-effective annotations instead of exact (i.e., full) but expensive annotations. We introduce a novel AL framework, Instance-wise Supervision-Level Optimization (ISO), which not only selects the instances to annotate but also determines their optimal annotation level within a fixed annotation budget. Its optimization criterion leverages the value-to-cost ratio (VCR) of each instance while ensuring diversity among the selected instances. In classification experiments, ISO consistently outperforms traditional AL methods and surpasses a state-of-the-art AL approach that combines full and weak supervision, achieving higher accuracy at a lower overall cost. This code is available at https://github.com/matsuo-shinnosuke/ISOAL.
Authors: Yuqi Liu, Bohao Peng, Zhisheng Zhong, Zihao Yue, Fanbin Lu, Bei Yu, Jiaya Jia
Abstract: Traditional methods for reasoning segmentation rely on supervised fine-tuning with categorical labels and simple descriptions, limiting its out-of-domain generalization and lacking explicit reasoning processes. To address these limitations, we propose Seg-Zero, a novel framework that demonstrates remarkable generalizability and derives explicit chain-of-thought reasoning through cognitive reinforcement. Seg-Zero introduces a decoupled architecture consisting of a reasoning model and a segmentation model. The reasoning model interprets user intentions, generates explicit reasoning chains, and produces positional prompts, which are subsequently used by the segmentation model to generate precious pixel-level masks. We design a sophisticated reward mechanism that integrates both format and accuracy rewards to effectively guide optimization directions. Trained exclusively via reinforcement learning with GRPO and without explicit reasoning data, Seg-Zero achieves robust zero-shot generalization and exhibits emergent test-time reasoning capabilities. Experiments show that Seg-Zero-7B achieves a zero-shot performance of 57.5 on the ReasonSeg benchmark, surpassing the prior LISA-7B by 18\%. This significant improvement highlights Seg-Zero's ability to generalize across domains while presenting an explicit reasoning process. Code is available at https://github.com/dvlab-research/Seg-Zero.
Authors: Yuchen Yang, Wei Wang, Yifei Liu, Linfeng Dong, Hao Wu, Mingxin Zhang, Zhihang Zhong, Xiao Sun
Abstract: Group Activity Understanding is predominantly studied as Group Activity Recognition (GAR) task. However, existing GAR benchmarks suffer from coarse-grained activity vocabularies and the only data form in single-view, which hinder the evaluation of state-of-the-art algorithms. To address these limitations, we introduce SGA-INTERACT, the first 3D skeleton-based benchmark for group activity understanding. It features complex activities inspired by basketball tactics, emphasizing rich spatial interactions and long-term dependencies. SGA-INTERACT introduces Temporal Group Activity Localization (TGAL) task, extending group activity understanding to untrimmed sequences, filling the gap left by GAR as a standalone task. In addition to the benchmark, we propose One2Many, a novel framework that employs a pretrained 3D skeleton backbone for unified individual feature extraction. This framework aligns with the feature extraction paradigm in RGB-based methods, enabling direct evaluation of RGB-based models on skeleton-based benchmarks. We conduct extensive evaluations on SGA-INTERACT using two skeleton-based methods, three RGB-based methods, and a proposed baseline within the One2Many framework. The general low performance of baselines highlights the benchmark's challenges, motivating advancements in group activity understanding.
Authors: Chen-Lin Zhang, Lin Sui, Shuming Liu, Fangzhou Mu, Zhangcheng Wang, Bernard Ghanem
Abstract: Temporal localization in untrimmed videos, which aims to identify specific timestamps, is crucial for video understanding but remains challenging. This task encompasses several subtasks, including temporal action localization, temporal video grounding, moment retrieval, and generic event boundary detection. Existing methods in each subfield are typically designed for specific tasks and lack generalizability across domains. In this paper, we propose TimeLoc, a unified end-to-end framework for timestamp localization that can handle multiple tasks. First, our approach employs a simple yet effective one-stage localization model that supports text queries as input and multiple actions as output. Second, we jointly train the video encoder and localization model in an end-to-end manner. To efficiently process long videos, we introduce temporal chunking, enabling the handling of videos with over 30k frames. Third, we find that fine-tuning pre-trained text encoders with a multi-stage training strategy further enhances text-conditioned localization. TimeLoc achieves state-of-the-art results across multiple benchmarks: +1.3% and +1.9% mAP over previous best methods on THUMOS14 and EPIC-Kitchens-100, +1.1% on Kinetics-GEBD, +2.94% mAP on QVHighlights, and significant improvements in temporal video grounding (+11.5% on TACoS and +6.7% on Charades-STA under R1@0.5). Our code and checkpoints will be released at https://github.com/sming256/TimeLoc.
Authors: Xiaoyang Liu, Yuquan Wang, Zheng Chen, Jiezhang Cao, He Zhang, Yulun Zhang, Xiaokang Yang
Abstract: Currently, methods for single-image deblurring based on CNNs and transformers have demonstrated promising performance. However, these methods often suffer from perceptual limitations, poor generalization ability, and struggle with heavy or complex blur. While diffusion-based methods can partially address these shortcomings, their multi-step denoising process limits their practical usage. In this paper, we conduct an in-depth exploration of diffusion models in deblurring and propose a one-step diffusion model for deblurring (OSDD), a novel framework that reduces the denoising process to a single step, significantly improving inference efficiency while maintaining high fidelity. To tackle fidelity loss in diffusion models, we introduce an enhanced variational autoencoder (eVAE), which improves structural restoration. Additionally, we construct a high-quality synthetic deblurring dataset to mitigate perceptual collapse and design a dynamic dual-adapter (DDA) to enhance perceptual quality while preserving fidelity. Extensive experiments demonstrate that our method achieves strong performance on both full and no-reference metrics. Our code and pre-trained model will be publicly available at https://github.com/xyLiu339/OSDD.
Authors: Jianwen Sun, Yukang Feng, Chuanhao Li, Fanrui Zhang, Zizhen Li, Jiaxin Ai, Sizhuo Zhou, Yu Dai, Shenglin Zhang, Kaipeng Zhang
Abstract: Unified models (UniMs) for multimodal understanding and generation have recently received much attention in the area of vision and language. Existing UniMs are designed to simultaneously learn both multimodal understanding and generation capabilities, demanding substantial computational resources, and often struggle to generate interleaved text-image. We present ARMOR, a resource-efficient and pure autoregressive framework that achieves both understanding and generation by fine-tuning existing multimodal large language models (MLLMs). Specifically, ARMOR extends existing MLLMs from three perspectives: (1) For model architecture, an asymmetric encoder-decoder architecture with a forward-switching mechanism is introduced to unify embedding space integrating textual and visual modalities for enabling natural text-image interleaved generation with minimal computational overhead. (2) For training data, a meticulously curated, high-quality interleaved dataset is collected for fine-tuning MLLMs. (3) For the training algorithm, we propose a ``what or how to generate" algorithm to empower existing MLLMs with multimodal generation capabilities while preserving their multimodal understanding capabilities, through three progressive training stages based on the collected dataset. Experimental results demonstrate that ARMOR upgrades existing MLLMs to UniMs with promising image generation capabilities, using limited training resources. Our code will be released soon at https://armor.github.io.
URLs: https://armor.github.io.
Authors: Junyi Wu, Zhiteng Li, Zheng Hui, Yulun Zhang, Linghe Kong, Xiaokang Yang
Abstract: Recently, Diffusion Transformers (DiTs) have emerged as a dominant architecture in video generation, surpassing U-Net-based models in terms of performance. However, the enhanced capabilities of DiTs come with significant drawbacks, including increased computational and memory costs, which hinder their deployment on resource-constrained devices. Current acceleration techniques, such as quantization and cache mechanism, offer limited speedup and are often applied in isolation, failing to fully address the complexities of DiT architectures. In this paper, we propose QuantCache, a novel training-free inference acceleration framework that jointly optimizes hierarchical latent caching, adaptive importance-guided quantization, and structural redundancy-aware pruning. QuantCache achieves an end-to-end latency speedup of 6.72$\times$ on Open-Sora with minimal loss in generation quality. Extensive experiments across multiple video generation benchmarks demonstrate the effectiveness of our method, setting a new standard for efficient DiT inference. The code and models will be available at https://github.com/JunyiWuCode/QuantCache.
Authors: Yuzheng Wang, Zhaoyu Chen, Dingkang Yang, Yuanhang Wang, Lizhe Qi
Abstract: Adversarial Robustness Distillation (ARD) is a promising task to boost the robustness of small-capacity models with the guidance of the pre-trained robust teacher. The ARD can be summarized as a min-max optimization process, i.e., synthesizing adversarial examples (inner) & training the student (outer). Although competitive robustness performance, existing ARD methods still have issues. In the inner process, the synthetic training examples are far from the teacher's decision boundary leading to important robust information missing. In the outer process, the student model is decoupled from learning natural and robust scenarios, leading to the robustness saturation, i.e., student performance is highly susceptible to customized teacher selection. To tackle these issues, this paper proposes a general Min-Max optimization Adversarial Robustness Distillation (MMARD) method. For the inner process, we introduce the teacher's robust predictions, which drive the training examples closer to the teacher's decision boundary to explore more robust knowledge. For the outer process, we propose a structured information modeling method based on triangular relationships to measure the mutual information of the model in natural and robust scenarios and enhance the model's ability to understand multi-scenario mapping relationships. Experiments show our MMARD achieves state-of-the-art performance on multiple benchmarks. Besides, MMARD is plug-and-play and convenient to combine with existing methods.
Authors: Yihua Shao, Deyang Lin, Fanhu Zeng, Minxi Yan, Muyang Zhang, Siyu Chen, Yuxuan Fan, Ziyang Yan, Haozhe Wang, Jingcai Guo, Yan Wang, Haotong Qin, Hao Tang
Abstract: Diffusion models have been widely adopted in image and video generation. However, their complex network architecture leads to high inference overhead for its generation process. Existing diffusion quantization methods primarily focus on the quantization of the model structure while ignoring the impact of time-steps variation during sampling. At the same time, most current approaches fail to account for significant activations that cannot be eliminated, resulting in substantial performance degradation after quantization. To address these issues, we propose Time-Rotation Diffusion Quantization (TR-DQ), a novel quantization method incorporating time-step and rotation-based optimization. TR-DQ first divides the sampling process based on time-steps and applies a rotation matrix to smooth activations and weights dynamically. For different time-steps, a dedicated hyperparameter is introduced for adaptive timing modeling, which enables dynamic quantization across different time steps. Additionally, we also explore the compression potential of Classifier-Free Guidance (CFG-wise) to establish a foundation for subsequent work. TR-DQ achieves state-of-the-art (SOTA) performance on image generation and video generation tasks and a 1.38-1.89x speedup and 1.97-2.58x memory reduction in inference compared to existing quantization methods.
Authors: Shijie Li, Xun Xu, Si Yong Yeo, Xulei Yang
Abstract: Motion forecasting is a crucial component of autonomous driving systems, enabling the generation of accurate and smooth future trajectories to ensure safe navigation to the destination. In previous methods, potential future trajectories are often absent in the scene encoding stage, which may lead to suboptimal outcomes. Additionally, prior approaches typically employ transformer architectures for spatiotemporal modeling of trajectories and map information, which suffer from the quadratic scaling complexity of the transformer architecture. In this work, we propose an interaction-based method, named Future-Aware Interaction Network, that introduces potential future trajectories into scene encoding for a comprehensive traffic representation. Furthermore, a State Space Model (SSM), specifically Mamba, is introduced for both spatial and temporal modeling. To adapt Mamba for spatial interaction modeling, we propose an adaptive reordering strategy that transforms unordered data into a structured sequence. Additionally, Mamba is employed to refine generated future trajectories temporally, ensuring more consistent predictions. These enhancements not only improve model efficiency but also enhance the accuracy and diversity of predictions. We conduct comprehensive experiments on the widely used Argoverse 1 and Argoverse 2 datasets, demonstrating that the proposed method achieves superior performance compared to previous approaches in a more efficient way. The code will be released according to the acceptance.
Authors: Qiyuan He, Angela Yao
Abstract: Personalized image generation with text-to-image diffusion models generates unseen images based on reference image content. Zero-shot adapter methods such as IP-Adapter and OminiControl are especially interesting because they do not require test-time fine-tuning. However, they struggle to balance preserving personalized content and adherence to the text prompt. We identify a critical design flaw resulting in this performance gap: current adapters inadequately integrate personalization images with the textual descriptions. The generated images, therefore, replicate the personalized content rather than adhere to the text prompt instructions. Yet the base text-to-image has strong conceptual understanding capabilities that can be leveraged. We propose Conceptrol, a simple yet effective framework that enhances zero-shot adapters without adding computational overhead. Conceptrol constrains the attention of visual specification with a textual concept mask that improves subject-driven generation capabilities. It achieves as much as 89% improvement on personalization benchmarks over the vanilla IP-Adapter and can even outperform fine-tuning approaches such as Dreambooth LoRA. The source code is available at https://github.com/QY-H00/Conceptrol.
Authors: Shijie Li, Zhongyao Cheng, Rong Li, Shuai Li, Juergen Gall, Xun Xu, Xulei Yang
Abstract: Monocular Semantic Scene Completion (MonoSSC) reconstructs and interprets 3D environments from a single image, enabling diverse real-world applications. However, existing methods are often constrained by the local receptive field of Convolutional Neural Networks (CNNs), making it challenging to handle the non-uniform distribution of projected points (Fig. \ref{fig:perspective}) and effectively reconstruct missing information caused by the 3D-to-2D projection. In this work, we introduce GA-MonoSSC, a hybrid architecture for MonoSSC that effectively captures global context in both the 2D image domain and 3D space. Specifically, we propose a Dual-Head Multi-Modality Encoder, which leverages a Transformer architecture to capture spatial relationships across all features in the 2D image domain, enabling more comprehensive 2D feature extraction. Additionally, we introduce the Frustum Mamba Decoder, built on the State Space Model (SSM), to efficiently capture long-range dependencies in 3D space. Furthermore, we propose a frustum reordering strategy within the Frustum Mamba Decoder to mitigate feature discontinuities in the reordered voxel sequence, ensuring better alignment with the scan mechanism of the State Space Model (SSM) for improved 3D representation learning. We conduct extensive experiments on the widely used Occ-ScanNet and NYUv2 datasets, demonstrating that our proposed method achieves state-of-the-art performance, validating its effectiveness. The code will be released upon acceptance.
Authors: Xiaoming Peng, Yixin Yang, Yang Zhou, Hui Huang
Abstract: Recently, 2D Gaussian Splatting (2DGS) has demonstrated superior geometry reconstruction quality than the popular 3DGS by using 2D surfels to approximate thin surfaces. However, it falls short when dealing with glossy surfaces, resulting in visible holes in these areas. We found the reflection discontinuity causes the issue. To fit the jump from diffuse to specular reflection at different viewing angles, depth bias is introduced in the optimized Gaussian primitives. To address that, we first replace the depth distortion loss in 2DGS with a novel depth convergence loss, which imposes a strong constraint on depth continuity. Then, we rectified the depth criterion in determining the actual surface, which fully accounts for all the intersecting Gaussians along the ray. Qualitative and quantitative evaluations across various datasets reveal that our method significantly improves reconstruction quality, with more complete and accurate surfaces than 2DGS.
Authors: Hao Xu, Tengfei Xue, Dongnan Liu, Yuqian Chen, Fan Zhang, Carl-Fredrik Westin, Ron Kikinis, Lauren J. O'Donnell, Weidong Cai
Abstract: 3D neuroimages provide a comprehensive view of brain structure and function, aiding in precise localization and functional connectivity analysis. Segmentation of white matter (WM) tracts using 3D neuroimages is vital for understanding the brain's structural connectivity in both healthy and diseased states. One-shot Class Incremental Semantic Segmentation (OCIS) refers to effectively segmenting new (novel) classes using only a single sample while retaining knowledge of old (base) classes without forgetting. Voxel-contrastive OCIS methods adjust the feature space to alleviate the feature overlap problem between the base and novel classes. However, since WM tract segmentation is a multi-label segmentation task, existing single-label voxel contrastive-based methods may cause inherent contradictions. To address this, we propose a new multi-label voxel contrast framework called MultiCo3D for one-shot class incremental tract segmentation. Our method utilizes uncertainty distillation to preserve base tract segmentation knowledge while adjusting the feature space with multi-label voxel contrast to alleviate feature overlap when learning novel tracts and dynamically weighting multi losses to balance overall loss. We compare our method against several state-of-the-art (SOTA) approaches. The experimental results show that our method significantly enhances one-shot class incremental tract segmentation accuracy across five different experimental setups on HCP and Preto datasets.
Authors: Yanqing Shen, Sanping Zhou, Jingwen Fu, Ruotong Wang, Shitao Chen, Nanning Zheng
Abstract: Visual place recognition is a challenging task for autonomous driving and robotics, which is usually considered as an image retrieval problem. A commonly used two-stage strategy involves global retrieval followed by re-ranking using patch-level descriptors. Most deep learning-based methods in an end-to-end manner cannot extract global features with sufficient semantic information from RGB images. In contrast, re-ranking can utilize more explicit structural and semantic information in one-to-one matching process, but it is time-consuming. To bridge the gap between global retrieval and re-ranking and achieve a good trade-off between accuracy and efficiency, we propose StructVPR++, a framework that embeds structural and semantic knowledge into RGB global representations via segmentation-guided distillation. Our key innovation lies in decoupling label-specific features from global descriptors, enabling explicit semantic alignment between image pairs without requiring segmentation during deployment. Furthermore, we introduce a sample-wise weighted distillation strategy that prioritizes reliable training pairs while suppressing noisy ones. Experiments on four benchmarks demonstrate that StructVPR++ surpasses state-of-the-art global methods by 5-23% in Recall@1 and even outperforms many two-stage approaches, achieving real-time efficiency with a single RGB input.
Authors: Renhao Lu
Abstract: Semantic segmentation is a core task in computer vision with applications in biomedical imaging, remote sensing, and autonomous driving. While standard loss functions such as cross-entropy and Dice loss perform well in general cases, they often struggle with fine structures, particularly in tasks involving thin structures or closely packed objects. Various weight map-based loss functions have been proposed to address this issue by assigning higher loss weights to pixels prone to misclassification. However, these methods typically rely on precomputed or runtime-generated weight maps based on distance transforms, which impose significant computational costs and fail to adapt to evolving network predictions. In this paper, we propose a novel steerable pyramid-based weighted (SPW) loss function that efficiently generates adaptive weight maps. Unlike traditional boundary-aware losses that depend on static or iteratively updated distance maps, our method leverages steerable pyramids to dynamically emphasize regions across multiple frequency bands (capturing features at different scales) while maintaining computational efficiency. Additionally, by incorporating network predictions into the weight computation, our approach enables adaptive refinement during training. We evaluate our method on the SNEMI3D, GlaS, and DRIVE datasets, benchmarking it against 11 state-of-the-art loss functions. Our results demonstrate that the proposed SPW loss function achieves superior pixel precision and segmentation accuracy with minimal computational overhead. This work provides an effective and efficient solution for improving semantic segmentation, particularly for applications requiring multiscale feature representation. The code is avaiable at https://anonymous.4open.science/r/SPW-0884
Authors: Ruchi Bhatt, Shreya Bansal, Amanpreet Chander, Rupinder Kaur, Malya Singh, Mohan Kankanhalli, Abdulmotaleb El Saddik, Mukesh Kumar Saini
Abstract: Understanding plant growth dynamics is essential for applications in agriculture and plant phenotyping. We present the Growth Modelling (GroMo) challenge, which is designed for two primary tasks: (1) plant age prediction and (2) leaf count estimation, both essential for crop monitoring and precision agriculture. For this challenge, we introduce GroMo25, a dataset with images of four crops: radish, okra, wheat, and mustard. Each crop consists of multiple plants (p1, p2, ..., pn) captured over different days (d1, d2, ..., dm) and categorized into five levels (L1, L2, L3, L4, L5). Each plant is captured from 24 different angles with a 15-degree gap between images. Participants are required to perform both tasks for all four crops with these multiview images. We proposed a Multiview Vision Transformer (MVVT) model for the GroMo challenge and evaluated the crop-wise performance on GroMo25. MVVT reports an average MAE of 7.74 for age prediction and an MAE of 5.52 for leaf count. The GroMo Challenge aims to advance plant phenotyping research by encouraging innovative solutions for tracking and predicting plant growth. The GitHub repository is publicly available at https://github.com/mriglab/GroMo-Plant-Growth-Modeling-with-Multiview-Images.
URLs: https://github.com/mriglab/GroMo-Plant-Growth-Modeling-with-Multiview-Images.
Authors: Long Peng, Anran Wu, Wenbo Li, Peizhe Xia, Xueyuan Dai, Xinjie Zhang, Xin Di, Haoze Sun, Renjing Pei, Yang Wang, Yang Cao, Zheng-Jun Zha
Abstract: Arbitrary-scale super-resolution (ASSR) aims to reconstruct high-resolution (HR) images from low-resolution (LR) inputs with arbitrary upsampling factors using a single model, addressing the limitations of traditional SR methods constrained to fixed-scale factors (\textit{e.g.}, $\times$ 2). Recent advances leveraging implicit neural representation (INR) have achieved great progress by modeling coordinate-to-pixel mappings. However, the efficiency of these methods may suffer from repeated upsampling and decoding, while their reconstruction fidelity and quality are constrained by the intrinsic representational limitations of coordinate-based functions. To address these challenges, we propose a novel ContinuousSR framework with a Pixel-to-Gaussian paradigm, which explicitly reconstructs 2D continuous HR signals from LR images using Gaussian Splatting. This approach eliminates the need for time-consuming upsampling and decoding, enabling extremely fast arbitrary-scale super-resolution. Once the Gaussian field is built in a single pass, ContinuousSR can perform arbitrary-scale rendering in just 1ms per scale. Our method introduces several key innovations. Through statistical ana
Authors: Xiaohai Li, Bineng Zhong, Qihua Liang, Zhiyi Mo, Jian Nong, Shuxiang Song
Abstract: The consistency between the semantic information provided by the multi-modal reference and the tracked object is crucial for visual-language (VL) tracking. However, existing VL tracking frameworks rely on static multi-modal references to locate dynamic objects, which can lead to semantic discrepancies and reduce the robustness of the tracker. To address this issue, we propose a novel vision-language tracking framework, named DUTrack, which captures the latest state of the target by dynamically updating multi-modal references to maintain consistency. Specifically, we introduce a Dynamic Language Update Module, which leverages a large language model to generate dynamic language descriptions for the object based on visual features and object category information. Then, we design a Dynamic Template Capture Module, which captures the regions in the image that highly match the dynamic language descriptions. Furthermore, to ensure the efficiency of description generation, we design an update strategy that assesses changes in target displacement, scale, and other factors to decide on updates. Finally, the dynamic template and language descriptions that record the latest state of the target are used to update the multi-modal references, providing more accurate reference information for subsequent inference and enhancing the robustness of the tracker. DUTrack achieves new state-of-the-art performance on four mainstream vision-language and two vision-only tracking benchmarks, including LaSOT, LaSOT$_{\rm{ext}}$, TNL2K, OTB99-Lang, GOT-10K, and UAV123. Code and models are available at https://github.com/GXNU-ZhongLab/DUTrack.
Authors: Sijie Zhao, Feng Liu, Xueliang Zhang, Hao Chen, Tao Han, Junchao Gong, Ran Tao, Pengfeng Xiao, Lei Bai, Wanli Ouyang
Abstract: The increasing impact of climate change and extreme weather events has spurred growing interest in deep learning for weather research. However, existing studies often rely on weather data in pixel space, which presents several challenges such as smooth outputs in model outputs, limited applicability to a single pressure-variable subset (PVS), and high data storage and computational costs. To address these challenges, we propose a novel Weather Latent Autoencoder (WLA) that transforms weather data from pixel space to latent space, enabling efficient weather task modeling. By decoupling weather reconstruction from downstream tasks, WLA improves the accuracy and sharpness of weather task model results. The incorporated Pressure-Variable Unified Module transforms multiple PVS into a unified representation, enhancing the adaptability of the model in multiple weather scenarios. Furthermore, weather tasks can be performed in a low-storage latent space of WLA rather than a high-storage pixel space, thus significantly reducing data storage and computational costs. Through extensive experimentation, we demonstrate its superior compression and reconstruction performance, enabling the creation of the ERA5-latent dataset with unified representations of multiple PVS from ERA5 data. The compressed full PVS in the ERA5-latent dataset reduces the original 244.34 TB of data to 0.43 TB. The downstream task further demonstrates that task models can apply to multiple PVS with low data costs in latent space and achieve superior performance compared to models in pixel space. Code, ERA5-latent data, and pre-trained models are available at https://anonymous.4open.science/r/Weather-Latent-Autoencoder-8467.
URLs: https://anonymous.4open.science/r/Weather-Latent-Autoencoder-8467.
Authors: Meiyu Zeng, Xingming Liao, Canyu Chen, Nankai Lin, Zhuowei Wang, Chong Chen, Aimin Yang
Abstract: Artificial intelligence generated content (AIGC), known as DeepFakes, has emerged as a growing concern because it is being utilized as a tool for spreading disinformation. While much research exists on identifying AI-generated text and images, research on detecting AI-generated videos is limited. Existing datasets for AI-generated videos detection exhibit limitations in terms of diversity, complexity, and realism. To address these issues, this paper focuses on AI-generated videos detection and constructs a diverse dataset named Chameleon. We generate videos through multiple generation tools and various real video sources. At the same time, we preserve the videos' real-world complexity, including scene switches and dynamic perspective changes, and expand beyond face-centered detection to include human actions and environment generation. Our work bridges the gap between AI-generated dataset construction and real-world forensic needs, offering a valuable benchmark to counteract the evolving threats of AI-generated content.
Authors: Chaocan Xue, Bineng Zhong, Qihua Liang, Yaozong Zheng, Ning Li, Yuanliang Xue, Shuxiang Song
Abstract: Vision transformers (ViTs) have emerged as a popular backbone for visual tracking. However, complete ViT architectures are too cumbersome to deploy for unmanned aerial vehicle (UAV) tracking which extremely emphasizes efficiency. In this study, we discover that many layers within lightweight ViT-based trackers tend to learn relatively redundant and repetitive target representations. Based on this observation, we propose a similarity-guided layer adaptation approach to optimize the structure of ViTs. Our approach dynamically disables a large number of representation-similar layers and selectively retains only a single optimal layer among them, aiming to achieve a better accuracy-speed trade-off. By incorporating this approach into existing ViTs, we tailor previously complete ViT architectures into an efficient similarity-guided layer-adaptive framework, namely SGLATrack, for real-time UAV tracking. Extensive experiments on six tracking benchmarks verify the effectiveness of the proposed approach, and show that our SGLATrack achieves a state-of-the-art real-time speed while maintaining competitive tracking precision. Codes and models are available at https://github.com/GXNU-ZhongLab/SGLATrack.
Authors: Hasan Abed Al Kader Hammoud, Bernard Ghanem
Abstract: We propose DiffCLIP, a novel vision-language model that extends the differential attention mechanism to CLIP architectures. Differential attention was originally developed for large language models to amplify relevant context while canceling out noisy information. In this work, we integrate this mechanism into CLIP's dual encoder (image and text) framework. With minimal additional parameters, DiffCLIP achieves superior performance on image-text understanding tasks. Across zero-shot classification, retrieval, and robustness benchmarks, DiffCLIP consistently outperforms baseline CLIP models. Notably, these gains come with negligible computational overhead, demonstrating that differential attention can significantly enhance multi-modal representations without sacrificing efficiency. Code can be found at https://github.com/hammoudhasan/DiffCLIP.
Authors: Mingxiao Li, Tingyu Qu, Tinne Tuytelaars, Marie-Francine Moens
Abstract: Personalized image generation via text prompts has great potential to improve daily life and professional work by facilitating the creation of customized visual content. The aim of image personalization is to create images based on a user-provided subject while maintaining both consistency of the subject and flexibility to accommodate various textual descriptions of that subject. However, current methods face challenges in ensuring fidelity to the text prompt while not overfitting to the training data. In this work, we introduce a novel training pipeline that incorporates an attractor to filter out distractions in training images, allowing the model to focus on learning an effective representation of the personalized subject. Moreover, current evaluation methods struggle due to the lack of a dedicated test set. The evaluation set-up typically relies on the training data of the personalization task to compute text-image and image-image similarity scores, which, while useful, tend to overestimate performance. Although human evaluations are commonly used as an alternative, they often suffer from bias and inconsistency. To address these issues, we curate a diverse and high-quality test set with well-designed prompts. With this new benchmark, automatic evaluation metrics can reliably assess model performance
Authors: Lei Shi, Andreas Bulling
Abstract: We propose CLAD -- a Constrained Latent Action Diffusion model for vision-language procedure planning in instructional videos. Procedure planning is the challenging task of predicting intermediate actions given a visual observation of a start and a goal state. However, future interactive AI systems must also be able to plan procedures using multi-modal input, e.g., where visual observations are augmented with language descriptions. To tackle this vision-language procedure planning task, our method uses a Variational Autoencoder (VAE) to learn the latent representation of actions and observations as constraints and integrate them into the diffusion process. This approach exploits that the latent space of diffusion models already has semantics that can be used. We use the latent constraints to steer the diffusion model to better generate actions. We report extensive experiments on the popular CrossTask, Coin, and NIV datasets and show that our method outperforms state-of-the-art methods by a large margin. By evaluating ablated versions of our method, we further show that the proposed integration of the action and observation representations learnt in the VAE latent space is key to these performance improvements.
Authors: Shipeng Liu, Liang Zhao, Dengfeng Chen
Abstract: Unsupervised image complexity representation often suffers from bias in positive sample selection and sensitivity to image content. We propose CLICv2, a contrastive learning framework that enforces content invariance for complexity representation. Unlike CLIC, which generates positive samples via cropping-introducing positive pairs bias-our shifted patchify method applies randomized directional shifts to image patches before contrastive learning. Patches at corresponding positions serve as positive pairs, ensuring content-invariant learning. Additionally, we propose patch-wise contrastive loss, which enhances local complexity representation while mitigating content interference. In order to further suppress the interference of image content, we introduce Masked Image Modeling as an auxiliary task, but we set its modeling objective as the entropy of masked patches, which recovers the entropy of the overall image by using the information of the unmasked patches, and then obtains the global complexity perception ability. Extensive experiments on IC9600 demonstrate that CLICv2 significantly outperforms existing unsupervised methods in PCC and SRCC, achieving content-invariant complexity representation without introducing positive pairs bias.
Authors: Hassan Kazemi Tehrani, Jun Cai, Abbas Yekanlou, Sylvia Santosa
Abstract: Accurate food intake monitoring is crucial for maintaining a healthy diet and preventing nutrition-related diseases. With the diverse range of foods consumed across various cultures, classic food classification models have limitations due to their reliance on fixed-sized food datasets. Studies show that people consume only a small range of foods across the existing ones, each consuming a unique set of foods. Existing class-incremental models have low accuracy for the new classes and lack personalization. This paper introduces a personalized, class-incremental food classification model designed to overcome these challenges and improve the performance of food intake monitoring systems. Our approach adapts itself to the new array of food classes, maintaining applicability and accuracy, both for new and existing classes by using personalization. Our model's primary focus is personalization, which improves classification accuracy by prioritizing a subset of foods based on an individual's eating habits, including meal frequency, times, and locations. A modified version of DSN is utilized to expand on the appearance of new food classes. Additionally, we propose a comprehensive framework that integrates this model into a food intake monitoring system. This system analyzes meal images provided by users, makes use of a smart scale to estimate food weight, utilizes a nutrient content database to calculate the amount of each macro-nutrient, and creates a dietary user profile through a mobile application. Finally, experimental evaluations on two new benchmark datasets FOOD101-Personal and VFN-Personal, personalized versions of well-known datasets for food classification, are conducted to demonstrate the effectiveness of our model in improving the classification accuracy of both new and existing classes, addressing the limitations of both conventional and class-incremental food classification models.
Authors: Yihong Luo, Tianyang Hu, Yifan Song, Jiacheng Sun, Zhenguo Li, Jing Tang
Abstract: While diffusion distillation has enabled one-step generation through methods like Variational Score Distillation, adapting distilled models to emerging new controls -- such as novel structural constraints or latest user preferences -- remains challenging. Conventional approaches typically requires modifying the base diffusion model and redistilling it -- a process that is both computationally intensive and time-consuming. To address these challenges, we introduce Joint Distribution Matching (JDM), a novel approach that minimizes the reverse KL divergence between image-condition joint distributions. By deriving a tractable upper bound, JDM decouples fidelity learning from condition learning. This asymmetric distillation scheme enables our one-step student to handle controls unknown to the teacher model and facilitates improved classifier-free guidance (CFG) usage and seamless integration of human feedback learning (HFL). Experimental results demonstrate that JDM surpasses baseline methods such as multi-step ControlNet by mere one-step in most cases, while achieving state-of-the-art performance in one-step text-to-image synthesis through improved usage of CFG or HFL integration.
Authors: Yang Zou, Zhaoshuai Qi, Yating Liu, Zihao Xu, Weipeng Sun, Weiyi Liu, Xingyuan Li, Jiaqi Yang, Yanning Zhang
Abstract: Object pose estimation, which plays a vital role in robotics, augmented reality, and autonomous driving, has been of great interest in computer vision. Existing studies either require multi-stage pose regression or rely on 2D-3D feature matching. Though these approaches have shown promising results, they rely heavily on appearance information, requiring complex input (i.e., multi-view reference input, depth, or CAD models) and intricate pipeline (i.e., feature extraction-SfM-2D to 3D matching-PnP). We propose AxisPose, a model-free, matching-free, single-shot solution for robust 6D pose estimation, which fundamentally diverges from the existing paradigm. Unlike existing methods that rely on 2D-3D or 2D-2D matching using 3D techniques, such as SfM and PnP, AxisPose directly infers a robust 6D pose from a single view by leveraging a diffusion model to learn the latent axis distribution of objects without reference views. Specifically, AxisPose constructs an Axis Generation Module (AGM) to capture the latent geometric distribution of object axes through a diffusion model. The diffusion process is guided by injecting the gradient of geometric consistency loss into the noise estimation to maintain the geometric consistency of the generated tri-axis. With the generated tri-axis projection, AxisPose further adopts a Triaxial Back-projection Module (TBM) to recover the 6D pose from the object tri-axis. The proposed AxisPose achieves robust performance at the cross-instance level (i.e., one model for N instances) using only a single view as input without reference images, with great potential for generalization to unseen-object level.
Authors: Wenxin Ma, Xu Zhang, Qingsong Yao, Fenghe Tang, Chenxu Wu, Yingtai Li, Rui Yan, Zihang Jiang, S. Kevin Zhou
Abstract: Anomaly detection (AD) identifies outliers for applications like defect and lesion detection. While CLIP shows promise for zero-shot AD tasks due to its strong generalization capabilities, its inherent Anomaly-Unawareness leads to limited discrimination between normal and abnormal features. To address this problem, we propose Anomaly-Aware CLIP (AA-CLIP), which enhances CLIP's anomaly discrimination ability in both text and visual spaces while preserving its generalization capability. AA-CLIP is achieved through a straightforward yet effective two-stage approach: it first creates anomaly-aware text anchors to differentiate normal and abnormal semantics clearly, then aligns patch-level visual features with these anchors for precise anomaly localization. This two-stage strategy, with the help of residual adapters, gradually adapts CLIP in a controlled manner, achieving effective AD while maintaining CLIP's class knowledge. Extensive experiments validate AA-CLIP as a resource-efficient solution for zero-shot AD tasks, achieving state-of-the-art results in industrial and medical applications. The code is available at https://github.com/Mwxinnn/AA-CLIP.
Authors: Roni Goldshmidt
Abstract: Interpretability in Vision-Language Models (VLMs) is crucial for trust, debugging, and decision-making in high-stakes applications. We introduce PixelSHAP, a model-agnostic framework extending Shapley-based analysis to structured visual entities. Unlike previous methods focusing on text prompts, PixelSHAP applies to vision-based reasoning by systematically perturbing image objects and quantifying their influence on a VLM's response. PixelSHAP requires no model internals, operating solely on input-output pairs, making it compatible with open-source and commercial models. It supports diverse embedding-based similarity metrics and scales efficiently using optimization techniques inspired by Shapley-based methods. We validate PixelSHAP in autonomous driving, highlighting its ability to enhance interpretability. Key challenges include segmentation sensitivity and object occlusion. Our open-source implementation facilitates further research.
Authors: Dongheon Lee, Seokju Yun, Youngmin Ro
Abstract: In this paper, we tackle the high computational overhead of transformers for lightweight image super-resolution. (SR). Motivated by the observations of self-attention's inter-layer repetition, we introduce a convolutionized self-attention module named Convolutional Attention (ConvAttn) that emulates self-attention's long-range modeling capability and instance-dependent weighting with a single shared large kernel and dynamic kernels. By utilizing the ConvAttn module, we significantly reduce the reliance on self-attention and its involved memory-bound operations while maintaining the representational capability of transformers. Furthermore, we overcome the challenge of integrating flash attention into the lightweight SR regime, effectively mitigating self-attention's inherent memory bottleneck. We scale up window size to 32$\times$32 with flash attention rather than proposing an intricated self-attention module, significantly improving PSNR by 0.31dB on Urban100$\times$2 while reducing latency and memory usage by 16$\times$ and 12.2$\times$. Building on these approaches, our proposed network, termed Emulating Self-attention with Convolution (ESC), notably improves PSNR by 0.27 dB on Urban100$\times$4 compared to HiT-SRF, reducing the latency and memory usage by 3.7$\times$ and 6.2$\times$, respectively. Extensive experiments demonstrate that our ESC maintains the ability for long-range modeling, data scalability, and the representational power of transformers despite most self-attentions being replaced by the ConvAttn module.
Authors: Yihong Luo, Tianyang Hu, Jiacheng Sun, Yujun Cai, Jing Tang
Abstract: Accelerating diffusion model sampling is crucial for efficient AIGC deployment. While diffusion distillation methods -- based on distribution matching and trajectory matching -- reduce sampling to as few as one step, they fall short on complex tasks like text-to-image generation. Few-step generation offers a better balance between speed and quality, but existing approaches face a persistent trade-off: distribution matching lacks flexibility for multi-step sampling, while trajectory matching often yields suboptimal image quality. To bridge this gap, we propose learning few-step diffusion models by Trajectory Distribution Matching (TDM), a unified distillation paradigm that combines the strengths of distribution and trajectory matching. Our method introduces a data-free score distillation objective, aligning the student's trajectory with the teacher's at the distribution level. Further, we develop a sampling-steps-aware objective that decouples learning targets across different steps, enabling more adjustable sampling. This approach supports both deterministic sampling for superior image quality and flexible multi-step adaptation, achieving state-of-the-art performance with remarkable efficiency. Our model, TDM, outperforms existing methods on various backbones, such as SDXL and PixArt-$\alpha$, delivering superior quality and significantly reduced training costs. In particular, our method distills PixArt-$\alpha$ into a 4-step generator that outperforms its teacher on real user preference at 1024 resolution. This is accomplished with 500 iterations and 2 A800 hours -- a mere 0.01% of the teacher's training cost. In addition, our proposed TDM can be extended to accelerate text-to-video diffusion. Notably, TDM can outperform its teacher model (CogVideoX-2B) by using only 4 NFE on VBench, improving the total score from 80.91 to 81.65. Project page: https://tdm-t2x.github.io/
Authors: Chenyu Huang, Peng Ye, Xiaohui Wang, Shenghe Zheng, Biqing Qi, Lei Bai, Wanli Ouyang, Tao Chen
Abstract: With transformer-based models and the pretrain-finetune paradigm becoming mainstream, the high storage and deployment costs of individual finetuned models on multiple tasks pose critical challenges. Delta compression attempts to lower the costs by reducing the redundancy of delta parameters (i.e., the difference between the finetuned and pre-trained model weights). However, existing methods usually face problems including data accessibility and training requirements. To tackle this issue, we introduce Delta-DCT, the first data-free delta compression method inspired by classic JPEG image compression, leveraging the Discrete Cosine Transform (DCT). We first (a) group delta parameters within a layer into patches. Then we (b) assess the importance of each patch and allocate them with different quantization bit-widths. Afterwards, we (c) convert these patches to the DCT domain and conduct quantization to each patch based on the allocated bit-width. The proposed Delta-DCT does not require any training or data calibration, while achieving performance comparable to or even surpassing original finetuned models under 1-bit equivalent delta compression ratios on different kinds of models including: (1) recently-released LLMs of different sizes from 7B to 13B, (2) relatively smaller language models including RoBERTa and T5 models, (3) variants of vision transformer models, and (4) multi-modal BEiT-3 models.
Authors: Di Wu, Liu Liu, Zhou Linli, Anran Huang, Liangtu Song, Qiaojun Yu, Qi Wu, Cewu Lu
Abstract: Articulated objects, as prevalent entities in human life, their 3D representations play crucial roles across various applications. However, achieving both high-fidelity textured surface reconstruction and dynamic generation for articulated objects remains challenging for existing methods. In this paper, we present REArtGS, a novel framework that introduces additional geometric and motion constraints to 3D Gaussian primitives, enabling high-quality textured surface reconstruction and generation for articulated objects. Specifically, given multi-view RGB images of arbitrary two states of articulated objects, we first introduce an unbiased Signed Distance Field (SDF) guidance to regularize Gaussian opacity fields, enhancing geometry constraints and improving surface reconstruction quality. Then we establish deformable fields for 3D Gaussians constrained by the kinematic structures of articulated objects, achieving unsupervised generation of surface meshes in unseen states. Extensive experiments on both synthetic and real datasets demonstrate our approach achieves high-quality textured surface reconstruction for given states, and enables high-fidelity surface generation for unseen states. Codes will be released within the next four months.
Authors: Hantao Zhou, Rui Yang, Longxiang Tang, Guanyi Qin, Yan Zhang, Runze Hu, Xiu Li
Abstract: Image assessment aims to evaluate the quality and aesthetics of images and has been applied across various scenarios, such as natural and AIGC scenes. Existing methods mostly address these sub-tasks or scenes individually. While some works attempt to develop unified image assessment models, they have struggled to achieve satisfactory performance or cover a broad spectrum of assessment scenarios. In this paper, we present \textbf{Gamma}, a \textbf{G}eneric im\textbf{A}ge assess\textbf{M}ent model using \textbf{M}ixture of \textbf{A}ssessment Experts, which can effectively assess images from diverse scenes through mixed-dataset training. Achieving unified training in image assessment presents significant challenges due to annotation biases across different datasets. To address this issue, we first propose a Mixture of Assessment Experts (MoAE) module, which employs shared and adaptive experts to dynamically learn common and specific knowledge for different datasets, respectively. In addition, we introduce a Scene-based Differential Prompt (SDP) strategy, which uses scene-specific prompts to provide prior knowledge and guidance during the learning process, further boosting adaptation for various scenes. Our Gamma model is trained and evaluated on 12 datasets spanning 6 image assessment scenarios. Extensive experiments show that our unified Gamma outperforms other state-of-the-art mixed-training methods by significant margins while covering more scenes. Code: https://github.com/zht8506/Gamma.
Authors: Xuechao Zou, Yue Li, Shun Zhang, Kai Li, Shiying Wang, Pin Tao, Junliang Xing, Congyan Lang
Abstract: Remote sensing image segmentation faces persistent challenges in distinguishing morphologically similar categories and adapting to diverse scene variations. While existing methods rely on implicit representation learning paradigms, they often fail to dynamically adjust semantic embeddings according to contextual cues, leading to suboptimal performance in fine-grained scenarios such as cloud thickness differentiation. This work introduces a dynamic dictionary learning framework that explicitly models class ID embeddings through iterative refinement. The core contribution lies in a novel dictionary construction mechanism, where class-aware semantic embeddings are progressively updated via multi-stage alternating cross-attention querying between image features and dictionary embeddings. This process enables adaptive representation learning tailored to input-specific characteristics, effectively resolving ambiguities in intra-class heterogeneity and inter-class homogeneity. To further enhance discriminability, a contrastive constraint is applied to the dictionary space, ensuring compact intra-class distributions while maximizing inter-class separability. Extensive experiments across both coarse- and fine-grained datasets demonstrate consistent improvements over state-of-the-art methods, particularly in two online test benchmarks (LoveDA and UAVid). Code is available at https://anonymous.4open.science/r/D2LS-8267/.
Authors: Yanjie Pan, Qingdong He, Zhengkai Jiang, Pengcheng Xu, Chaoyi Wang, Jinlong Peng, Haoxuan Wang, Yun Cao, Zhenye Gan, Mingmin Chi, Bo Peng, Yabiao Wang
Abstract: Recent advances in diffusion-based text-to-image generation have demonstrated promising results through visual condition control. However, existing ControlNet-like methods struggle with compositional visual conditioning - simultaneously preserving semantic fidelity across multiple heterogeneous control signals while maintaining high visual quality, where they employ separate control branches that often introduce conflicting guidance during the denoising process, leading to structural distortions and artifacts in generated images. To address this issue, we present PixelPonder, a novel unified control framework, which allows for effective control of multiple visual conditions under a single control structure. Specifically, we design a patch-level adaptive condition selection mechanism that dynamically prioritizes spatially relevant control signals at the sub-region level, enabling precise local guidance without global interference. Additionally, a time-aware control injection scheme is deployed to modulate condition influence according to denoising timesteps, progressively transitioning from structural preservation to texture refinement and fully utilizing the control information from different categories to promote more harmonious image generation. Extensive experiments demonstrate that PixelPonder surpasses previous methods across different benchmark datasets, showing superior improvement in spatial alignment accuracy while maintaining high textual semantic consistency.
Authors: Zhaowei Chen, Borui Zhao, Yuchen Ge, Yuhao Chen, Renjie Song, Jiajun Liang
Abstract: Online Knowledge Distillation (OKD) methods streamline the distillation training process into a single stage, eliminating the need for knowledge transfer from a pretrained teacher network to a more compact student network. This paper presents an innovative approach to leverage intermediate spatial representations. Our analysis of the intermediate features from both teacher and student models reveals two pivotal insights: (1) the similar features between students and teachers are predominantly focused on foreground objects. (2) teacher models emphasize foreground objects more than students. Building on these findings, we propose Asymmetric Decision-Making (ADM) to enhance feature consensus learning for student models while continuously promoting feature diversity in teacher models. Specifically, Consensus Learning for student models prioritizes spatial features with high consensus relative to teacher models. Conversely, Divergence Learning for teacher models highlights spatial features with lower similarity compared to student models, indicating superior performance by teacher models in these regions. Consequently, ADM facilitates the student models to catch up with the feature learning process of the teacher models. Extensive experiments demonstrate that ADM consistently surpasses existing OKD methods across various online knowledge distillation settings and also achieves superior results when applied to offline knowledge distillation, semantic segmentation and diffusion distillation tasks.
Authors: Chenfei Liao, Xu Zheng, Yuanhuiyi Lyu, Haiwei Xue, Yihong Cao, Jiawen Wang, Kailun Yang, Xuming Hu
Abstract: Research has focused on Multi-Modal Semantic Segmentation (MMSS), where pixel-wise predictions are derived from multiple visual modalities captured by diverse sensors. Recently, the large vision model, Segment Anything Model 2 (SAM2), has shown strong zero-shot segmentation performance on both images and videos. When extending SAM2 to MMSS, two issues arise: 1. How can SAM2 be adapted to multi-modal data? 2. How can SAM2 better understand semantics? Inspired by cross-frame correlation in videos, we propose to treat multi-modal data as a sequence of frames representing the same scene. Our key idea is to ''memorize'' the modality-agnostic information and 'memorize' the semantics related to the targeted scene. To achieve this, we apply SAM2's memory mechanisms across multi-modal data to capture modality-agnostic features. Meanwhile, to memorize the semantic knowledge, we propose a training-only Semantic Prototype Memory Module (SPMM) to store category-level prototypes across training for facilitating SAM2's transition from instance to semantic segmentation. A prototypical adaptation loss is imposed between global and local prototypes iteratively to align and refine SAM2's semantic understanding. Extensive experimental results demonstrate that our proposed MemorySAM outperforms SoTA methods by large margins on both synthetic and real-world benchmarks (65.38% on DELIVER, 52.88% on MCubeS). Source code will be made publicly available.
Authors: Wentian Xu, Ziyun Liang, Harry Anthony, Yasin Ibrahim, Felix Cohen, Guang Yang, Daniel Whitehouse, David Menon, Virginia Newcombe, Konstantinos Kamnitsas
Abstract: Interactive segmentation models use real-time user interactions, such as mouse clicks, as extra inputs to dynamically refine the model predictions. After model deployment, user corrections of model predictions could be used to adapt the model to the post-deployment data distribution, countering distribution-shift and enhancing reliability. Motivated by this, we introduce an online adaptation framework that enables an interactive segmentation model to continuously learn from user interaction and improve its performance on new data distributions, as it processes a sequence of test images. We introduce the Gaussian Point Loss function to train the model how to leverage user clicks, along with a two-stage online optimization method that adapts the model using the corrected predictions generated via user interactions. We demonstrate that this simple and therefore practical approach is very effective. Experiments on 5 fundus and 4 brain MRI databases demonstrate that our method outperforms existing approaches under various data distribution shifts, including segmentation of image modalities and pathologies not seen during training.
Authors: Vsevolod Skorokhodov, Nikita Durasov, Pascal Fua
Abstract: Gaussian Splatting has become a popular technique for various 3D Computer Vision tasks, including novel view synthesis, scene reconstruction, and dynamic scene rendering. However, the challenge of natural-looking object insertion, where the object's appearance seamlessly matches the scene, remains unsolved. In this work, we propose a method, dubbed D3DR, for inserting a 3DGS-parametrized object into 3DGS scenes while correcting its lighting, shadows, and other visual artifacts to ensure consistency, a problem that has not been successfully addressed before. We leverage advances in diffusion models, which, trained on real-world data, implicitly understand correct scene lighting. After inserting the object, we optimize a diffusion-based Delta Denoising Score (DDS)-inspired objective to adjust its 3D Gaussian parameters for proper lighting correction. Utilizing diffusion model personalization techniques to improve optimization quality, our approach ensures seamless object insertion and natural appearance. Finally, we demonstrate the method's effectiveness by comparing it to existing approaches, achieving 0.5 PSNR and 0.15 SSIM improvements in relighting quality.
Authors: Rui Song, Chenwei Liang, Yan Xia, Walter Zimmer, Hu Cao, Holger Caesar, Andreas Festag, Alois Knoll
Abstract: Dynamic scene rendering opens new avenues in autonomous driving by enabling closed-loop simulations with photorealistic data, which is crucial for validating end-to-end algorithms. However, the complex and highly dynamic nature of traffic environments presents significant challenges in accurately rendering these scenes. In this paper, we introduce a novel 4D Gaussian Splatting (4DGS) approach, which incorporates context and temporal deformation awareness to improve dynamic scene rendering. Specifically, we employ a 2D semantic segmentation foundation model to self-supervise the 4D semantic features of Gaussians, ensuring meaningful contextual embedding. Simultaneously, we track the temporal deformation of each Gaussian across adjacent frames. By aggregating and encoding both semantic and temporal deformation features, each Gaussian is equipped with cues for potential deformation compensation within 3D space, facilitating a more precise representation of dynamic scenes. Experimental results show that our method improves 4DGS's ability to capture fine details in dynamic scene rendering for autonomous driving and outperforms other self-supervised methods in 4D reconstruction and novel view synthesis. Furthermore, CoDa-4DGS deforms semantic features with each Gaussian, enabling broader applications.
Authors: Ka Chun Shum, Binh-Son Hua, Duc Thanh Nguyen, Sai-Kit Yeung
Abstract: Diffusion models have shown great promise in synthesizing visually appealing images. However, it remains challenging to condition the synthesis at a fine-grained level, for instance, synthesizing image pixels following some generic color pattern. Existing image synthesis methods often produce contents that fall outside the desired pixel conditions. To address this, we introduce a novel color alignment algorithm that confines the generative process in diffusion models within a given color pattern. Specifically, we project diffusion terms, either imagery samples or latent representations, into a conditional color space to align with the input color distribution. This strategy simplifies the prediction in diffusion models within a color manifold while still allowing plausible structures in generated contents, thus enabling the generation of diverse contents that comply with the target color pattern. Experimental results demonstrate our state-of-the-art performance in conditioning and controlling of color pixels, while maintaining on-par generation quality and diversity in comparison with regular diffusion models.
Authors: Hantao Zhang, Yuhe Liu, Jiancheng Yang, Weidong Guo, Xinyuan Wang, Pascal Fua
Abstract: Accurate medical image segmentation is crucial for precise anatomical delineation. Deep learning models like U-Net have shown great success but depend heavily on large datasets and struggle with domain shifts, complex structures, and limited training samples. Recent studies have explored diffusion models for segmentation by iteratively refining masks. However, these methods still retain the conventional image-to-mask mapping, making them highly sensitive to input data, which hampers stability and generalization. In contrast, we introduce DiffAtlas, a novel generative framework that models both images and masks through diffusion during training, effectively ``GenAI-fying'' atlas-based segmentation. During testing, the model is guided to generate a specific target image-mask pair, from which the corresponding mask is obtained. DiffAtlas retains the robustness of the atlas paradigm while overcoming its scalability and domain-specific limitations. Extensive experiments on CT and MRI across same-domain, cross-modality, varying-domain, and different data-scale settings using the MMWHS and TotalSegmentator datasets demonstrate that our approach outperforms existing methods, particularly in limited-data and zero-shot modality segmentation. Code is available at https://github.com/M3DV/DiffAtlas.
Authors: Wenxuan Huang, Bohan Jia, Zijie Zhai, Shaosheng Cao, Zheyu Ye, Fei Zhao, Yao Hu, Shaohui Lin
Abstract: DeepSeek-R1-Zero has successfully demonstrated the emergence of reasoning capabilities in LLMs purely through Reinforcement Learning (RL). Inspired by this breakthrough, we explore how RL can be utilized to enhance the reasoning capability of MLLMs. However, direct training with RL struggles to activate complex reasoning capabilities such as questioning and reflection in MLLMs, due to the absence of substantial high-quality multimodal reasoning data. To address this issue, we propose the reasoning MLLM, Vision-R1, to improve multimodal reasoning capability. Specifically, we first construct a high-quality multimodal CoT dataset without human annotations by leveraging an existing MLLM and DeepSeek-R1 through modality bridging and data filtering to obtain a 200K multimodal CoT dataset, Vision-R1-cold dataset. It serves as cold-start initialization data for Vision-R1. To mitigate the optimization challenges caused by overthinking after cold start, we propose Progressive Thinking Suppression Training (PTST) strategy and employ Group Relative Policy Optimization (GRPO) with the hard formatting result reward function to gradually refine the model's ability to learn correct and complex reasoning processes on a 10K multimodal math dataset. Comprehensive experiments show our model achieves an average improvement of $\sim$6% across various multimodal math reasoning benchmarks. Vision-R1-7B achieves a 73.5% accuracy on the widely used MathVista benchmark, which is only 0.4% lower than the leading reasoning model, OpenAI O1. The datasets and code will be released in: https://github.com/Osilly/Vision-R1 .
Authors: Hung Q. Vo, Samira Zare, Son T. Ly, Lin Wang, Chika F. Ezeana, Xiaohui Yu, Kelvin K. Wong, Stephen T. C. Wong, Hien V. Nguyen
Abstract: Despite significant progress in robust deep learning techniques for mammogram breast cancer classification, their reliability in real-world clinical development settings remains uncertain. The translation of these models to clinical practice faces challenges due to variations in medical centers, imaging protocols, and patient populations. To enhance their robustness, invariant learning methods have been proposed, prioritizing causal factors over misleading features. However, their effectiveness in clinical development and impact on mammogram classification require investigation. This paper reassesses the application of invariant learning for breast cancer risk estimation based on mammograms. Utilizing diverse multi-site public datasets, it represents the first study in this area. The objective is to evaluate invariant learning's benefits in developing robust models. Invariant learning methods, including Invariant Risk Minimization and Variance Risk Extrapolation, are compared quantitatively against Empirical Risk Minimization. Evaluation metrics include accuracy, average precision, and area under the curve. Additionally, interpretability is examined through class activation maps and visualization of learned representations. This research examines the advantages, limitations, and challenges of invariant learning for mammogram classification, guiding future studies to develop generalized methods for breast cancer prediction on whole mammograms in out-of-domain scenarios.
Authors: Abdelaziz Bouzidi, Hamid Laga, Hazem Wannous
Abstract: Neural fields such as DeepSDF and Neural Radiance Fields have recently revolutionized novel-view synthesis and 3D reconstruction from RGB images and videos. However, achieving high-quality representation, reconstruction, and rendering requires deep neural networks, which are slow to train and evaluate. Although several acceleration techniques have been proposed, they often trade off speed for memory. Gaussian splatting-based methods, on the other hand, accelerate the rendering time but remain costly in terms of training speed and memory needed to store the parameters of a large number of Gaussians. In this paper, we introduce a novel neural representation that is fast, both at training and inference times, and lightweight. Our key observation is that the neurons used in traditional MLPs perform simple computations (a dot product followed by ReLU activation) and thus one needs to use either wide and deep MLPs or high-resolution and high-dimensional feature grids to parameterize complex nonlinear functions. We show in this paper that by replacing traditional neurons with Radial Basis Function (RBF) kernels, one can achieve highly accurate representation of 2D (RGB images), 3D (geometry), and 5D (radiance fields) signals with just a single layer of such neurons. The representation is highly parallelizable, operates on low-resolution feature grids, and is compact and memory-efficient. We demonstrate that the proposed novel representation can be trained for 3D geometry representation in less than 15 seconds and for novel view synthesis in less than 15 mins. At runtime, it can synthesize novel views at more than 60 fps without sacrificing quality.
Authors: Zisheng Chen, Chunwei Wang, Xiuwei Chen, Hang Xu, Jianhua Han, Xiandan Liang
Abstract: We present SemHiTok, a unified image Tokenizer via Semantic-Guided Hierarchical codebook that provides consistent discrete feature representations for multimodal understanding and generation tasks. Recently, unified multimodal large models (MLLMs) for understanding and generation have sparked exploration within research community. Previous works attempt to train a unified image tokenizer by combining loss functions for semantic feature reconstruction and pixel reconstruction. However, due to the differing levels of features prioritized by multimodal understanding and generation tasks, joint training methods face significant challenges in achieving a good trade-off. SemHiTok addresses this challenge through Semantic-Guided Hierarchical codebook which builds texture sub-codebooks on pre-trained semantic codebook. This design decouples the training of semantic reconstruction and pixel reconstruction and equips the tokenizer with low-level texture feature extraction capability without degradation of high-level semantic feature extraction ability. Our experiments demonstrate that SemHiTok achieves state-of-the-art rFID score at 256X256resolution compared to other unified tokenizers, and exhibits competitive performance on multimodal understanding and generation tasks.
Authors: Benjamin Beaudett, Shenyuan Liang, Anuj Srivastava
Abstract: Despite high-dimensionality of images, the sets of images of 3D objects have long been hypothesized to form low-dimensional manifolds. What is the nature of such manifolds? How do they differ across objects and object classes? Answering these questions can provide key insights in explaining and advancing success of machine learning algorithms in computer vision. This paper investigates dual tasks -- learning and analyzing shapes of image manifolds -- by revisiting a classical problem of manifold learning but from a novel geometrical perspective. It uses geometry-preserving transformations to map the pose image manifolds, sets of images formed by rotating 3D objects, to low-dimensional latent spaces. The pose manifolds of different objects in latent spaces are found to be nonlinear, smooth manifolds. The paper then compares shapes of these manifolds for different objects using Kendall's shape analysis, modulo rigid motions and global scaling, and clusters objects according to these shape metrics. Interestingly, pose manifolds for objects from the same classes are frequently clustered together. The geometries of image manifolds can be exploited to simplify vision and image processing tasks, to predict performances, and to provide insights into learning methods.
Authors: Yan Wang, Shijie Zhao, Kai Chen, Kexin Zhang, Junlin Li, Li Zhang
Abstract: Recent research applying text-to-image (T2I) diffusion models to real-world super-resolution (SR) has achieved remarkable success. However, fundamental misalignments between T2I and SR targets result in a dilemma between inference speed and detail fidelity. Specifically, T2I tasks prioritize multi-step inversion to synthesize coherent outputs aligned with textual prompts and shrink the latent space to reduce generating complexity. Contrariwise, SR tasks preserve most information from low-resolution input while solely restoring high-frequency details, thus necessitating sufficient latent space and fewer inference steps. To bridge the gap, we present a one-step diffusion model for generative detail restoration, GenDR, distilled from a tailored diffusion model with larger latent space. In detail, we train a new SD2.1-VAE16 (0.9B) via representation alignment to expand latent space without enlarging the model size. Regarding step-distillation, we propose consistent score identity distillation (CiD) that incorporates SR task-specific loss into score distillation to leverage more SR priors and align the training target. Furthermore, we extend CiD with adversarial learning and representation alignment (CiDA) to enhance perceptual quality and accelerate training. We also polish the pipeline to achieve a more efficient inference. Experimental results demonstrate that GenDR achieves state-of-the-art performance in both quantitative metrics and visual fidelity.
Authors: Yizheng Sun, Hao Li, Chang Xu, Chenghua Lin, Riza Batista-Navarro, Jingyuan Sun
Abstract: Vision language models (VLMs) have excelled in visual reasoning but often incur high computational costs. One key reason is the redundancy of visual tokens. Although recent token reduction methods claim to achieve minimal performance loss, our extensive experiments reveal that token reduction can substantially alter a model's output distribution, leading to changes in prediction patterns that standard metrics such as accuracy loss do not fully capture. Such inconsistencies are especially concerning for practical applications where system stability is critical. To investigate this phenomenon, we analyze how token reduction influences the energy distribution of a VLM's internal representations using a lower-rank approximation via Singular Value Decomposition (SVD). Our results show that changes in the Inverse Participation Ratio of the singular value spectrum are strongly correlated with the model's consistency after token reduction. Based on these insights, we propose LoFi--a training-free visual token reduction method that utilizes the leverage score from SVD for token pruning. Experimental evaluations demonstrate that LoFi not only reduces computational costs with minimal performance degradation but also significantly outperforms state-of-the-art methods in terms of output consistency.
Authors: Hritik Bansal, Clark Peng, Yonatan Bitton, Roman Goldenberg, Aditya Grover, Kai-Wei Chang
Abstract: Large-scale video generative models, capable of creating realistic videos of diverse visual concepts, are strong candidates for general-purpose physical world simulators. However, their adherence to physical commonsense across real-world actions remains unclear (e.g., playing tennis, backflip). Existing benchmarks suffer from limitations such as limited size, lack of human evaluation, sim-to-real gaps, and absence of fine-grained physical rule analysis. To address this, we introduce VideoPhy-2, an action-centric dataset for evaluating physical commonsense in generated videos. We curate 200 diverse actions and detailed prompts for video synthesis from modern generative models. We perform human evaluation that assesses semantic adherence, physical commonsense, and grounding of physical rules in the generated videos. Our findings reveal major shortcomings, with even the best model achieving only 22% joint performance (i.e., high semantic and physical commonsense adherence) on the hard subset of VideoPhy-2. We find that the models particularly struggle with conservation laws like mass and momentum. Finally, we also train VideoPhy-AutoEval, an automatic evaluator for fast, reliable assessment on our dataset. Overall, VideoPhy-2 serves as a rigorous benchmark, exposing critical gaps in video generative models and guiding future research in physically-grounded video generation. The data and code is available at https://videophy2.github.io/.
Authors: Aref Farhadipour, Hossein Ranjbar, Masoumeh Chapariniya, Teodora Vukovic, Sarah Ebling, Volker Dellwo
Abstract: Emotion recognition and sentiment analysis are pivotal tasks in speech and language processing, particularly in real-world scenarios involving multi-party, conversational data. This paper presents a multimodal approach to tackle these challenges on a well-known dataset. We propose a system that integrates four key modalities/channels using pre-trained models: RoBERTa for text, Wav2Vec2 for speech, a proposed FacialNet for facial expressions, and a CNN+Transformer architecture trained from scratch for video analysis. Feature embeddings from each modality are concatenated to form a multimodal vector, which is then used to predict emotion and sentiment labels. The multimodal system demonstrates superior performance compared to unimodal approaches, achieving an accuracy of 66.36% for emotion recognition and 72.15% for sentiment analysis.
Authors: Yanwei Wang
Abstract: 3D reconstruction of high-resolution target remains a challenge task due to the large memory required from the large input image size. Recently developed learning based algorithms provide promising reconstruction performance than traditional ones, however, they generally require more memory than the traditional algorithms and facing scalability issue. In this paper, we developed a generic approach, sub-image recapture (SIR), to split large image into smaller sub-images and process them individually. As a result of this framework, the existing 3D reconstruction algorithms can be implemented based on sub-image recapture with significantly reduced memory and substantially improved scalability
Authors: Wei Dai, Alan Luo, Zane Durante, Debadutta Dash, Arnold Milstein, Kevin Schulman, Ehsan Adeli, Li Fei-Fei
Abstract: In the rapidly evolving domain of video understanding, Video Question Answering (VideoQA) remains a focal point. However, existing datasets exhibit gaps in temporal and spatial granularity, which consequently limits the capabilities of existing VideoQA methods. This paper introduces the Multi-Object Multi-Actor Question Answering (MOMA-QA) dataset, which is designed to address these shortcomings by emphasizing temporal localization, spatial relationship reasoning, and entity-centric queries. With ground truth scene graphs and temporal interval annotations, MOMA-QA is ideal for developing models for fine-grained video understanding. Furthermore, we present a novel video-language model, SGVLM, which incorporates a scene graph predictor, an efficient frame retriever, and a pre-trained large language model for temporal localization and fine-grained relationship understanding. Evaluations on MOMA-QA and other public datasets demonstrate the superior performance of our model, setting new benchmarks for VideoQA.
Authors: Siyu Li, Yihong Cao, Hao Shi, Yongsheng Zang, Xuan He, Kailun Yang, Zhiyong Li
Abstract: The exploration of Bird's-Eye View (BEV) mapping technology has driven significant innovation in visual perception technology for autonomous driving. BEV mapping models need to be applied to the unlabeled real world, making the study of unsupervised domain adaptation models an essential path. However, research on unsupervised domain adaptation for BEV mapping remains limited and cannot perfectly accommodate all BEV mapping tasks. To address this gap, this paper proposes HierDAMap, a universal and holistic BEV domain adaptation framework with hierarchical perspective priors. Unlike existing research that solely focuses on image-level learning using prior knowledge, this paper explores the guiding role of perspective prior knowledge across three distinct levels: global, sparse, and instance levels. With these priors, HierDA consists of three essential components, including Semantic-Guided Pseudo Supervision (SGPS), Dynamic-Aware Coherence Learning (DACL), and Cross-Domain Frustum Mixing (CDFM). SGPS constrains the cross-domain consistency of perspective feature distribution through pseudo labels generated by vision foundation models in 2D space. To mitigate feature distribution discrepancies caused by spatial variations, DACL employs uncertainty-aware predicted depth as an intermediary to derive dynamic BEV labels from perspective pseudo-labels, thereby constraining the coarse BEV features derived from corresponding perspective features. CDFM, on the other hand, leverages perspective masks of view frustum to mix multi-view perspective images from both domains, which guides cross-domain view transformation and encoding learning through mixed BEV labels. The proposed method is verified on multiple BEV mapping tasks, such as BEV semantic segmentation, high-definition semantic, and vectorized mapping. The source code will be made publicly available at https://github.com/lynn-yu/HierDAMap.
Authors: Sungsik Kim, Janghyun Baek, Jinkyu Kim, Jaekoo Lee
Abstract: While Large Language Models (LLMs) have recently shown impressive results in reasoning tasks, their application to pedestrian trajectory prediction remains challenging due to two key limitations: insufficient use of visual information and the difficulty of predicting entire trajectories. To address these challenges, we propose Goal-driven and User-Informed Dynamic Estimation for pedestrian trajectory using Chain-of-Thought (GUIDE-CoT). Our approach integrates two innovative modules: (1) a goal-oriented visual prompt, which enhances goal prediction accuracy combining visual prompts with a pretrained visual encoder, and (2) a chain-of-thought (CoT) LLM for trajectory generation, which generates realistic trajectories toward the predicted goal. Moreover, our method introduces controllable trajectory generation, allowing for flexible and user-guided modifications to the predicted paths. Through extensive experiments on the ETH/UCY benchmark datasets, our method achieves state-of-the-art performance, delivering both high accuracy and greater adaptability in pedestrian trajectory prediction. Our code is publicly available at https://github.com/ai-kmu/GUIDE-CoT.
Authors: Zhuowen Zheng, Yain-Whar Si, Xiaochen Yuan, Junwei Duan, Ke Wang, Xiaofan Li, Xinyuan Zhang, Xueyuan Gong
Abstract: Nowadays, with the advancement of deep neural networks (DNNs) and the availability of large-scale datasets, the face recognition (FR) model has achieved exceptional performance. However, since the parameter magnitude of the fully connected (FC) layer directly depends on the number of identities in the dataset. If training the FR model on large-scale datasets, the size of the model parameter will be excessively huge, leading to substantial demand for computational resources, such as time and memory. This paper proposes the attention fully connected (AttFC) layer, which could significantly reduce computational resources. AttFC employs an attention loader to generate the generative class center (GCC), and dynamically store the class center with Dynamic Class Container (DCC). DCC only stores a small subset of all class centers in FC, thus its parameter count is substantially less than the FC layer. Also, training face recognition models on large-scale datasets with one GPU often encounter out-of-memory (OOM) issues. AttFC overcomes this and achieves comparable performance to state-of-the-art methods.
Authors: Somayeh Hussaini, Tobias Fischer, Michael Milford
Abstract: In visual place recognition (VPR), filtering and sequence-based matching approaches can improve performance by integrating temporal information across image sequences, especially in challenging conditions. While these methods are commonly applied, their effects on system behavior can be unpredictable and can actually make performance worse in certain situations. In this work, we present a new supervised learning approach that learns to predict the per-frame sequence matching receptiveness (SMR) of VPR techniques, enabling the system to selectively decide when to trust the output of a sequence matching system. The approach is agnostic to the underlying VPR technique. Our approach predicts SMR-and hence significantly improves VPR performance-across a large range of state-of-the-art and classical VPR techniques (namely CosPlace, MixVPR, EigenPlaces, SALAD, AP-GeM, NetVLAD and SAD), and across three benchmark VPR datasets (Nordland, Oxford RobotCar, and SFU-Mountain). We also provide insights into a complementary approach that uses the predictor to replace discarded matches, as well as ablation studies, including an analysis of the interactions between our SMR predictor and the selected sequence length. We will release our code upon acceptance.
Authors: Xiangyan Qu, Jing Yu, Jiamin Zhuang, Gaopeng Gou, Gang Xiong, Qi Wu
Abstract: Zero-shot learning (ZSL) aims to train a model on seen classes and recognize unseen classes by knowledge transfer through shared auxiliary information. Recent studies reveal that documents from encyclopedias provide helpful auxiliary information. However, existing methods align noisy documents, entangled in visual and non-visual descriptions, with image regions, yet solely depend on implicit learning. These models fail to filter non-visual noise reliably and incorrectly align non-visual words to image regions, which is harmful to knowledge transfer. In this work, we propose a novel multi-attribute document supervision framework to remove noises at both document collection and model learning stages. With the help of large language models, we introduce a novel prompt algorithm that automatically removes non-visual descriptions and enriches less-described documents in multiple attribute views. Our proposed model, MADS, extracts multi-view transferable knowledge with information decoupling and semantic interactions for semantic alignment at local and global levels. Besides, we introduce a model-agnostic focus loss to explicitly enhance attention to visually discriminative information during training, also improving existing methods without additional parameters. With comparable computation costs, MADS consistently outperforms the SOTA by 7.2% and 8.2% on average in three benchmarks for document-based ZSL and GZSL settings, respectively. Moreover, we qualitatively offer interpretable predictions from multiple attribute views.
Authors: Yihong Leng, Jiaojiao Li, Haitao Xu, Rui Song
Abstract: Current hyperspectral image (HSI) reconstruction methods primarily rely on image-level approaches, which are time-consuming to form abundant high-quality HSIs through imagers. In contrast, spectrometers offer a more efficient alternative by capturing high-fidelity point spectra, enabling pixel-level HSI reconstruction that balances accuracy and label efficiency. To this end, we introduce a pixel-level spectral super-resolution (Pixel-SSR) paradigm that reconstructs HSI from RGB and point spectra. Despite its advantages, Pixel-SSR presents two key challenges: 1) generalizability to novel scenes lacking point spectra, and 2) effective information extraction to promote reconstruction accuracy. To address the first challenge, a Gamma-modeled strategy is investigated to synthesize point spectra based on their intrinsic properties, including nonnegativity, a skewed distribution, and a positive correlation. Furthermore, complementary three-branch prompts from RGB and point spectra are extracted with a Dynamic Prompt Mamba (DyPro-Mamba), which progressively directs the reconstruction with global spatial distributions, edge details, and spectral dependency. Comprehensive evaluations, including horizontal comparisons with leading methods and vertical assessments across unsupervised and image-level supervised paradigms, demonstrate that ours achieves competitive reconstruction accuracy with efficient label consumption.
Authors: Konstantinos D. Polyzos, Athanasios Bacharis, Saketh Madhuvarasu, Nikos Papanikolopoulos, Tara Javidi
Abstract: Gaussian splatting (GS) along with its extensions and variants provides outstanding performance in real-time scene rendering while meeting reduced storage demands and computational efficiency. While the selection of 2D images capturing the scene of interest is crucial for the proper initialization and training of GS, hence markedly affecting the rendering performance, prior works rely on passively and typically densely selected 2D images. In contrast, this paper proposes `ActiveInitSplat', a novel framework for active selection of training images for proper initialization and training of GS. ActiveInitSplat relies on density and occupancy criteria of the resultant 3D scene representation from the selected 2D images, to ensure that the latter are captured from diverse viewpoints leading to better scene coverage and that the initialized Gaussian functions are well aligned with the actual 3D structure. Numerical tests on well-known simulated and real environments demonstrate the merits of ActiveInitSplat resulting in significant GS rendering performance improvement over passive GS baselines, in the widely adopted LPIPS, SSIM, and PSNR metrics.
Authors: Cagri Gungor, Derek Eppinger, Adriana Kovashka
Abstract: Tactile sensing, which relies on direct physical contact, is critical for human perception and underpins applications in computer vision, robotics, and multimodal learning. Because tactile data is often scarce and costly to acquire, generating synthetic tactile images provides a scalable solution to augment real-world measurements. However, ensuring robust generalization in synthesizing tactile images-capturing subtle, material-specific contact features-remains challenging. We demonstrate that overlapping training and test samples in commonly used datasets inflate performance metrics, obscuring the true generalizability of tactile models. To address this, we propose a leakage-free evaluation protocol coupled with novel, reference-free metrics-TMMD, I-TMMD, CI-TMMD, and D-TMMD-tailored for tactile generation. Moreover, we propose a vision-to-touch generation method that leverages text as an intermediate modality by incorporating concise, material-specific descriptions during training to better capture essential tactile features. Experiments on two popular visuo-tactile datasets, Touch and Go and HCT, show that our approach achieves superior performance and enhanced generalization in a leakage-free setting.
Authors: Ta Duc Huy, Sen Kim Tran, Phan Nguyen, Nguyen Hoang Tran, Tran Bao Sam, Anton van den Hengel, Zhibin Liao, Johan W. Verjans, Minh-Son To, Vu Minh Hieu Phan
Abstract: The ability to interpret and intervene model decisions is important for the adoption of computer-aided diagnosis methods in clinical workflows. Recent concept-based methods link the model predictions with interpretable concepts and modify their activation scores to interact with the model. However, these concepts are at the image level, which hinders the model from pinpointing the exact patches the concepts are activated. Alternatively, prototype-based methods learn representations from training image patches and compare these with test image patches, using the similarity scores for final class prediction. However, interpreting the underlying concepts of these patches can be challenging and often necessitates post-hoc guesswork. To address this issue, this paper introduces the novel Concept-based Similarity Reasoning network (CSR), which offers (i) patch-level prototype with intrinsic concept interpretation, and (ii) spatial interactivity. First, the proposed CSR provides localized explanation by grounding prototypes of each concept on image regions. Second, our model introduces novel spatial-level interaction, allowing doctors to engage directly with specific image areas, making it an intuitive and transparent tool for medical imaging. CSR improves upon prior state-of-the-art interpretable methods by up to 4.5\% across three biomedical datasets. Our code is released at https://github.com/tadeephuy/InteractCSR.
Authors: Yuefan Cao, Xuyang Guo, Jiayan Huo, Yingyu Liang, Zhenmei Shi, Zhao Song, Jiahao Zhang, Zhen Zhuang
Abstract: Generative modeling is widely regarded as one of the most essential problems in today's AI community, with text-to-image generation having gained unprecedented real-world impacts. Among various approaches, diffusion models have achieved remarkable success and have become the de facto solution for text-to-image generation. However, despite their impressive performance, these models exhibit fundamental limitations in adhering to numerical constraints in user instructions, frequently generating images with an incorrect number of objects. While several prior works have mentioned this issue, a comprehensive and rigorous evaluation of this limitation remains lacking. To address this gap, we introduce T2ICountBench, a novel benchmark designed to rigorously evaluate the counting ability of state-of-the-art text-to-image diffusion models. Our benchmark encompasses a diverse set of generative models, including both open-source and private systems. It explicitly isolates counting performance from other capabilities, provides structured difficulty levels, and incorporates human evaluations to ensure high reliability. Extensive evaluations with T2ICountBench reveal that all state-of-the-art diffusion models fail to generate the correct number of objects, with accuracy dropping significantly as the number of objects increases. Additionally, an exploratory study on prompt refinement demonstrates that such simple interventions generally do not improve counting accuracy. Our findings highlight the inherent challenges in numerical understanding within diffusion models and point to promising directions for future improvements.
Authors: Yan Yang, Dongxu Li, Haoning Wu, Bei Chen, Liu Liu, Liyuan Pan, Junnan Li
Abstract: Solving expert-level multimodal tasks is a key milestone towards general intelligence. As the capabilities of multimodal large language models (MLLMs) continue to improve, evaluation of such advanced multimodal intelligence becomes necessary yet challenging. In this work, we introduce ProBench, a benchmark of open-ended user queries that require professional expertise and advanced reasoning. ProBench consists of 4,000 high-quality samples independently submitted by professionals based on their daily productivity demands. It spans across 10 fields and 56 sub-fields, including science, arts, humanities, coding, mathematics, and creative writing. Experimentally, we evaluate and compare 24 latest models using MLLM-as-a-Judge. Our results reveal that although the best open-source models rival the proprietary ones, ProBench presents significant challenges in visual perception, textual understanding, domain knowledge and advanced reasoning, thus providing valuable directions for future multimodal AI research efforts.
Authors: Nasla Saleem, Talukder Zaki Jubery, Aditya Balu, Yan Zhou, Yawei Li, Patrick S. Schnable, Adarsh Krishnamurthy, Baskar Ganapathysubramanian
Abstract: High-density planting is a widely adopted strategy to enhance maize productivity, yet it introduces challenges such as increased interplant competition and shading, which can limit light capture and overall yield potential. In response, some maize plants naturally reorient their canopies to optimize light capture, a process known as canopy reorientation. Understanding this adaptive response and its impact on light capture is crucial for maximizing agricultural yield potential. This study introduces an end-to-end framework that integrates realistic 3D reconstructions of field-grown maize with photosynthetically active radiation (PAR) modeling to assess the effects of phyllotaxy and planting density on light interception. In particular, using 3D point clouds derived from field data, virtual fields for a diverse set of maize genotypes were constructed and validated against field PAR measurements. Using this framework, we present detailed analyses of the impact of canopy orientations, plant and row spacings, and planting row directions on PAR interception throughout a typical growing season. Our findings highlight significant variations in light interception efficiency across different planting densities and canopy orientations. By elucidating the relationship between canopy architecture and light capture, this study offers valuable guidance for optimizing maize breeding and cultivation strategies across diverse agricultural settings.
Authors: Xiaoqian Hu
Abstract: In Recent Years, Digital Technologies Have Made Significant Strides In Augmenting-Human-Health, Cognition, And Perception, Particularly Within The Field Of Computational-Pathology. This Paper Presents A Novel Approach To Enhancing The Analysis Of Histopathology Images By Leveraging A Mult-modal-Model That Combines Vision Transformers (Vit) With Gpt-2 For Image Captioning. The Model Is Fine-Tuned On The Specialized Arch-Dataset, Which Includes Dense Image Captions Derived From Clinical And Academic Resources, To Capture The Complexities Of Pathology Images Such As Tissue Morphologies, Staining Variations, And Pathological Conditions. By Generating Accurate, Contextually Captions, The Model Augments The Cognitive Capabilities Of Healthcare Professionals, Enabling More Efficient Disease Classification, Segmentation, And Detection. The Model Enhances The Perception Of Subtle Pathological Features In Images That Might Otherwise Go Unnoticed, Thereby Improving Diagnostic Accuracy. Our Approach Demonstrates The Potential For Digital Technologies To Augment Human Cognitive Abilities In Medical Image Analysis, Providing Steps Toward More Personalized And Accurate Healthcare Outcomes.
Authors: Xin Liu, Jie Liu, Jie Tang, Gangshan Wu
Abstract: Transformer-based methods have demonstrated impressive performance in low-level visual tasks such as Image Super-Resolution (SR). However, its computational complexity grows quadratically with the spatial resolution. A series of works attempt to alleviate this problem by dividing Low-Resolution images into local windows, axial stripes, or dilated windows. SR typically leverages the redundancy of images for reconstruction, and this redundancy appears not only in local regions but also in long-range regions. However, these methods limit attention computation to content-agnostic local regions, limiting directly the ability of attention to capture long-range dependency. To address these issues, we propose a lightweight Content-Aware Token Aggregation Network (CATANet). Specifically, we propose an efficient Content-Aware Token Aggregation module for aggregating long-range content-similar tokens, which shares token centers across all image tokens and updates them only during the training phase. Then we utilize intra-group self-attention to enable long-range information interaction. Moreover, we design an inter-group cross-attention to further enhance global information interaction. The experimental results show that, compared with the state-of-the-art cluster-based method SPIN, our method achieves superior performance, with a maximum PSNR improvement of 0.33dB and nearly double the inference speed.
Authors: Xingzu Zhan, Chen Xie, Haoran Sun, Xiaochun Mai
Abstract: Text-to-motion generation is a rapidly growing field at the nexus of multimodal learning and computer graphics, promising flexible and cost-effective applications in gaming, animation, robotics, and virtual reality. Existing approaches often rely on simple spatiotemporal stacking, which introduces feature redundancy, while subtle joint-level details remain overlooked from a spatial perspective. To this end, we propose a novel HiSTF Mamba framework. The framework is composed of three key modules: Dual-Spatial Mamba, Bi-Temporal Mamba, and Dynamic Spatiotemporal Fusion Module (DSFM). Dual-Spatial Mamba incorporates ``Part-based + Whole-based'' parallel modeling to represent both whole-body coordination and fine-grained joint dynamics. Bi-Temporal Mamba adopts a bidirectional scanning strategy, effectively encoding short-term motion details and long-term dependencies. DSFM further performs redundancy removal and extraction of complementary information for temporal features, then fuses them with spatial features, yielding an expressive spatio-temporal representation. Experimental results on the HumanML3D dataset demonstrate that HiSTF Mamba achieves state-of-the-art performance across multiple metrics. In particular, it reduces the FID score from 0.283 to 0.189, a relative decrease of nearly 30%. These findings validate the effectiveness of HiSTF Mamba in achieving high fidelity and strong semantic alignment in text-to-motion generation.
Authors: S M A Sharif, Abdur Rehman, Zain Ul Abidin, Rizwan Ali Naqvi, Fayaz Ali Dharejo, Radu Timofte
Abstract: Digital cameras often struggle to produce plausible images in low-light conditions. Improving these single-shot images remains challenging due to a lack of diverse real-world pair data samples. To address this limitation, we propose a large-scale high-resolution (i.e., beyond 4k) pair Single-Shot Low-Light Enhancement (SLLIE) dataset. Our dataset comprises 6,425 unique focus-aligned image pairs captured with smartphone sensors in dynamic settings under challenging lighting conditions (0.1--200 lux), covering various indoor and outdoor scenes with varying noise and intensity. We extracted and refined around 180,000 non-overlapping patches from 6,025 collected scenes for training while reserving 400 pairs for benchmarking. In addition to that, we collected 2,117 low-light scenes from different sources for extensive real-world aesthetic evaluation. To our knowledge, this is the largest real-world dataset available for SLLIE research. We also propose learning luminance-chrominance (LC) attributes separately through a tuning fork-shaped transformer model to enhance real-world low-light images, addressing challenges like denoising and over-enhancement in complex scenes. We also propose an LC cross-attention block for feature fusion, an LC refinement block for enhanced reconstruction, and LC-guided supervision to ensure perceptually coherent enhancements. We demonstrated our method's effectiveness across various hardware and scenarios, proving its practicality in real-world applications. Code and dataset available at https://github.com/sharif-apu/LSD-TFFormer.
Authors: Xiaoliang Ju, Hongsheng Li
Abstract: We present DirectTriGS, a novel framework designed for 3D object generation with Gaussian Splatting (GS). GS-based rendering for 3D content has gained considerable attention recently. However, there has been limited exploration in directly generating 3D Gaussians compared to traditional generative modeling approaches. The main challenge lies in the complex data structure of GS represented by discrete point clouds with multiple channels. To overcome this challenge, we propose employing the triplane representation, which allows us to represent Gaussian Splatting as an image-like continuous field. This representation effectively encodes both the geometry and texture information, enabling smooth transformation back to Gaussian point clouds and rendering into images by a TriRenderer, with only 2D supervisions. The proposed TriRenderer is fully differentiable, so that the rendering loss can supervise both texture and geometry encoding. Furthermore, the triplane representation can be compressed using a Variational Autoencoder (VAE), which can subsequently be utilized in latent diffusion to generate 3D objects. The experiments demonstrate that the proposed generation framework can produce high-quality 3D object geometry and rendering results in the text-to-3D task.
Authors: Chikai Shang, Mengke Li, Yiqun Zhang, Zhen Chen, Jinlin Wu, Fangqing Gu, Yang Lu, Yiu-ming Cheung
Abstract: Visual prompt tuning (VPT) provides an efficient and effective solution for adapting pre-trained models to various downstream tasks by incorporating learnable prompts. However, most prior art indiscriminately applies a fixed prompt distribution across different tasks, neglecting the importance of each block differing depending on the task. In this paper, we investigate adaptive distribution optimization (ADO) by addressing two key questions: (1) How to appropriately and formally define ADO, and (2) How to design an adaptive distribution strategy guided by this definition? Through in-depth analysis, we provide an affirmative answer that properly adjusting the distribution significantly improves VPT performance, and further uncover a key insight that a nested relationship exists between ADO and VPT. Based on these findings, we propose a new VPT framework, termed PRO-VPT (iterative Prompt RelOcation-based VPT), which adaptively adjusts the distribution building upon a nested optimization formulation. Specifically, we develop a prompt relocation strategy for ADO derived from this formulation, comprising two optimization steps: identifying and pruning idle prompts, followed by determining the optimal blocks for their relocation. By iteratively performing prompt relocation and VPT, our proposal adaptively learns the optimal prompt distribution, thereby unlocking the full potential of VPT. Extensive experiments demonstrate that our proposal significantly outperforms state-of-the-art VPT methods, e.g., PRO-VPT surpasses VPT by 1.6% average accuracy, leading prompt-based methods to state-of-the-art performance on the VTAB-1k benchmark. The code is available at https://github.com/ckshang/PRO-VPT.
Authors: Hanqing Liu, Shouwei Ruan, Yao Huang, Shiji Zhao, Xingxing Wei
Abstract: Vision-Language Models (VLMs) have achieved remarkable success in various tasks, yet their robustness to real-world illumination variations remains largely unexplored. To bridge this gap, we propose \textbf{I}llumination \textbf{T}ransformation \textbf{A}ttack (\textbf{ITA}), the first framework to systematically assess VLMs' robustness against illumination changes. However, there still exist two key challenges: (1) how to model global illumination with fine-grained control to achieve diverse lighting conditions and (2) how to ensure adversarial effectiveness while maintaining naturalness. To address the first challenge, we innovatively decompose global illumination into multiple parameterized point light sources based on the illumination rendering equation. This design enables us to model more diverse lighting variations that previous methods could not capture. Then, by integrating these parameterized lighting variations with physics-based lighting reconstruction techniques, we could precisely render such light interactions in the original scenes, finally meeting the goal of fine-grained lighting control. For the second challenge, by controlling illumination through the lighting reconstrution model's latent space rather than direct pixel manipulation, we inherently preserve physical lighting priors. Furthermore, to prevent potential reconstruction artifacts, we design additional perceptual constraints for maintaining visual consistency with original images and diversity constraints for avoiding light source convergence. Extensive experiments demonstrate that our ITA could significantly reduce the performance of advanced VLMs, e.g., LLaVA-1.6, while possessing competitive naturalness, exposing VLMS' critical illuminiation vulnerabilities.
Authors: Jiacheng Liu, Chang Zou, Yuanhuiyi Lyu, Junjie Chen, Linfeng Zhang
Abstract: Diffusion Transformers (DiT) have revolutionized high-fidelity image and video synthesis, yet their computational demands remain prohibitive for real-time applications. To solve this problem, feature caching has been proposed to accelerate diffusion models by caching the features in the previous timesteps and then reusing them in the following timesteps. However, at timesteps with significant intervals, the feature similarity in diffusion models decreases substantially, leading to a pronounced increase in errors introduced by feature caching, significantly harming the generation quality. To solve this problem, we propose TaylorSeer, which firstly shows that features of diffusion models at future timesteps can be predicted based on their values at previous timesteps. Based on the fact that features change slowly and continuously across timesteps, TaylorSeer employs a differential method to approximate the higher-order derivatives of features and predict features in future timesteps with Taylor series expansion. Extensive experiments demonstrate its significant effectiveness in both image and video synthesis, especially in high acceleration ratios. For instance, it achieves an almost lossless acceleration of 4.99$\times$ on FLUX and 5.00$\times$ on HunyuanVideo without additional training. On DiT, it achieves $3.41$ lower FID compared with previous SOTA at $4.53$$\times$ acceleration. %Our code is provided in the supplementary materials and will be made publicly available on GitHub. Our codes have been released in Github:https://github.com/Shenyi-Z/TaylorSeer
Authors: Ning Ding, Jing Han, Yuchuan Tian, Chao Xu, Kai Han, Yehui Tang
Abstract: Diffusion Transformer (DiT) has now become the preferred choice for building image generation models due to its great generation capability. Unlike previous convolution-based UNet models, DiT is purely composed of a stack of transformer blocks, which renders DiT excellent in scalability like large language models. However, the growing model size and multi-step sampling paradigm bring about considerable pressure on deployment and inference. In this work, we propose a post-training quantization framework tailored for Diffusion Transforms to tackle these challenges. We firstly locate that the quantization difficulty of DiT mainly originates from the time-dependent channel-specific outliers. We propose a timestep-aware shift-and-scale strategy to smooth the activation distribution to reduce the quantization error. Secondly, based on the observation that activations of adjacent timesteps have similar distributions, we utilize a hierarchical clustering scheme to divide the denoising timesteps into multiple groups. We further design a re-parameterization scheme which absorbs the quantization parameters into nearby module to avoid redundant computations. Comprehensive experiments demonstrate that out PTQ method successfully quantize the Diffusion Transformer into 8-bit weight and 8-bit activation (W8A8) with state-of-the-art FiD score. And our method can further quantize DiT model into 4-bit weight and 8-bit activation (W4A8) without sacrificing generation quality.
Authors: Hanyu Zhou, Gim Hee Lee
Abstract: Large multimodal models (LMMs) excel in scene understanding but struggle with fine-grained spatiotemporal reasoning due to weak alignment between linguistic and visual representations. Existing methods map textual positions and durations into the visual space encoded from frame-based videos, but suffer from temporal sparsity that limits language-vision temporal coordination. To address this issue, we introduce LLaFEA (Large Language and Frame-Event Assistant) to leverage event cameras for temporally dense perception and frame-event fusion. Our approach employs a cross-attention mechanism to integrate complementary spatial and temporal features, followed by self-attention matching for global spatio-temporal associations. We further embed textual position and duration tokens into the fused visual space to enhance fine-grained alignment. This unified framework ensures robust spatio-temporal coordinate alignment, enabling LMMs to interpret scenes at any position and any time. In addition, we construct a dataset of real-world frames-events with coordinate instructions and conduct extensive experiments to validate the effectiveness of the proposed method.
Authors: Sania Zahan, Ghulam Mubashar Hassan, Ajmal Mian
Abstract: The increasing pace of population aging calls for better care and support systems. Falling is a frequent and critical problem for elderly people causing serious long-term health issues. Fall detection from video streams is not an attractive option for real-life applications due to privacy issues. Existing methods try to resolve this issue by using very low-resolution cameras or video encryption. However, privacy cannot be ensured completely with such approaches. Key points on the body, such as skeleton joints, can convey significant information about motion dynamics and successive posture changes which are crucial for fall detection. Skeleton joints have been explored for feature extraction but with image recognition models that ignore joint dependency across frames which is important for the classification of actions. Moreover, existing models are over-parameterized or evaluated on small datasets with very few activity classes. We propose an efficient graph convolution network model that exploits spatio-temporal joint dependencies and dynamics of human skeleton joints for accurate fall detection. Our method leverages dynamic representation with robust concurrent spatio-temporal characteristics of skeleton joints. We performed extensive experiments on three large-scale datasets. With a significantly smaller model size than most existing methods, our proposed method achieves state-of-the-art results on the large scale NTU datasets.
Authors: Jianxiong Gao, Yichang Liu, Baofeng Yang, Jianfeng Feng, Yanwei Fu
Abstract: In this paper, we introduce CineBrain, the first large-scale dataset featuring simultaneous EEG and fMRI recordings during dynamic audiovisual stimulation. Recognizing the complementary strengths of EEG's high temporal resolution and fMRI's deep-brain spatial coverage, CineBrain provides approximately six hours of narrative-driven content from the popular television series The Big Bang Theory for each of six participants. Building upon this unique dataset, we propose CineSync, an innovative multimodal decoding framework integrates a Multi-Modal Fusion Encoder with a diffusion-based Neural Latent Decoder. Our approach effectively fuses EEG and fMRI signals, significantly improving the reconstruction quality of complex audiovisual stimuli. To facilitate rigorous evaluation, we introduce Cine-Benchmark, a comprehensive evaluation protocol that assesses reconstructions across semantic and perceptual dimensions. Experimental results demonstrate that CineSync achieves state-of-the-art video reconstruction performance and highlight our initial success in combining fMRI and EEG for reconstructing both video and audio stimuli. Project Page: https://jianxgao.github.io/CineBrain.
Authors: Jiaxin Li, Hongxing Wang, Jiawei Tan, Zhilong Ou, Junsong Yuan
Abstract: Understanding 3D object shapes necessitates shape representation by object parts abstracted from results of instance and semantic segmentation. Promising shape representations enable computers to interpret a shape with meaningful parts and identify their repeatability. However, supervised shape representations depend on costly annotation efforts, while current unsupervised methods work under strong semantic priors and involve multi-stage training, thereby limiting their generalization and deployment in shape reasoning and understanding. Driven by the tendency of high-dimensional semantically similar features to lie in or near low-dimensional subspaces, we introduce a one-stage, fully unsupervised framework towards semantic-aware shape representation. This framework produces joint instance segmentation, semantic segmentation, and shape abstraction through sparse representation and feature alignment of object parts in a high-dimensional space. For sparse representation, we devise a sparse latent membership pursuit method that models each object part feature as a sparse convex combination of point features at either the semantic or instance level, promoting part features in the same subspace to exhibit similar semantics. For feature alignment, we customize an attention-based strategy in the feature space to align instance- and semantic-level object part features and reconstruct the input shape using both of them, ensuring geometric reusability and semantic consistency of object parts. To firm up semantic disambiguation, we construct cascade unfrozen learning on geometric parameters of object parts.
Authors: Wentao Wu, Chenglong Li, Xiao Wang, Bin Luo, Qi Liu
Abstract: Existing multimodal UAV object detection methods often overlook the impact of semantic gaps between modalities, which makes it difficult to achieve accurate semantic and spatial alignments, limiting detection performance. To address this problem, we propose a Large Language Model (LLM) guided Progressive feature Alignment Network called LPANet, which leverages the semantic features extracted from a large language model to guide the progressive semantic and spatial alignment between modalities for multimodal UAV object detection. To employ the powerful semantic representation of LLM, we generate the fine-grained text descriptions of each object category by ChatGPT and then extract the semantic features using the large language model MPNet. Based on the semantic features, we guide the semantic and spatial alignments in a progressive manner as follows. First, we design the Semantic Alignment Module (SAM) to pull the semantic features and multimodal visual features of each object closer, alleviating the semantic differences of objects between modalities. Second, we design the Explicit Spatial alignment Module (ESM) by integrating the semantic relations into the estimation of feature-level offsets, alleviating the coarse spatial misalignment between modalities. Finally, we design the Implicit Spatial alignment Module (ISM), which leverages the cross-modal correlations to aggregate key features from neighboring regions to achieve implicit spatial alignment. Comprehensive experiments on two public multimodal UAV object detection datasets demonstrate that our approach outperforms state-of-the-art multimodal UAV object detectors.
Authors: Xingye Fan (Rex), Zhongwen (Rex), Zhang, Yuri Boykov
Abstract: This paper demonstrates a surprising result for segmentation with image-level targets: extending binary class tags to approximate relative object-size distributions allows off-the-shelf architectures to solve the segmentation problem. A straightforward zero-avoiding KL-divergence loss for average predictions produces segmentation accuracy comparable to the standard pixel-precise supervision with full ground truth masks. In contrast, current results based on class tags typically require complex non-reproducible architectural modifications and specialized multi-stage training procedures. Our ideas are validated on PASCAL VOC using our new human annotations of approximate object sizes. We also show the results on COCO and medical data using synthetically corrupted size targets. All standard networks demonstrate robustness to the size targets' errors. For some classes, the validation accuracy is significantly better than the pixel-level supervision; the latter is not robust to errors in the masks. Our work provides new ideas and insights on image-level supervision in segmentation and may encourage other simple general solutions to the problem.
Authors: Zeyu Zhang, Yiran Wang, Wei Mao, Danning Li, Rui Zhao, Biao Wu, Zirui Song, Bohan Zhuang, Ian Reid, Richard Hartley
Abstract: Conditional motion generation has been extensively studied in computer vision, yet two critical challenges remain. First, while masked autoregressive methods have recently outperformed diffusion-based approaches, existing masking models lack a mechanism to prioritize dynamic frames and body parts based on given conditions. Second, existing methods for different conditioning modalities often fail to integrate multiple modalities effectively, limiting control and coherence in generated motion. To address these challenges, we propose Motion Anything, a multimodal motion generation framework that introduces an Attention-based Mask Modeling approach, enabling fine-grained spatial and temporal control over key frames and actions. Our model adaptively encodes multimodal conditions, including text and music, improving controllability. Additionally, we introduce Text-Motion-Dance (TMD), a new motion dataset consisting of 2,153 pairs of text, music, and dance, making it twice the size of AIST++, thereby filling a critical gap in the community. Extensive experiments demonstrate that Motion Anything surpasses state-of-the-art methods across multiple benchmarks, achieving a 15% improvement in FID on HumanML3D and showing consistent performance gains on AIST++ and TMD. See our project website https://steve-zeyu-zhang.github.io/MotionAnything
Authors: Jian Jin, Zhenbo Yu, Yang Shen, Zhenyong Fu, Jian Yang
Abstract: Customized text-to-image generation renders user-specified concepts into novel contexts based on textual prompts. Scaling the number of concepts in customized generation meets a broader demand for user creation, whereas existing methods face challenges with generation quality and computational efficiency. In this paper, we propose LaTexBlend, a novel framework for effectively and efficiently scaling multi-concept customized generation. The core idea of LaTexBlend is to represent single concepts and blend multiple concepts within a Latent Textual space, which is positioned after the text encoder and a linear projection. LaTexBlend customizes each concept individually, storing them in a concept bank with a compact representation of latent textual features that captures sufficient concept information to ensure high fidelity. At inference, concepts from the bank can be freely and seamlessly combined in the latent textual space, offering two key merits for multi-concept generation: 1) excellent scalability, and 2) significant reduction of denoising deviation, preserving coherent layouts. Extensive experiments demonstrate that LaTexBlend can flexibly integrate multiple customized concepts with harmonious structures and high subject fidelity, substantially outperforming baselines in both generation quality and computational efficiency. Our code will be publicly available.
Authors: Xin Wen, Bingchen Zhao, Yilun Chen, Jiangmiao Pang, Xiaojuan Qi
Abstract: Pre-trained vision models (PVMs) are fundamental to modern robotics, yet their optimal configuration remains unclear. Through systematic evaluation, we find that while DINO and iBOT outperform MAE across visuomotor control and perception tasks, they struggle when trained on non-(single-)object-centric (NOC) data--a limitation strongly correlated with their diminished ability to learn object-centric representations. This investigation indicates that the ability to form object-centric representations from the non-object-centric robotics dataset is the key to success for PVMs. Motivated by this discovery, we designed SlotMIM, a method that induces object-centric representations by introducing a semantic bottleneck to reduce the number of prototypes to encourage the emergence of objectness as well as cross-view consistency regularization for encouraging multiview invariance. Our experiments encompass pre-training on object-centric, scene-centric, web-crawled, and ego-centric data. Across all settings, our approach learns transferrable representations and achieves significant improvements over prior work in image recognition, scene understanding, and robot learning evaluations. When scaled up with million-scale datasets, our method also demonstrates superior data efficiency and scalability. Our code and models are publicly available at https://github.com/CVMI-Lab/SlotMIM.
Authors: Shining Wang, Yunlong Wang, Ruiqi Wu, Bingliang Jiao, Wenxuan Wang, Peng Wang
Abstract: When discussing the Aerial-Ground Person Re-identification (AGPReID) task, we face the main challenge of the significant appearance variations caused by different viewpoints, making identity matching difficult. To address this issue, previous methods attempt to reduce the differences between viewpoints by critical attributes and decoupling the viewpoints. While these methods can mitigate viewpoint differences to some extent, they still face two main issues: (1) difficulty in handling viewpoint diversity and (2) neglect of the contribution of local features. To effectively address these challenges, we design and implement the Self-Calibrating and Adaptive Prompt (SeCap) method for the AGPReID task. The core of this framework relies on the Prompt Re-calibration Module (PRM), which adaptively re-calibrates prompts based on the input. Combined with the Local Feature Refinement Module (LFRM), SeCap can extract view-invariant features from local features for AGPReID. Meanwhile, given the current scarcity of datasets in the AGPReID field, we further contribute two real-world Large-scale Aerial-Ground Person Re-Identification datasets, LAGPeR and G2APS-ReID. The former is collected and annotated by us independently, covering $4,231$ unique identities and containing $63,841$ high-quality images; the latter is reconstructed from the person search dataset G2APS. Through extensive experiments on AGPReID datasets, we demonstrate that SeCap is a feasible and effective solution for the AGPReID task. The datasets and source code available on https://github.com/wangshining681/SeCap-AGPReID.
Authors: Guanghao Li, Mingzhi Chen, Hao Yu, Shuting Dong, Wenhao Jiang, Ming Tang, Chun Yuan
Abstract: Deep learning-based denoising models have been widely employed in vision tasks, functioning as filters to eliminate noise while retaining crucial semantic information. Additionally, they play a vital role in defending against adversarial perturbations that threaten downstream tasks. However, these models can be intrinsically susceptible to adversarial attacks due to their dependence on specific noise assumptions. Existing attacks on denoising models mainly aim at deteriorating visual clarity while neglecting semantic manipulation, rendering them either easily detectable or limited in effectiveness. In this paper, we propose Mutual Information-Guided Attack (MIGA), the first method designed to directly attack deep denoising models by strategically disrupting their ability to preserve semantic content via adversarial perturbations. By minimizing the mutual information between the original and denoised images, a measure of semantic similarity. MIGA forces the denoiser to produce perceptually clean yet semantically altered outputs. While these images appear visually plausible, they encode systematically distorted semantics, revealing a fundamental vulnerability in denoising models. These distortions persist in denoised outputs and can be quantitatively assessed through downstream task performance. We propose new evaluation metrics and systematically assess MIGA on four denoising models across five datasets, demonstrating its consistent effectiveness in disrupting semantic fidelity. Our findings suggest that denoising models are not always robust and can introduce security risks in real-world applications. Code is available in the Supplementary Material.
Authors: Xiang Liu, Zhaoxiang Liu, Huan Hu, Zezhou Chen, Kohou Wang, Kai Wang, Shiguo Lian
Abstract: While conversational generative AI has shown considerable potential in enhancing decision-making for agricultural professionals, its exploration has predominantly been anchored in text-based interactions. The evolution of multimodal conversational AI, leveraging vast amounts of image-text data from diverse sources, marks a significant stride forward. However, the application of such advanced vision-language models in the agricultural domain, particularly for crop disease diagnosis, remains underexplored. In this work, we present the crop disease domain multimodal (CDDM) dataset, a pioneering resource designed to advance the field of agricultural research through the application of multimodal learning techniques. The dataset comprises 137,000 images of various crop diseases, accompanied by 1 million question-answer pairs that span a broad spectrum of agricultural knowledge, from disease identification to management practices. By integrating visual and textual data, CDDM facilitates the development of sophisticated question-answering systems capable of providing precise, useful advice to farmers and agricultural professionals. We demonstrate the utility of the dataset by finetuning state-of-the-art multimodal models, showcasing significant improvements in crop disease diagnosis. Specifically, we employed a novel finetuning strategy that utilizes low-rank adaptation (LoRA) to finetune the visual encoder, adapter and language model simultaneously. Our contributions include not only the dataset but also a finetuning strategy and a benchmark to stimulate further research in agricultural technology, aiming to bridge the gap between advanced AI techniques and practical agricultural applications. The dataset is available at https: //github.com/UnicomAI/UnicomBenchmark/tree/main/CDDMBench.
Authors: Yang Liu, Mengyuan Liu, Shudong Huang, Jiancheng Lv
Abstract: Learning visual semantic similarity is a critical challenge in bridging the gap between images and texts. However, there exist inherent variations between vision and language data, such as information density, i.e., images can contain textual information from multiple different views, which makes it difficult to compute the similarity between these two modalities accurately and efficiently. In this paper, we propose a novel framework called Asymmetric Visual Semantic Embedding (AVSE) to dynamically select features from various regions of images tailored to different textual inputs for similarity calculation. To capture information from different views in the image, we design a radial bias sampling module to sample image patches and obtain image features from various views, Furthermore, AVSE introduces a novel module for efficient computation of visual semantic similarity between asymmetric image and text embeddings. Central to this module is the presumption of foundational semantic units within the embeddings, denoted as ``meta-semantic embeddings." It segments all embeddings into meta-semantic embeddings with the same dimension and calculates visual semantic similarity by finding the optimal match of meta-semantic embeddings of two modalities. Our proposed AVSE model is extensively evaluated on the large-scale MS-COCO and Flickr30K datasets, demonstrating its superiority over recent state-of-the-art methods.
Authors: Pengchen Liang, Haishan Huang, Bin Pu, Jianguo Chen, Xiang Hua, Jing Zhang, Weibo Ma, Zhuangzhuang Chen, Yiwei Li, Qing Chang
Abstract: Large-scale pre-trained models, such as Vision Foundation Models (VFMs), have demonstrated impressive performance across various downstream tasks by transferring generalized knowledge, especially when target data is limited. However, their high computational cost and the domain gap between natural and medical images limit their practical application in medical segmentation tasks. Motivated by this, we pose the following important question: "How can we effectively utilize the knowledge of large pre-trained VFMs to train a small, task-specific model for medical image segmentation when training data is limited?" To address this problem, we propose a novel and generalizable task-specific knowledge distillation framework. Our method fine-tunes the VFM on the target segmentation task to capture task-specific features before distilling the knowledge to smaller models, leveraging Low-Rank Adaptation (LoRA) to reduce the computational cost of fine-tuning. Additionally, we incorporate synthetic data generated by diffusion models to augment the transfer set, enhancing model performance in data-limited scenarios. Experimental results across five medical image datasets demonstrate that our method consistently outperforms task-agnostic knowledge distillation and self-supervised pretraining approaches like MoCo v3 and Masked Autoencoders (MAE). For example, on the KidneyUS dataset, our method achieved a 28% higher Dice score than task-agnostic KD using 80 labeled samples for fine-tuning. On the CHAOS dataset, it achieved an 11% improvement over MAE with 100 labeled samples. These results underscore the potential of task-specific knowledge distillation to train accurate, efficient models for medical image segmentation in data-constrained settings.
Authors: Xinyu Xi, Hua Yang, Shentai Zhang, Yijie Liu, Sijin Sun, Xiuju Fu
Abstract: Maritime Multi-Scene Recognition is crucial for enhancing the capabilities of intelligent marine robotics, particularly in applications such as marine conservation, environmental monitoring, and disaster response. However, this task presents significant challenges due to environmental interference, where marine conditions degrade image quality, and the complexity of maritime scenes, which requires deeper reasoning for accurate recognition. Pure vision models alone are insufficient to address these issues. To overcome these limitations, we propose a novel multimodal Artificial Intelligence (AI) framework that integrates image data, textual descriptions and classification vectors generated by a Multimodal Large Language Model (MLLM), to provide richer semantic understanding and improve recognition accuracy. Our framework employs an efficient multimodal fusion mechanism to further enhance model robustness and adaptability in complex maritime environments. Experimental results show that our model achieves 98$\%$ accuracy, surpassing previous SOTA models by 3.5$\%$. To optimize deployment on resource-constrained platforms, we adopt activation-aware weight quantization (AWQ) as a lightweight technique, reducing the model size to 68.75MB with only a 0.5$\%$ accuracy drop while significantly lowering computational overhead. This work provides a high-performance solution for real-time maritime scene recognition, enabling Autonomous Surface Vehicles (ASVs) to support environmental monitoring and disaster response in resource-limited settings.
Authors: Jiahao Wang, Xiangyu Cao, Jiaru Zhong, Yuner Zhang, Haibao Yu, Lei He, Shaobing Xu
Abstract: Despite significant advancements, autonomous driving systems continue to struggle with occluded objects and long-range detection due to the inherent limitations of single-perspective sensing. Aerial-ground cooperation offers a promising solution by integrating UAVs' aerial views with ground vehicles' local observations. However, progress in this emerging field has been hindered by the absence of public datasets and standardized evaluation benchmarks. To address this gap, this paper presents a comprehensive solution for aerial-ground cooperative 3D perception through three key contributions: (1) Griffin, a large-scale multi-modal dataset featuring over 200 dynamic scenes (30k+ frames) with varied UAV altitudes (20-60m), diverse weather conditions, and occlusion-aware 3D annotations, enhanced by CARLA-AirSim co-simulation for realistic UAV dynamics; (2) A unified benchmarking framework for aerial-ground cooperative detection and tracking tasks, including protocols for evaluating communication efficiency, latency tolerance, and altitude adaptability; (3) AGILE, an instance-level intermediate fusion baseline that dynamically aligns cross-view features through query-based interaction, achieving an advantageous balance between communication overhead and perception accuracy. Extensive experiments prove the effectiveness of aerial-ground cooperative perception and demonstrate the direction of further research. The dataset and codes are available at https://github.com/wang-jh18-SVM/Griffin.
Authors: Youngseok Kim, Sunwook Hwang, Hyung-Sin Kim, Saewoong Bahk
Abstract: The growing use of 3D point cloud data in autonomous vehicles (AVs) has raised serious privacy concerns, particularly due to the sensitive information that can be extracted from 3D data. While model inversion attacks have been widely studied in the context of 2D data, their application to 3D point clouds remains largely unexplored. To fill this gap, we present the first in-depth study of model inversion attacks aimed at restoring 3D point cloud scenes. Our analysis reveals the unique challenges, the inherent sparsity of 3D point clouds and the ambiguity between empty and non-empty voxels after voxelization, which are further exacerbated by the dispersion of non-empty voxels across feature extractor layers. To address these challenges, we introduce ConcreTizer, a simple yet effective model inversion attack designed specifically for voxel-based 3D point cloud data. ConcreTizer incorporates Voxel Occupancy Classification to distinguish between empty and non-empty voxels and Dispersion-Controlled Supervision to mitigate non-empty voxel dispersion. Extensive experiments on widely used 3D feature extractors and benchmark datasets, such as KITTI and Waymo, demonstrate that ConcreTizer concretely restores the original 3D point cloud scene from disrupted 3D feature data. Our findings highlight both the vulnerability of 3D data to inversion attacks and the urgent need for robust defense strategies.
Authors: Hanyu Zhou, Haonan Wang, Haoyue Liu, Yuxing Duan, Yi Chang, Luxin Yan
Abstract: High-dynamic scene optical flow is a challenging task, which suffers spatial blur and temporal discontinuous motion due to large displacement in frame imaging, thus deteriorating the spatiotemporal feature of optical flow. Typically, existing methods mainly introduce event camera to directly fuse the spatiotemporal features between the two modalities. However, this direct fusion is ineffective, since there exists a large gap due to the heterogeneous data representation between frame and event modalities. To address this issue, we explore a common-latent space as an intermediate bridge to mitigate the modality gap. In this work, we propose a novel common spatiotemporal fusion between frame and event modalities for high-dynamic scene optical flow, including visual boundary localization and motion correlation fusion. Specifically, in visual boundary localization, we figure out that frame and event share the similar spatiotemporal gradients, whose similarity distribution is consistent with the extracted boundary distribution. This motivates us to design the common spatiotemporal gradient to constrain the reference boundary localization. In motion correlation fusion, we discover that the frame-based motion possesses spatially dense but temporally discontinuous correlation, while the event-based motion has spatially sparse but temporally continuous correlation. This inspires us to use the reference boundary to guide the complementary motion knowledge fusion between the two modalities. Moreover, common spatiotemporal fusion can not only relieve the cross-modal feature discrepancy, but also make the fusion process interpretable for dense and continuous optical flow. Extensive experiments have been performed to verify the superiority of the proposed method.
Authors: Stylianos Zindros, Christos Chronis, Panagiotis Radoglou-Grammatikis, Vasileios Argyriou, Panagiotis Sarigiannidis, Iraklis Varlamis, Georgios Th. Papadopoulos
Abstract: As the security of public spaces remains a critical issue in today's world, Digital Twin technologies have emerged in recent years as a promising solution for detecting and predicting potential future threats. The applied methodology leverages a Digital Twin of a metro station in Athens, Greece, using the FlexSim simulation software. The model encompasses points of interest and passenger flows, and sets their corresponding parameters. These elements influence and allow the model to provide reasonable predictions on the security management of the station under various scenarios. Experimental tests are conducted with different configurations of surveillance cameras and optimizations of camera angles to evaluate the effectiveness of the space surveillance setup. The results show that the strategic positioning of surveillance cameras and the adjustment of their angles significantly improves the detection of suspicious behaviors and with the use of the DT it is possible to evaluate different scenarios and find the optimal camera setup for each case. In summary, this study highlights the value of Digital Twins in real-time simulation and data-driven security management. The proposed approach contributes to the ongoing development of smart security solutions for public spaces and provides an innovative framework for threat detection and prevention.
Authors: Haoyu Zheng, Qifan Yu, Binghe Yu, Yang Dai, Wenqiao Zhang, Juncheng Li, Siliang Tang, Yueting Zhuang
Abstract: Diffusion models have achieved remarkable progress in image and video stylization. However, most existing methods focus on single-style transfer, while video stylization involving multiple styles necessitates seamless transitions between them. We refer to this smooth style transition between video frames as video style morphing. Current approaches often generate stylized video frames with discontinuous structures and abrupt style changes when handling such transitions. To address these limitations, we introduce SOYO, a novel diffusion-based framework for video style morphing. Our method employs a pre-trained text-to-image diffusion model without fine-tuning, combining attention injection and AdaIN to preserve structural consistency and enable smooth style transitions across video frames. Moreover, we notice that applying linear equidistant interpolation directly induces imbalanced style morphing. To harmonize across video frames, we propose a novel adaptive sampling scheduler operating between two style images. Extensive experiments demonstrate that SOYO outperforms existing methods in open-domain video style morphing, better preserving the structural coherence of video frames while achieving stable and smooth style transitions.
Authors: Zhaojie Zeng, Yuesong Wang, Lili Ju, Tao Guan
Abstract: By adaptively controlling the density and generating more Gaussians in regions with high-frequency information, 3D Gaussian Splatting (3DGS) can better represent scene details. From the signal processing perspective, representing details usually needs more Gaussians with relatively smaller scales. However, 3DGS currently lacks an explicit constraint linking the density and scale of 3D Gaussians across the domain, leading to 3DGS using improper-scale Gaussians to express frequency information, resulting in the loss of accuracy. In this paper, we propose to establish a direct relation between density and scale through the reparameterization of the scaling parameters and ensure the consistency between them via explicit constraints (i.e., density responds well to changes in frequency). Furthermore, we develop a frequency-aware density control strategy, consisting of densification and deletion, to improve representation quality with fewer Gaussians. A dynamic threshold encourages densification in high-frequency regions, while a scale-based filter deletes Gaussians with improper scale. Experimental results on various datasets demonstrate that our method outperforms existing state-of-the-art methods quantitatively and qualitatively.
Authors: Jiazheng Liu, Sipeng Zheng, B\"orje F. Karlsson, Zongqing Lu
Abstract: Multimodal large language models (MLLMs), built on large-scale pre-trained vision towers and language models, have shown great capabilities in multimodal understanding. However, most existing MLLMs are trained on single-turn vision question-answering tasks, which do not accurately reflect real-world human conversations. In this paper, we introduce MMDiag, a multi-turn multimodal dialogue dataset. This dataset is collaboratively generated through deliberately designed rules and GPT assistance, featuring strong correlations between questions, between questions and images, and among different image regions; thus aligning more closely with real-world scenarios. MMDiag serves as a strong benchmark for multi-turn multimodal dialogue learning and brings more challenges to the grounding and reasoning capabilities of MLLMs. Further, inspired by human vision processing, we present DiagNote, an MLLM equipped with multimodal grounding and reasoning capabilities. DiagNote consists of two modules (Deliberate and Gaze) interacting with each other to perform Chain-of-Thought and annotations respectively, throughout multi-turn dialogues. We empirically demonstrate the advantages of DiagNote in both grounding and jointly processing and reasoning with vision and language information over existing MLLMs.
Authors: Jiaojiao Li, Shiyao Duan, Haitao XU, Rui Song
Abstract: The inherent difficulty in acquiring accurately co-registered RGB-hyperspectral image (HSI) pairs has significantly impeded the practical deployment of current data-driven Hyperspectral Image Generation (HIG) networks in engineering applications. Gleichzeitig, the ill-posed nature of the aligning constraints, compounded with the complexities of mining cross-domain features, also hinders the advancement of unpaired HIG (UnHIG) tasks. In this paper, we conquer these challenges by modeling the UnHIG to range space interaction and compensations of null space through Range-Null Space Decomposition (RND) methodology. Specifically, the introduced contrastive learning effectively aligns the geometric and spectral distributions of unpaired data by building the interaction of range space, considering the consistent feature in degradation process. Following this, we map the frequency representations of dual-domain input and thoroughly mining the null space, like degraded and high-frequency components, through the proposed Non-uniform Kolmogorov-Arnold Networks. Extensive comparative experiments demonstrate that it establishes a new benchmark in UnHIG.
Authors: Sania Zahan, Ghulam Mubashar Hassan, Ajmal Mian
Abstract: Older people are susceptible to fall due to instability in posture and deteriorating health. Immediate access to medical support can greatly reduce repercussions. Hence, there is an increasing interest in automated fall detection, often incorporated into a smart healthcare system to provide better monitoring. Existing systems focus on wearable devices which are inconvenient or video monitoring which has privacy concerns. Moreover, these systems provide a limited perspective of their generalization ability as they are tested on datasets containing few activities that have wide disparity in the action space and are easy to differentiate. Complex daily life scenarios pose much greater challenges with activities that overlap in action spaces due to similar posture or motion. To overcome these limitations, we propose a fall detection model, coined SDFA, based on human skeletons extracted from low-resolution videos. The use of skeleton data ensures privacy and low-resolution videos ensures low hardware and computational cost. Our model captures discriminative structural displacements and motion trends using unified joint and motion features projected onto a shared high dimensional space. Particularly, the use of separable convolution combined with a powerful GCN architecture provides improved performance. Extensive experiments on five large-scale datasets with a wide range of evaluation settings show that our model achieves competitive performance with extremely low computational complexity and runs faster than existing models.
Authors: Keyu Du, Hao Xu, Haipeng Li, Hong Qu, Chi-Wing Fu, Shuaicheng Liu
Abstract: Scene-level point cloud registration is very challenging when considering dynamic foregrounds. Existing indoor datasets mostly assume rigid motions, so the trained models cannot robustly handle scenes with non-rigid motions. On the other hand, non-rigid datasets are mainly object-level, so the trained models cannot generalize well to complex scenes. This paper presents HybridReg, a new approach to 3D point cloud registration, learning uncertainty mask to account for hybrid motions: rigid for backgrounds and non-rigid/rigid for instance-level foregrounds. First, we build a scene-level 3D registration dataset, namely HybridMatch, designed specifically with strategies to arrange diverse deforming foregrounds in a controllable manner. Second, we account for different motion types and formulate a mask-learning module to alleviate the interference of deforming outliers. Third, we exploit a simple yet effective negative log-likelihood loss to adopt uncertainty to guide the feature extraction and correlation computation. To our best knowledge, HybridReg is the first work that exploits hybrid motions for robust point cloud registration. Extensive experiments show HybridReg's strengths, leading it to achieve state-of-the-art performance on both widely-used indoor and outdoor datasets.
Authors: Yi Liu, Hao Zhou, Wenxiang Shang, Ran Lin, Benlei Cui
Abstract: Erase inpainting, or object removal, aims to precisely remove target objects within masked regions while preserving the overall consistency of the surrounding content. Despite diffusion-based methods have made significant strides in the field of image inpainting, challenges remain regarding the emergence of unexpected objects or artifacts. We assert that the inexact diffusion pathways established by existing standard optimization paradigms constrain the efficacy of object removal. To tackle these challenges, we propose a novel Erase Diffusion, termed EraDiff, aimed at unleashing the potential power of standard diffusion in the context of object removal. In contrast to standard diffusion, the EraDiff adapts both the optimization paradigm and the network to improve the coherence and elimination of the erasure results. We first introduce a Chain-Rectifying Optimization (CRO) paradigm, a sophisticated diffusion process specifically designed to align with the objectives of erasure. This paradigm establishes innovative diffusion transition pathways that simulate the gradual elimination of objects during optimization, allowing the model to accurately capture the intent of object removal. Furthermore, to mitigate deviations caused by artifacts during the sampling pathways, we develop a simple yet effective Self-Rectifying Attention (SRA) mechanism. The SRA calibrates the sampling pathways by altering self-attention activation, allowing the model to effectively bypass artifacts while further enhancing the coherence of the generated content. With this design, our proposed EraDiff achieves state-of-the-art performance on the OpenImages V5 dataset and demonstrates significant superiority in real-world scenarios.
Authors: Yuxuan Zhang, Yirui Yuan, Yiren Song, Haofan Wang, Jiaming Liu
Abstract: Recent advancements in Unet-based diffusion models, such as ControlNet and IP-Adapter, have introduced effective spatial and subject control mechanisms. However, the DiT (Diffusion Transformer) architecture still struggles with efficient and flexible control. To tackle this issue, we propose EasyControl, a novel framework designed to unify condition-guided diffusion transformers with high efficiency and flexibility. Our framework is built on three key innovations. First, we introduce a lightweight Condition Injection LoRA Module. This module processes conditional signals in isolation, acting as a plug-and-play solution. It avoids modifying the base model weights, ensuring compatibility with customized models and enabling the flexible injection of diverse conditions. Notably, this module also supports harmonious and robust zero-shot multi-condition generalization, even when trained only on single-condition data. Second, we propose a Position-Aware Training Paradigm. This approach standardizes input conditions to fixed resolutions, allowing the generation of images with arbitrary aspect ratios and flexible resolutions. At the same time, it optimizes computational efficiency, making the framework more practical for real-world applications. Third, we develop a Causal Attention Mechanism combined with the KV Cache technique, adapted for conditional generation tasks. This innovation significantly reduces the latency of image synthesis, improving the overall efficiency of the framework. Through extensive experiments, we demonstrate that EasyControl achieves exceptional performance across various application scenarios. These innovations collectively make our framework highly efficient, flexible, and suitable for a wide range of tasks.
Authors: Dong-Hee Paek, Seung-Hyun Kong
Abstract: Sensor fusion of camera, LiDAR, and 4-dimensional (4D) Radar has brought a significant performance improvement in autonomous driving (AD). However, there still exist fundamental challenges: deeply coupled fusion methods assume continuous sensor availability, making them vulnerable to sensor degradation and failure, whereas sensor-wise cross-attention fusion methods struggle with computational cost and unified feature representation. This paper presents availability-aware sensor fusion (ASF), a novel method that employs unified canonical projection (UCP) to enable consistency in all sensor features for fusion and cross-attention across sensors along patches (CASAP) to enhance robustness of sensor fusion against sensor degradation and failure. As a result, the proposed ASF shows a superior object detection performance to the existing state-of-the-art fusion methods under various weather and sensor degradation (or failure) conditions; Extensive experiments on the K-Radar dataset demonstrate that ASF achieves improvements of 9.7% in AP BEV (87.2%) and 20.1% in AP 3D (73.6%) in object detection at IoU=0.5, while requiring a low computational cost. The code will be available at https://github.com/kaist-avelab/K-Radar.
Authors: Haolong Ma, Hui Li, Chunyang Cheng, Zeyang Zhang, Xiaoning Song, Xiao-Jun Wu
Abstract: All-in-One Degradation-Aware Fusion Models (ADFMs), a class of multi-modal image fusion models, address complex scenes by mitigating degradations from source images and generating high-quality fused images. Mainstream ADFMs often rely on highly synthetic multi-modal multi-quality images for supervision, limiting their effectiveness in cross-modal and rare degradation scenarios. The inherent relationship among these multi-modal, multi-quality images of the same scene provides explicit supervision for training, but also raises above problems. To address these limitations, we present LURE, a Learning-driven Unified Representation model for infrared and visible Image Fusion, which is degradation-aware. LURE decouples multi-modal multi-quality data at the data level and recouples this relationship in a unified latent feature space (ULFS) by proposing a novel unified loss. This decoupling circumvents data-level limitations of prior models and allows leveraging real-world restoration datasets for training high-quality degradation-aware models, sidestepping above issues. To enhance text-image interaction, we refine image-text interaction and residual structures via Text-Guided Attention (TGA) and an inner residual structure. These enhances text's spatial perception of images and preserve more visual details. Experiments show our method outperforms state-of-the-art (SOTA) methods across general fusion, degradation-aware fusion, and downstream tasks. The code will be publicly available.
Authors: Sheng Luo, Yi Zhou, Tao Zhou
Abstract: Incremental learning (IL) aims to overcome catastrophic forgetting of previous tasks while learning new ones. Existing IL methods make strong assumptions that the incoming task type will either only increases new classes or domains (i.e. Class IL, Domain IL), or increase by a static scale in a class- and domain-agnostic manner (i.e. Versatile IL (VIL)), which greatly limit their applicability in the unpredictable and dynamic wild. In this work, we investigate $\textbf{Universal Incremental Learning (UIL)}$, where a model neither knows which new classes or domains will increase along sequential tasks, nor the scale of the increments within each task. This uncertainty prevents the model from confidently learning knowledge from all task distributions and symmetrically focusing on the diverse knowledge within each task distribution. Consequently, UIL presents a more general and realistic IL scenario, making the model face confusion arising from inter-task and intra-task distribution randomness. To $\textbf{Mi}$tigate both $\textbf{Co}$nfusion, we propose a simple yet effective framework for UIL, named $\textbf{MiCo}$. At the inter-task distribution level, we employ a multi-objective learning scheme to enforce accurate and deterministic predictions, and its effectiveness is further enhanced by a direction recalibration module that reduces conflicting gradients. Moreover, at the intra-task distribution level, we introduce a magnitude recalibration module to alleviate asymmetrical optimization towards imbalanced class distribution. Extensive experiments on three benchmarks demonstrate the effectiveness of our method, outperforming existing state-of-the-art methods in both the UIL scenario and the VIL scenario. Our code will be available at $\href{https://github.com/rolsheng/UIL}{here}$.
Authors: Yan Jiang, Zhongmiao Qi, Jianhao Li, Jiangbo Qian, Chong Wang, Yu Xin
Abstract: Hashing algorithms have been widely used in large-scale image retrieval tasks, especially for seen class data. Zero-shot hashing algorithms have been proposed to handle unseen class data. The key technique in these algorithms involves learning features from seen classes and transferring them to unseen classes, that is, aligning the feature embeddings between the seen and unseen classes. Most existing zero-shot hashing algorithms use the shared attributes between the two classes of interest to complete alignment tasks. However, the attributes are always described for a whole image, even though they represent specific parts of the image. Hence, these methods ignore the importance of aligning attributes with the corresponding image parts, which explicitly introduces noise and reduces the accuracy achieved when aligning the features of seen and unseen classes. To address this problem, we propose a new zero-shot hashing method called RAZH. We first use a clustering algorithm to group similar patches to image parts for attribute matching and then replace the image parts with the corresponding attribute vectors, gradually aligning each part with its nearest attribute. Extensive evaluation results demonstrate the superiority of the RAZH method over several state-of-the-art methods.
Authors: Mihcael Green, Matan Levy, Issar Tzachor, Dvir Samuel, Nir Darshan, Rami Ben-Ari
Abstract: We address the challenge of Small Object Image Retrieval (SoIR), where the goal is to retrieve images containing a specific small object, in a cluttered scene. The key challenge in this setting is constructing a single image descriptor, for scalable and efficient search, that effectively represents all objects in the image. In this paper, we first analyze the limitations of existing methods on this challenging task and then introduce new benchmarks to support SoIR evaluation. Next, we introduce Multi-object Attention Optimization (MaO), a novel retrieval framework which incorporates a dedicated multi-object pre-training phase. This is followed by a refinement process that leverages attention-based feature extraction with object masks, integrating them into a single unified image descriptor. Our MaO approach significantly outperforms existing retrieval methods and strong baselines, achieving notable improvements in both zero-shot and lightweight multi-object fine-tuning. We hope this work will lay the groundwork and inspire further research to enhance retrieval performance for this highly practical task.
Authors: Juntian Du, Yuan Sun, Zhihu Zhou, Pinyi Chen, Runzhe Zhang, Keji Mao
Abstract: Optical flow estimation based on deep learning, particularly the recently proposed top-performing methods that incorporate the Transformer, has demonstrated impressive performance, due to the Transformer's powerful global modeling capabilities. However, the quadratic computational complexity of attention mechanism in the Transformers results in time-consuming training and inference. To alleviate these issues, we propose a novel MambaFlow framework that leverages the high accuracy and efficiency of Mamba architecture to capture features with local correlation while preserving its global information, achieving remarkable performance. To the best of our knowledge, the proposed method is the first Mamba-centric architecture for end-to-end optical flow estimation. It comprises two primary contributed components, both of which are Mamba-centric: a feature enhancement Mamba (FEM) module designed to optimize feature representation quality and a flow propagation Mamba (FPM) module engineered to address occlusion issues by facilitate effective flow information dissemination. Extensive experiments demonstrate that our approach achieves state-of-the-art results, despite encountering occluded regions. On the Sintel benchmark, MambaFlow achieves an EPE all of 1.60, surpassing the leading 1.74 of GMFlow. Additionally, MambaFlow significantly improves inference speed with a runtime of 0.113 seconds, making it 18% faster than GMFlow. The source code will be made publicly available upon acceptance of the paper.
Authors: Yongle Zhang, Yimin Liu, Qiang Wu
Abstract: Diffusion models have become widely adopted in image completion tasks, with text prompts commonly employed to ensure semantic coherence by providing high-level guidance. However, a persistent challenge arises when an object is partially obscured in the damaged region, yet its remaining parts are still visible in the background. While text prompts offer semantic direction, they often fail to precisely recover fine-grained structural details, such as the object's overall posture, ensuring alignment with the visible object information in the background. This limitation stems from the inability of text prompts to provide pixel-level specificity. To address this, we propose supplementing text-based guidance with a novel visual aid: a casual sketch, which can be roughly drawn by anyone based on visible object parts. This sketch supplies critical structural cues, enabling the generative model to produce an object structure that seamlessly integrates with the existing background. We introduce the Visual Sketch Self-Aware (VSSA) model, which integrates the casual sketch into each iterative step of the diffusion process, offering distinct advantages for partially corrupted scenarios. By blending sketch-derived features with those of the corrupted image, and leveraging text prompt guidance, the VSSA assists the diffusion model in generating images that preserve both the intended object semantics and structural consistency across the restored objects and original regions. To support this research, we created two datasets, CUB-sketch and MSCOCO-sketch, each combining images, sketches, and text. Extensive qualitative and quantitative experiments demonstrate that our approach outperforms several state-of-the-art methods.
Authors: Victor Shea-Jay Huang, Le Zhuo, Yi Xin, Zhaokai Wang, Peng Gao, Hongsheng Li
Abstract: Diffusion Transformers (DiTs) are a powerful yet underexplored class of generative models compared to U-Net-based diffusion models. To bridge this gap, we introduce TIDE (Temporal-aware Sparse Autoencoders for Interpretable Diffusion transformErs), a novel framework that enhances temporal reconstruction within DiT activation layers across denoising steps. TIDE employs Sparse Autoencoders (SAEs) with a sparse bottleneck layer to extract interpretable and hierarchical features, revealing that diffusion models inherently learn hierarchical features at multiple levels (e.g., 3D, semantic, class) during generative pre-training. Our approach achieves state-of-the-art reconstruction performance, with a mean squared error (MSE) of 1e-3 and a cosine similarity of 0.97, demonstrating superior accuracy in capturing activation dynamics along the denoising trajectory. Beyond interpretability, we showcase TIDE's potential in downstream applications such as sparse activation-guided image editing and style transfer, enabling improved controllability for generative systems. By providing a comprehensive training and evaluation protocol tailored for DiTs, TIDE contributes to developing more interpretable, transparent, and trustworthy generative models.
Authors: Amira Guesmi, Bassem Ouni, Muhammad Shafique
Abstract: Quantized Neural Networks (QNNs) have emerged as a promising solution for reducing model size and computational costs, making them well-suited for deployment in edge and resource-constrained environments. While quantization is known to disrupt gradient propagation and enhance robustness against pixel-level adversarial attacks, its effectiveness against patch-based adversarial attacks remains largely unexplored. In this work, we demonstrate that adversarial patches remain highly transferable across quantized models, achieving over 70\% attack success rates (ASR) even at extreme bit-width reductions (e.g., 2-bit). This challenges the common assumption that quantization inherently mitigates adversarial threats. To address this, we propose Quantization-Aware Defense Training with Randomization (QADT-R), a novel defense strategy that integrates Adaptive Quantization-Aware Patch Generation (A-QAPA), Dynamic Bit-Width Training (DBWT), and Gradient-Inconsistent Regularization (GIR) to enhance resilience against highly transferable patch-based attacks. A-QAPA generates adversarial patches within quantized models, ensuring robustness across different bit-widths. DBWT introduces bit-width cycling during training to prevent overfitting to a specific quantization setting, while GIR injects controlled gradient perturbations to disrupt adversarial optimization. Extensive evaluations on CIFAR-10 and ImageNet show that QADT-R reduces ASR by up to 25\% compared to prior defenses such as PBAT and DWQ. Our findings further reveal that PBAT-trained models, while effective against seen patch configurations, fail to generalize to unseen patches due to quantization shift. Additionally, our empirical analysis of gradient alignment, spatial sensitivity, and patch visibility provides insights into the mechanisms that contribute to the high transferability of patch-based attacks in QNNs.
Authors: Huilin Deng, Ding Zou, Rui Ma, Hongchen Luo, Yang Cao, Yu Kang
Abstract: While state-of-the-art vision-language models (VLMs) have demonstrated remarkable capabilities in complex visual-text tasks, their success heavily relies on massive model scaling, limiting their practical deployment. Small-scale VLMs offer a more practical alternative but face significant challenges when trained with traditional supervised fine-tuning (SFT), particularly in two aspects: out-of-domain (OOD) generalization and reasoning abilities, which significantly lags behind the contemporary Large language models (LLMs). To address these challenges, we propose Curriculum Reinforcement Finetuning (Curr-ReFT), a novel post-training paradigm specifically designed for small-scale VLMs. Inspired by the success of reinforcement learning in LLMs, Curr-ReFT comprises two sequential stages: (1) Curriculum Reinforcement Learning, which ensures steady progression of model capabilities through difficulty-aware reward design, transitioning from basic visual perception to complex reasoning tasks; and (2) Rejected Sampling-based Self-improvement, which maintains the fundamental capabilities of VLMs through selective learning from high-quality multimodal and language examples. Extensive experiments demonstrate that models trained with Curr-ReFT paradigm achieve state-of-the-art performance across various visual tasks in both in-domain and out-of-domain settings. Moreover, our Curr-ReFT enhanced 3B model matches the performance of 32B-parameter models, demonstrating that efficient training paradigms can effectively bridge the gap between small and large models.
Authors: Chuanming Wang, Henming Mao, Huanhuan Zhang, Huiyuan Fu, Huadong Ma
Abstract: Vision-Language Models (VLMs) have demonstrated impressive performance on various visual tasks, yet they still require adaptation on downstream tasks to achieve optimal performance. Recently, various adaptation technologies have been proposed, but we observe they often underperform in fine-grained visual recognition, which requires models to capture subtle yet discriminative features to distinguish similar sub-categories. Current adaptation methods typically rely on an alignment-based prediction framework, \ie the visual feature is compared with each class prompt for similarity calculation as the final prediction, which lacks class interaction during the forward pass. Besides, learning single uni-modal feature further restricts the model's expressive capacity. Therefore, we propose a novel mechanism, XR-VLM, to discover subtle differences by modeling cross-relationships, which specifically excels in scenarios involving multiple features. Our method introduces a unified multi-part visual feature extraction module designed to seamlessly integrate with the diverse backbones inherent in VLMs. Additionally, we develop a multi-part prompt learning module to capture multi-perspective descriptions of sub-categories. To further enhance discriminative capability, we propose a cross relationship modeling pattern that combines visual feature with all class prompt features, enabling a deeper exploration of the relationships between these two modalities. Extensive experiments have been conducted on various fine-grained datasets, and the results demonstrate that our method achieves significant improvements compared to current state-of-the-art approaches. Code will be released.
Authors: Zhihao Huang, Xi Qiu, Yukuo Ma, Yifu Zhou, Chi Zhang, Xuelong Li
Abstract: Autoregressive models have achieved promising results in natural language processing. However, for image generation tasks, they encounter substantial challenges in effectively capturing long-range dependencies, managing computational costs, and most crucially, defining meaningful autoregressive sequences that reflect natural image hierarchies. To address these issues, we present \textbf{N}ext-\textbf{F}requency \textbf{I}mage \textbf{G}eneration (\textbf{NFIG}), a novel framework that decomposes the image generation process into multiple frequency-guided stages. Our approach first generates low-frequency components to establish global structure with fewer tokens, then progressively adds higher-frequency details, following the natural spectral hierarchy of images. This principled autoregressive sequence not only improves the quality of generated images by better capturing true causal relationships between image components, but also significantly reduces computational overhead during inference. Extensive experiments demonstrate that NFIG achieves state-of-the-art performance with fewer steps, offering a more efficient solution for image generation, with 1.25$\times$ speedup compared to VAR-d20 while achieving better performance (FID: 2.81) on the ImageNet-256 benchmark. We hope that our insight of incorporating frequency-domain knowledge to guide autoregressive sequence design will shed light on future research. We will make our code publicly available upon acceptance of the paper.
Authors: Spyros Kondylatos, Nikolaos Ioannis Bountos, Dimitrios Michail, Xiao Xiang Zhu, Gustau Camps-Valls, Ioannis Papoutsis
Abstract: Recent advances in Computer Vision have introduced the concept of pretrained representation uncertainty, enabling zero-shot uncertainty estimation. This holds significant potential for Earth Observation (EO), where trustworthiness is critical, yet the complexity of EO data poses challenges to uncertainty-aware methods. In this work, we investigate the generalization of representation uncertainty in EO, considering the domain's unique semantic characteristics. We pretrain uncertainties on large EO datasets and propose an evaluation framework to assess their zero-shot performance in multi-label classification and segmentation EO tasks. Our findings reveal that, unlike uncertainties pretrained on natural images, EO-pretraining exhibits strong generalization across unseen EO domains, geographic locations, and target granularities, while maintaining sensitivity to variations in ground sampling distance. We demonstrate the practical utility of pretrained uncertainties showcasing their alignment with task-specific uncertainties in downstream tasks, their sensitivity to real-world EO image noise, and their ability to generate spatial uncertainty estimates out-of-the-box. Initiating the discussion on representation uncertainty in EO, our study provides insights into its strengths and limitations, paving the way for future research in the field. Code and weights are available at: https://github.com/Orion-AI-Lab/EOUncertaintyGeneralization.
URLs: https://github.com/Orion-AI-Lab/EOUncertaintyGeneralization.
Authors: Shuhe Wang, Xiaoya Li, Jiwei Li, Guoyin Wang, Xiaofei Sun, Bob Zhu, Han Qiu, Mo Yu, Shengjie Shen, Eduard Hovy
Abstract: Due to the data-driven nature of current face identity (FaceID) customization methods, all state-of-the-art models rely on large-scale datasets containing millions of high-quality text-image pairs for training. However, none of these datasets are publicly available, which restricts transparency and hinders further advancements in the field. To address this issue, in this paper, we collect and release FaceID-6M, the first large-scale, open-source FaceID dataset containing 6 million high-quality text-image pairs. Filtered from LAION-5B \cite{schuhmann2022laion}, FaceID-6M undergoes a rigorous image and text filtering steps to ensure dataset quality, including resolution filtering to maintain high-quality images and faces, face filtering to remove images that lack human faces, and keyword-based strategy to retain descriptions containing human-related terms (e.g., nationality, professions and names). Through these cleaning processes, FaceID-6M provides a high-quality dataset optimized for training powerful FaceID customization models, facilitating advancements in the field by offering an open resource for research and development. We conduct extensive experiments to show the effectiveness of our FaceID-6M, demonstrating that models trained on our FaceID-6M dataset achieve performance that is comparable to, and slightly better than currently available industrial models. Additionally, to support and advance research in the FaceID customization community, we make our code, datasets, and models fully publicly available. Our codes, models, and datasets are available at: https://github.com/ShuheSH/FaceID-6M.
Authors: Ding Zhong, Xu Zheng, Chenfei Liao, Yuanhuiyi Lyu, Jialei Chen, Shengyang Wu, Linfeng Zhang, Xuming Hu
Abstract: Segment Anything Model 2 (SAM2) has emerged as a strong base model in various pinhole imaging segmentation tasks. However, when applying it to $360^\circ$ domain, the significant field-of-view (FoV) gap between pinhole ($70^\circ \times 70^\circ$) and panoramic images ($180^\circ \times 360^\circ$) poses unique challenges. Two major concerns for this application includes 1) inevitable distortion and object deformation brought by the large FoV disparity between domains; 2) the lack of pixel-level semantic understanding that the original SAM2 cannot provide. To address these issues, we propose a novel OmniSAM framework, which makes the first attempt to apply SAM2 for panoramic semantic segmentation. Specifically, to bridge the first gap, OmniSAM first divides the panorama into sequences of patches. These patches are then treated as image sequences in similar manners as in video segmentation tasks. We then leverage the SAM2's memory mechanism to extract cross-patch correspondences that embeds the cross-FoV dependencies, improving feature continuity and the prediction consistency along mask boundaries. For the second gap, OmniSAM fine-tunes the pretrained image encoder and reutilize the mask decoder for semantic prediction. An FoV-based prototypical adaptation module with dynamic pseudo label update mechanism is also introduced to facilitate the alignment of memory and backbone features, thereby improving model generalization ability across different sizes of source models. Extensive experimental results demonstrate that OmniSAM outperforms the state-of-the-art methods by large margins, e.g., 79.06% (+10.22%) on SPin8-to-SPan8, 62.46% (+6.58%) on CS13-to-DP13.
Authors: Haiyang Xie, Xi Shen, Shihua Huang, Zheng Wang
Abstract: Most visual models are designed for sRGB images, yet RAW data offers significant advantages for object detection by preserving sensor information before ISP processing. This enables improved detection accuracy and more efficient hardware designs by bypassing the ISP. However, RAW object detection is challenging due to limited training data, unbalanced pixel distributions, and sensor noise. To address this, we propose SimROD, a lightweight and effective approach for RAW object detection. We introduce a Global Gamma Enhancement (GGE) module, which applies a learnable global gamma transformation with only four parameters, improving feature representation while keeping the model efficient. Additionally, we leverage the green channel's richer signal to enhance local details, aligning with the human eye's sensitivity and Bayer filter design. Extensive experiments on multiple RAW object detection datasets and detectors demonstrate that SimROD outperforms state-of-the-art methods like RAW-Adapter and DIAP while maintaining efficiency. Our work highlights the potential of RAW data for real-world object detection.
Authors: Hanqing Guo, Xiuxiu Lin, Shiyu Zhao
Abstract: Vision-based drone-to-drone detection has attracted increasing attention due to its importance in numerous tasks such as vision-based swarming, aerial see-and-avoid, and malicious drone detection. However, existing methods often encounter failures when the background is complex or the target is tiny. This paper proposes a novel end-to-end framework that accurately identifies small drones in complex environments using motion guidance. It starts by creating a motion difference map to capture the motion characteristics of tiny drones. Next, this motion difference map is combined with an RGB image using a bimodal fusion module, allowing for adaptive feature learning of the drone. Finally, the fused feature map is processed through an enhanced backbone and detection head based on the YOLOv5 framework to achieve accurate detection results. To validate our method, we propose a new dataset, named ARD100, which comprises 100 videos (202,467 frames) covering various challenging conditions and has the smallest average object size compared with the existing drone detection datasets. Extensive experiments on the ARD100 and NPS-Drones datasets show that our proposed detector performs exceptionally well under challenging conditions and surpasses state-of-the-art algorithms across various metrics. We publicly release the codes and ARD100 dataset at https://github.com/Irisky123/YOLOMG.
Authors: Zhen Zou, Hu Yu, Jie Xiao, Feng Zhao
Abstract: Diffusion Transformer (DiT) has exhibited impressive generation capabilities but faces great challenges due to its high computational complexity. To address this problem, various methods, notably feature caching, have been introduced. However, these approaches focus on aligning non-cache diffusion without analyzing the impact of caching on the generation of intermediate processes. So the lack of exploration provides us with room for analysis and improvement. In this paper, we analyze the impact of caching on the SNR of the diffusion process and discern that feature caching intensifies the denoising procedure, and we further identify this as a more severe exposure bias issue. Drawing on this insight, we introduce EB-Cache, a joint cache strategy that aligns the Non-exposure bias (which gives us a higher performance ceiling) diffusion process. Our approach incorporates a comprehensive understanding of caching mechanisms and offers a novel perspective on leveraging caches to expedite diffusion processes. Empirical results indicate that EB-Cache optimizes model performance while concurrently facilitating acceleration. Specifically, in the 50-step generation process, EB-Cache achieves 1.49$\times$ acceleration with 0.63 FID reduction from 3.69, surpassing prior acceleration methods. Code will be available at \href{https://github.com/aSleepyTree/EB-Cache}{https://github.com/aSleepyTree/EB-Cache}.
URLs: https://github.com/aSleepyTree/EB-Cache, https://github.com/aSleepyTree/EB-Cache
Authors: Sihao Lin, Daqi Liu, Ruochong Fu, Dongrui Liu, Andy Song, Hongwei Xie, Zhihui Li, Bing Wang, Xiaojun Chang
Abstract: Estimating the 3D world from 2D monocular images is a fundamental yet challenging task due to the labour-intensive nature of 3D annotations. To simplify label acquisition, this work proposes a novel approach that bridges 2D vision foundation models (VFMs) with 3D tasks by decoupling 3D supervision into an ensemble of image-level primitives, e.g., semantic and geometric components. As a key motivator, we leverage the zero-shot capabilities of vision-language models for image semantics. However, due to the notorious ill-posed problem - multiple distinct 3D scenes can produce identical 2D projections, directly inferring metric depth from a monocular image in a zero-shot manner is unsuitable. In contrast, 2D VFMs provide promising sources of relative depth, which theoretically aligns with metric depth when properly scaled and offset. Thus, we adapt the relative depth derived from VFMs into metric depth by optimising the scale and offset using temporal consistency, also known as novel view synthesis, without access to ground-truth metric depth. Consequently, we project the semantics into 3D space using the reconstructed metric depth, thereby providing 3D supervision. Extensive experiments on nuScenes and SemanticKITTI demonstrate the effectiveness of our framework. For instance, the proposed method surpasses the current state-of-the-art by 3.34% mIoU on nuScenes for voxel occupancy prediction.
Authors: Marcelo Eduardo Pederiva, Jos\'e Mario De Martino, Alessandro Zimmer
Abstract: Comprehending the environment and accurately detecting objects in 3D space are essential for advancing autonomous vehicle technologies. Integrating Camera and LIDAR data has emerged as an effective approach for achieving high accuracy in 3D Object Detection models. However, existing methodologies often rely on heavy, traditional backbones that are computationally demanding. This paper introduces a novel approach that incorporates cutting-edge Deep Learning techniques into the feature extraction process, aiming to create more efficient models without compromising performance. Our model, NextBEV, surpasses established feature extractors like ResNet50 and MobileNetV2. On the KITTI 3D Monocular detection benchmark, NextBEV achieves an accuracy improvement of 2.39%, having less than 10% of the MobileNetV3 parameters. Moreover, we propose changes in LIDAR backbones that decreased the original inference time to 10 ms. Additionally, by fusing these lightweight proposals, we have enhanced the accuracy of the VoxelNet-based model by 2.93% and improved the F1-score of the PointPillar-based model by approximately 20%. Therefore, this work contributes to establishing lightweight and powerful models for individual or fusion techniques, making them more suitable for onboard implementations.
Authors: Yuheng Liu, Xinke Li, Yuning Zhang, Lu Qi, Xin Li, Wenping Wang, Chongshou Li, Xueting Li, Ming-Hsuan Yang
Abstract: Three-dimensional scene generation is crucial in computer vision, with applications spanning autonomous driving, gaming and the metaverse. Current methods either lack user control or rely on imprecise, non-intuitive conditions. In this work, we propose a method that uses, scene graphs, an accessible, user friendly control format to generate outdoor 3D scenes. We develop an interactive system that transforms a sparse scene graph into a dense BEV (Bird's Eye View) Embedding Map, which guides a conditional diffusion model to generate 3D scenes that match the scene graph description. During inference, users can easily create or modify scene graphs to generate large-scale outdoor scenes. We create a large-scale dataset with paired scene graphs and 3D semantic scenes to train the BEV embedding and diffusion models. Experimental results show that our approach consistently produces high-quality 3D urban scenes closely aligned with the input scene graphs. To the best of our knowledge, this is the first approach to generate 3D outdoor scenes conditioned on scene graphs.
Authors: Hung Q. Vo, Pengyu Yuan, Zheng Yin, Kelvin K. Wong, Chika F. Ezeana, Son T. Ly, Stephen T. C. Wong, Hien V. Nguyen
Abstract: Self-supervised learning (SSL) has garnered substantial interest within the machine learning and computer vision communities. Two prominent approaches in SSL include contrastive-based learning and self-distillation utilizing cropping augmentation. Lately, masked image modeling (MIM) has emerged as a more potent SSL technique, employing image inpainting as a pretext task. MIM creates a strong inductive bias toward meaningful spatial and semantic understanding. This has opened up new opportunities for SSL to contribute not only to classification tasks but also to more complex applications like object detection and image segmentation. Building upon this progress, our research paper introduces a scalable and practical SSL approach centered around more challenging pretext tasks that facilitate the acquisition of robust features. Specifically, we leverage multi-scale image reconstruction from randomly masked input images as the foundation for feature learning. Our hypothesis posits that reconstructing high-resolution images enables the model to attend to finer spatial details, particularly beneficial for discerning subtle intricacies within medical images. The proposed SSL features help improve classification performance on the Curated Breast Imaging Subset of Digital Database for Screening Mammography (CBIS-DDSM) dataset. In pathology classification, our method demonstrates a 3\% increase in average precision (AP) and a 1\% increase in the area under the receiver operating characteristic curve (AUC) when compared to state-of-the-art (SOTA) algorithms. Moreover, in mass margins classification, our approach achieves a 4\% increase in AP and a 2\% increase in AUC.
Authors: Ziliang Miao, Runjian Chen, Yixi Cai, Buwei He, Wenquan Zhao, Wenqi Shao, Bo Zhang, Fu Zhang
Abstract: Moving object segmentation (MOS) on LiDAR point clouds is crucial for autonomous systems like self-driving vehicles. Previous supervised approaches rely heavily on costly manual annotations, while LiDAR sequences naturally capture temporal motion cues that can be leveraged for self-supervised learning. In this paper, we propose \textbf{T}emporal \textbf{O}verlapping \textbf{P}rediction (\textbf{TOP}), a self-supervised pre-training method that alleviate the labeling burden for MOS. \textbf{TOP} explores the temporal overlapping points that commonly observed by current and adjacent scans, and learns spatiotemporal representations by predicting the occupancy states of temporal overlapping points. Moreover, we utilize current occupancy reconstruction as an auxiliary pre-training objective, which enhances the current structural awareness of the model. We conduct extensive experiments and observe that the conventional metric Intersection-over-Union (IoU) shows strong bias to objects with more scanned points, which might neglect small or distant objects. To compensate for this bias, we introduce an additional metric called $\text{mIoU}_{\text{obj}}$ to evaluate object-level performance. Experiments on nuScenes and SemanticKITTI show that \textbf{TOP} outperforms both supervised training-from-scratch baseline and other self-supervised pre-training baselines by up to 28.77\% relative improvement, demonstrating strong transferability across LiDAR setups and generalization to other tasks. Code and pre-trained models will be publicly available upon publication.
Authors: Jing Yang, Sen Yang, Xiao Tan, Hanli Wang
Abstract: As an essential component of autonomous driving systems, high-definition (HD) maps provide rich and precise environmental information for auto-driving scenarios; however, existing methods, which primarily rely on query-based detection frameworks to directly model map elements or implicitly propagate queries over time, often struggle to maintain consistent temporal perception outcomes. These inconsistencies pose significant challenges to the stability and reliability of real-world autonomous driving and map data collection systems. To address this limitation, we propose a novel end-to-end tracking framework for global map construction by temporally tracking map elements' historical trajectories. Firstly, instance-level historical rasterization map representation is designed to explicitly store previous perception results, which can control and maintain different global instances' history information in a fine-grained way. Secondly, we introduce a Map-Trajectory Prior Fusion module within this tracking framework, leveraging historical priors for tracked instances to improve temporal smoothness and continuity. Thirdly, we propose a global perspective metric to evaluate the quality of temporal geometry construction in HD maps, filling the gap in current metrics for assessing global geometric perception results. Substantial experiments on the nuScenes and Argoverse2 datasets demonstrate that the proposed method outperforms state-of-the-art (SOTA) methods in both single-frame and temporal metrics. our project page: $\href{https://yj772881654.github.io/HisTrackMap/}{https://yj772881654.github.io/HisTrackMap.}$
URLs: https://yj772881654.github.io/HisTrackMap/, https://yj772881654.github.io/HisTrackMap.
Authors: Kazuya Nishimura, Ryoma Bise, Yasuhiro Kojima
Abstract: Spatial transcriptomics (ST) is a novel technique that simultaneously captures pathological images and gene expression profiling with spatial coordinates. Since ST is closely related to pathological features such as disease subtypes, it may be valuable to augment image representation with pathological information. However, there are no attempts to leverage ST for image recognition ({\it i.e,} patch-level classification of subtypes of pathological image.). One of the big challenges is significant batch effects in spatial transcriptomics that make it difficult to extract pathological features of images from ST. In this paper, we propose a batch-agnostic contrastive learning framework that can extract consistent signals from gene expression of ST in multiple patients. To extract consistent signals from ST, we utilize the batch-agnostic gene encoder that is trained in a variational inference manner. Experiments demonstrated the effectiveness of our framework on a publicly available dataset. Code is publicly available at https://github.com/naivete5656/TPIRBAE
Authors: Vasiliki Sideri-Lampretsa, Daniel Rueckert, Huaqi Qiu
Abstract: Evaluating deformable image registration (DIR) is challenging due to the inherent trade-off between achieving high alignment accuracy and maintaining deformation regularity. In this work, we introduce a novel evaluation scheme based on the alignment-regularity characteristic (ARC) to systematically capture and analyze this trade-off. We first introduce the ARC curves, which describe the performance of a given registration algorithm as a spectrum measured by alignment and regularity metrics. We further adopt a HyperNetwork-based approach that learns to continuously interpolate across the full regularization range, accelerating the construction and improving the sample density of ARC curves. We empirically demonstrate our evaluation scheme using representative learning-based deformable image registration methods with various network architectures and transformation models on two public datasets. We present a range of findings not evident from existing evaluation practices and provide general recommendations for model evaluation and selection using our evaluation scheme. All code relevant is made publicly available.
Authors: Melvin Reka, Tessa Pulli, Markus Vincze
Abstract: 6D object pose estimation for unseen objects is essential in robotics but traditionally relies on trained models that require large datasets, high computational costs, and struggle to generalize. Zero-shot approaches eliminate the need for training but depend on pre-existing 3D object models, which are often impractical to obtain. To address this, we propose a language-guided few-shot 3D reconstruction method, reconstructing a 3D mesh from few input images. In the proposed pipeline, receives a set of input images and a language query. A combination of GroundingDINO and Segment Anything Model outputs segmented masks from which a sparse point cloud is reconstructed with VGGSfM. Subsequently, the mesh is reconstructed with the Gaussian Splatting method SuGAR. In a final cleaning step, artifacts are removed, resulting in the final 3D mesh of the queried object. We evaluate the method in terms of accuracy and quality of the geometry and texture. Furthermore, we study the impact of imaging conditions such as viewing angle, number of input images, and image overlap on 3D object reconstruction quality, efficiency, and computational scalability.
Authors: Zebin You, Jingyang Ou, Xiaolu Zhang, Jun Hu, Jun Zhou, Chongxuan Li
Abstract: Although masked image generation models and masked diffusion models are designed with different motivations and objectives, we observe that they can be unified within a single framework. Building upon this insight, we carefully explore the design space of training and sampling, identifying key factors that contribute to both performance and efficiency. Based on the improvements observed during this exploration, we develop our model, referred to as eMIGM. Empirically, eMIGM demonstrates strong performance on ImageNet generation, as measured by Fr\'echet Inception Distance (FID). In particular, on ImageNet 256x256, with similar number of function evaluations (NFEs) and model parameters, eMIGM outperforms the seminal VAR. Moreover, as NFE and model parameters increase, eMIGM achieves performance comparable to the state-of-the-art continuous diffusion models while requiring less than 40% of the NFE. Additionally, on ImageNet 512x512, with only about 60% of the NFE, eMIGM outperforms the state-of-the-art continuous diffusion models.
Authors: Mona Sheikh Zeinoddin, Mobarakol Islam, Zafer Tandogdu, Greg Shaw, Mathew J. Clarkson, Evangelos Mazomenos, Danail Stoyanov
Abstract: Accurate depth and camera pose estimation is essential for achieving high-quality 3D visualisations in robotic-assisted surgery. Despite recent advancements in foundation model adaptation to monocular depth estimation of endoscopic scenes via self-supervised learning (SSL), no prior work has explored their use for pose estimation. These methods rely on low rank-based adaptation approaches, which constrain model updates to a low-rank space. We propose Endo-FASt3r, the first monocular SSL depth and pose estimation framework that uses foundation models for both tasks. We extend the Reloc3r relative pose estimation foundation model by designing Reloc3rX, introducing modifications necessary for convergence in SSL. We also present DoMoRA, a novel adaptation technique that enables higher-rank updates and faster convergence. Experiments on the SCARED dataset show that Endo-FASt3r achieves a substantial $10\%$ improvement in pose estimation and a $2\%$ improvement in depth estimation over prior work. Similar performance gains on the Hamlyn and StereoMIS datasets reinforce the generalisability of Endo-FASt3r across different datasets.
Authors: Ruochen Pi, Lianlei Shan
Abstract: Collecting and annotating medical images is a time-consuming and resource-intensive task. However, generating synthetic data through models such as Diffusion offers a cost-effective alternative. This paper introduces a new method for the automatic generation of accurate semantic masks from synthetic lung X-ray images based on a stable diffusion model trained on text-image pairs. This method uses cross-attention mapping between text and image to extend text-driven image synthesis to semantic mask generation. It employs text-guided cross-attention information to identify specific areas in an image and combines this with innovative techniques to produce high-resolution, class-differentiated pixel masks. This approach significantly reduces the costs associated with data collection and annotation. The experimental results demonstrate that segmentation models trained on synthetic data generated using the method are comparable to, and in some cases even better than, models trained on real datasets. This shows the effectiveness of the method and its potential to revolutionize medical image analysis.
Authors: Luigi Russo, Antonietta Sorriso, Silvia Liberata Ullo, Paolo Gamba
Abstract: Land Cover (LC) mapping using satellite imagery is critical for environmental monitoring and management. Deep Learning (DL), particularly Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs), have revolutionized this field by enhancing the accuracy of classification tasks. In this work, a novel approach combining a transformer-based Swin-Unet architecture with seasonal synthesized spatio-temporal images has been employed to classify LC types using spatio-temporal features extracted from Sentinel-1 (S1) Synthetic Aperture Radar (SAR) data, organized into seasonal clusters. The study focuses on three distinct regions - Amazonia, Africa, and Siberia - and evaluates the model performance across diverse ecoregions within these areas. By utilizing seasonal feature sequences instead of dense temporal sequences, notable performance improvements have been achieved, especially in regions with temporal data gaps like Siberia, where S1 data distribution is uneven and non-uniform. The results demonstrate the effectiveness and the generalization capabilities of the proposed methodology in achieving high overall accuracy (O.A.) values, even in regions with limited training data.
Authors: Chenglu Pan, Xiaogang Xu, Ganggui Ding, Yunke Zhang, Wenbo Li, Jiarong Xu, Qingbiao Wu
Abstract: Restoring low-resolution text images presents a significant challenge, as it requires maintaining both the fidelity and stylistic realism of the text in restored images. Existing text image restoration methods often fall short in hard situations, as the traditional super-resolution models cannot guarantee clarity, while diffusion-based methods fail to maintain fidelity. In this paper, we introduce a novel framework aimed at improving the generalization ability of diffusion models for text image super-resolution (SR), especially promoting fidelity. First, we propose a progressive data sampling strategy that incorporates diverse image types at different stages of training, stabilizing the convergence and improving the generalization. For the network architecture, we leverage a pre-trained SR prior to provide robust spatial reasoning capabilities, enhancing the model's ability to preserve textual information. Additionally, we employ a cross-attention mechanism to better integrate textual priors. To further reduce errors in textual priors, we utilize confidence scores to dynamically adjust the importance of textual features during training. Extensive experiments on real-world datasets demonstrate that our approach not only produces text images with more realistic visual appearances but also improves the accuracy of text structure.
Authors: Haicheng Liao, Hanlin Kong, Bonan Wang, Chengyue Wang, Wang Ye, Zhengbing He, Chengzhong Xu, Zhenning Li
Abstract: Accurate motion forecasting is crucial for safe autonomous driving (AD). This study proposes CoT-Drive, a novel approach that enhances motion forecasting by leveraging large language models (LLMs) and a chain-of-thought (CoT) prompting method. We introduce a teacher-student knowledge distillation strategy to effectively transfer LLMs' advanced scene understanding capabilities to lightweight language models (LMs), ensuring that CoT-Drive operates in real-time on edge devices while maintaining comprehensive scene understanding and generalization capabilities. By leveraging CoT prompting techniques for LLMs without additional training, CoT-Drive generates semantic annotations that significantly improve the understanding of complex traffic environments, thereby boosting the accuracy and robustness of predictions. Additionally, we present two new scene description datasets, Highway-Text and Urban-Text, designed for fine-tuning lightweight LMs to generate context-specific semantic annotations. Comprehensive evaluations of five real-world datasets demonstrate that CoT-Drive outperforms existing models, highlighting its effectiveness and efficiency in handling complex traffic scenarios. Overall, this study is the first to consider the practical application of LLMs in this field. It pioneers the training and use of a lightweight LLM surrogate for motion forecasting, setting a new benchmark and showcasing the potential of integrating LLMs into AD systems.
Authors: Haowen Bai, Jiangshe Zhang, Zixiang Zhao, Lilun Deng, Yukun Cui, Shuang Xu
Abstract: Multi-exposure image fusion consolidates multiple low dynamic range images of the same scene into a singular high dynamic range image. Retinex theory, which separates image illumination from scene reflectance, is naturally adopted to ensure consistent scene representation and effective information fusion across varied exposure levels. However, the conventional pixel-wise multiplication of illumination and reflectance inadequately models the glare effect induced by overexposure. To better adapt this theory for multi-exposure image fusion, we introduce an unsupervised and controllable method termed~\textbf{(Retinex-MEF)}. Specifically, our method decomposes multi-exposure images into separate illumination components and a shared reflectance component, and effectively modeling the glare induced by overexposure. Employing a bidirectional loss constraint to learn the common reflectance component, our approach effectively mitigates the glare effect. Furthermore, we establish a controllable exposure fusion criterion, enabling global exposure adjustments while preserving contrast, thus overcoming the constraints of fixed-level fusion. A series of experiments across multiple datasets, including underexposure-overexposure fusion, exposure control fusion, and homogeneous extreme exposure fusion, demonstrate the effective decomposition and flexible fusion capability of our model.
Authors: Feng Huang, Shuyuan Zheng, Zhaobing Qiu, Huanxian Liu, Huanxin Bai, Liqiong Chen
Abstract: Infrared small target detection is currently a hot and challenging task in computer vision. Existing methods usually focus on mining visual features of targets, which struggles to cope with complex and diverse detection scenarios. The main reason is that infrared small targets have limited image information on their own, thus relying only on visual features fails to discriminate targets and interferences, leading to lower detection performance. To address this issue, we introduce a novel approach leveraging semantic text to guide infrared small target detection, called Text-IRSTD. It innovatively expands classical IRSTD to text-guided IRSTD, providing a new research idea. On the one hand, we devise a novel fuzzy semantic text prompt to accommodate ambiguous target categories. On the other hand, we propose a progressive cross-modal semantic interaction decoder (PCSID) to facilitate information fusion between texts and images. In addition, we construct a new benchmark consisting of 2,755 infrared images of different scenarios with fuzzy semantic textual annotations, called FZDT. Extensive experimental results demonstrate that our method achieves better detection performance and target contour recovery than the state-of-the-art methods. Moreover, proposed Text-IRSTD shows strong generalization and wide application prospects in unseen detection scenarios. The dataset and code will be publicly released after acceptance of this paper.
Authors: Yubo Peng, Luping Xiang, Kun Yang, Kezhi Wang, Merouane Debbah
Abstract: Despite the widespread adoption of vision sensors in edge applications, such as surveillance, the transmission of video data consumes substantial spectrum resources. Semantic communication (SC) offers a solution by extracting and compressing information at the semantic level, preserving the accuracy and relevance of transmitted data while significantly reducing the volume of transmitted information. However, traditional SC methods face inefficiencies due to the repeated transmission of static frames in edge videos, exacerbated by the absence of sensing capabilities, which results in spectrum inefficiency. To address this challenge, we propose a SC with computer vision sensing (SCCVS) framework for edge video transmission. The framework first introduces a compression ratio (CR) adaptive SC (CRSC) model, capable of adjusting CR based on whether the frames are static or dynamic, effectively conserving spectrum resources. Additionally, we implement an object detection and semantic segmentation models-enabled sensing (OSMS) scheme, which intelligently senses the changes in the scene and assesses the significance of each frame through in-context analysis. Hence, The OSMS scheme provides CR prompts to the CRSC model based on real-time sensing results. Moreover, both CRSC and OSMS are designed as lightweight models, ensuring compatibility with resource-constrained sensors commonly used in practical edge applications. Experimental simulations validate the effectiveness of the proposed SCCVS framework, demonstrating its ability to enhance transmission efficiency without sacrificing critical semantic information.
Authors: Zhangyu Lai, Yilin Lu, Xinyang Li, Jianghang Lin, Yansong Qu, Liujuan Cao, Ming Li, Rongrong Ji
Abstract: While existing anomaly synthesis methods have made remarkable progress, achieving both realism and diversity in synthesis remains a major obstacle. To address this, we propose AnomalyPainter, a zero-shot framework that breaks the diversity-realism trade-off dilemma through synergizing Vision Language Large Model (VLLM), Latent Diffusion Model (LDM), and our newly introduced texture library Tex-9K. Tex-9K is a professional texture library containing 75 categories and 8,792 texture assets crafted for diverse anomaly synthesis. Leveraging VLLM's general knowledge, reasonable anomaly text descriptions are generated for each industrial object and matched with relevant diverse textures from Tex-9K. These textures then guide the LDM via ControlNet to paint on normal images. Furthermore, we introduce Texture-Aware Latent Init to stabilize the natural-image-trained ControlNet for industrial images. Extensive experiments show that AnomalyPainter outperforms existing methods in realism, diversity, and generalization, achieving superior downstream performance.
Authors: Baiyu Chen, Wilson Wongso, Zechen Li, Yonchanok Khaokaew, Hao Xue, Flora Salim
Abstract: Egocentric video-based models capture rich semantic information and have demonstrated strong performance in human activity recognition (HAR). However, their high power consumption, privacy concerns, and dependence on lighting conditions limit their feasibility for continuous on-device recognition. In contrast, inertial measurement unit (IMU) sensors offer an energy-efficient and privacy-preserving alternative, yet they suffer from limited large-scale annotated datasets, leading to weaker generalization in downstream tasks. To bridge this gap, we propose COMODO, a cross-modal self-supervised distillation framework that transfers rich semantic knowledge from the video modality to the IMU modality without requiring labeled annotations. COMODO leverages a pretrained and frozen video encoder to construct a dynamic instance queue, aligning the feature distributions of video and IMU embeddings. By distilling knowledge from video representations, our approach enables the IMU encoder to inherit rich semantic information from video while preserving its efficiency for real-world applications. Experiments on multiple egocentric HAR datasets demonstrate that COMODO consistently improves downstream classification performance, achieving results comparable to or exceeding fully supervised fine-tuned models. Moreover, COMODO exhibits strong cross-dataset generalization. Benefiting from its simplicity, our method is also generally applicable to various video and time-series pre-trained models, offering the potential to leverage more powerful teacher and student foundation models in future research. The code is available at https://github.com/Breezelled/COMODO .
Authors: Yuwei Niu, Munan Ning, Mengren Zheng, Bin Lin, Peng Jin, Jiaqi Liao, Kunpeng Ning, Bin Zhu, Li Yuan
Abstract: Text-to-Image (T2I) models are capable of generating high-quality artistic creations and visual content. However, existing research and evaluation standards predominantly focus on image realism and shallow text-image alignment, lacking a comprehensive assessment of complex semantic understanding and world knowledge integration in text to image generation. To address this challenge, we propose $\textbf{WISE}$, the first benchmark specifically designed for $\textbf{W}$orld Knowledge-$\textbf{I}$nformed $\textbf{S}$emantic $\textbf{E}$valuation. WISE moves beyond simple word-pixel mapping by challenging models with 1000 meticulously crafted prompts across 25 sub-domains in cultural common sense, spatio-temporal reasoning, and natural science. To overcome the limitations of traditional CLIP metric, we introduce $\textbf{WiScore}$, a novel quantitative metric for assessing knowledge-image alignment. Through comprehensive testing of 20 models (10 dedicated T2I models and 10 unified multimodal models) using 1,000 structured prompts spanning 25 subdomains, our findings reveal significant limitations in their ability to effectively integrate and apply world knowledge during image generation, highlighting critical pathways for enhancing knowledge incorporation and application in next-generation T2I models. Code and data are available at https://github.com/PKU-YuanGroup/WISE.
Authors: Fu Rong, Meng Lan, Qian Zhang, Lefei Zhang
Abstract: Referring Remote Sensing Image Segmentation (RRSIS) aims to segment target objects in remote sensing (RS) images based on textual descriptions. Although Segment Anything Model 2 (SAM 2) has shown remarkable performance in various segmentation tasks, its application to RRSIS presents several challenges, including understanding the text-described RS scenes and generating effective prompts from text descriptions. To address these issues, we propose RS2-SAM 2, a novel framework that adapts SAM 2 to RRSIS by aligning the adapted RS features and textual features, providing pseudo-mask-based dense prompts, and enforcing boundary constraints. Specifically, we first employ a union encoder to jointly encode the visual and textual inputs, generating aligned visual and text embeddings as well as multimodal class tokens. Then, we design a bidirectional hierarchical fusion module to adapt SAM 2 to RS scenes and align adapted visual features with the visually enhanced text embeddings, improving the model's interpretation of text-described RS scenes. Additionally, a mask prompt generator is introduced to take the visual embeddings and class tokens as input and produce a pseudo-mask as the dense prompt of SAM 2. To further refine segmentation, we introduce a text-guided boundary loss to optimize segmentation boundaries by computing text-weighted gradient differences. Experimental results on several RRSIS benchmarks demonstrate that RS2-SAM 2 achieves state-of-the-art performance.
Authors: Guilherme Silva, Pedro Silva, Gladston Moreira, Vander Freitas, Jadson Gertrudes, Eduardo Luz
Abstract: The classification of electrocardiogram (ECG) signals is crucial for early detection of arrhythmias and other cardiac conditions. However, despite advances in machine learning, many studies fail to follow standardization protocols, leading to inconsistencies in performance evaluation and real-world applicability. Additionally, hardware constraints essential for practical deployment, such as in pacemakers, Holter monitors, and wearable ECG patches, are often overlooked. Since real-world impact depends on feasibility in resource-constrained devices, ensuring efficient deployment is critical for continuous monitoring. This review systematically analyzes ECG classification studies published between 2017 and 2024, focusing on those adhering to the E3C (Embedded, Clinical, and Comparative Criteria), which include inter-patient paradigm implementation, compliance with Association for the Advancement of Medical Instrumentation (AAMI) recommendations, and model feasibility for embedded systems. While many studies report high accuracy, few properly consider patient-independent partitioning and hardware limitations. We identify state-of-the-art methods meeting E3C criteria and conduct a comparative analysis of accuracy, inference time, energy consumption, and memory usage. Finally, we propose standardized reporting practices to ensure fair comparisons and practical applicability of ECG classification models. By addressing these gaps, this study aims to guide future research toward more robust and clinically viable ECG classification systems.
Authors: Thomas Boucher, Evangelos B. Mazomenos
Abstract: Quantum vision transformers (QViTs) build on vision transformers (ViTs) by replacing linear layers within the self-attention mechanism with parameterised quantum neural networks (QNNs), harnessing quantum mechanical properties to improve feature representation. This hybrid approach aims to achieve superior performance, with significantly reduced model complexity as a result of the enriched feature representation, requiring fewer parameters. This paper proposes a novel QViT model for biomedical image classification and investigates its performance against comparable ViTs across eight diverse datasets, encompassing various modalities and classification tasks. We assess models trained from scratch and those pre-trained using knowledge distillation (KD) from high-quality teacher models. Our findings demonstrate that QViTs outperform comparable ViTs with average ROC AUC (0.863 vs 0.846) and accuracy (0.710 vs 0.687) when trained from scratch, and even compete with state-of-the-art classical models in multiple tasks, whilst being significantly more efficient (89% reduction in GFLOPs and 99.99% in parameter number). Additionally, we find that QViTs and ViTs respond equally well to KD, with QViT pre-training performance scaling with model complexity. This is the first investigation into the efficacy of deploying QViTs with KD for computer-aided diagnosis. Our results highlight the enormous potential of quantum machine learning (QML) in biomedical image analysis.
Authors: Xichen Tan, Yuanjing Luo, Yunfan Ye, Fang Liu, Zhiping Cai
Abstract: From image to video understanding, the capabilities of Multi-modal LLMs (MLLMs) are increasingly powerful. However, most existing video understanding benchmarks are relatively short, which makes them inadequate for effectively evaluating the long-sequence modeling capabilities of MLLMs. This highlights the urgent need for a comprehensive and integrated long video understanding benchmark to assess the ability of MLLMs thoroughly. To this end, we propose ALLVB (ALL-in-One Long Video Understanding Benchmark). ALLVB's main contributions include: 1) It integrates 9 major video understanding tasks. These tasks are converted into video QA formats, allowing a single benchmark to evaluate 9 different video understanding capabilities of MLLMs, highlighting the versatility, comprehensiveness, and challenging nature of ALLVB. 2) A fully automated annotation pipeline using GPT-4o is designed, requiring only human quality control, which facilitates the maintenance and expansion of the benchmark. 3) It contains 1,376 videos across 16 categories, averaging nearly 2 hours each, with a total of 252k QAs. To the best of our knowledge, it is the largest long video understanding benchmark in terms of the number of videos, average duration, and number of QAs. We have tested various mainstream MLLMs on ALLVB, and the results indicate that even the most advanced commercial models have significant room for improvement. This reflects the benchmark's challenging nature and demonstrates the substantial potential for development in long video understanding.
Authors: Bo Huang, Wenlun Xu, Qizhuo Han, Haodong Jing, Ying Li
Abstract: While diffusion models have achieved remarkable progress in style transfer tasks, existing methods typically rely on fine-tuning or optimizing pre-trained models during inference, leading to high computational costs and challenges in balancing content preservation with style integration. To address these limitations, we introduce AttenST, a training-free attention-driven style transfer framework. Specifically, we propose a style-guided self-attention mechanism that conditions self-attention on the reference style by retaining the query of the content image while substituting its key and value with those from the style image, enabling effective style feature integration. To mitigate style information loss during inversion, we introduce a style-preserving inversion strategy that refines inversion accuracy through multiple resampling steps. Additionally, we propose a content-aware adaptive instance normalization, which integrates content statistics into the normalization process to optimize style fusion while mitigating the content degradation. Furthermore, we introduce a dual-feature cross-attention mechanism to fuse content and style features, ensuring a harmonious synthesis of structural fidelity and stylistic expression. Extensive experiments demonstrate that AttenST outperforms existing methods, achieving state-of-the-art performance in style transfer dataset.
Authors: Weijia Wu, Zeyu Zhu, Mike Zheng Shou
Abstract: Existing long-form video generation frameworks lack automated planning, requiring manual input for storylines, scenes, cinematography, and character interactions, resulting in high costs and inefficiencies. To address these challenges, we present MovieAgent, an automated movie generation via multi-agent Chain of Thought (CoT) planning. MovieAgent offers two key advantages: 1) We firstly explore and define the paradigm of automated movie/long-video generation. Given a script and character bank, our MovieAgent can generates multi-scene, multi-shot long-form videos with a coherent narrative, while ensuring character consistency, synchronized subtitles, and stable audio throughout the film. 2) MovieAgent introduces a hierarchical CoT-based reasoning process to automatically structure scenes, camera settings, and cinematography, significantly reducing human effort. By employing multiple LLM agents to simulate the roles of a director, screenwriter, storyboard artist, and location manager, MovieAgent streamlines the production pipeline. Experiments demonstrate that MovieAgent achieves new state-of-the-art results in script faithfulness, character consistency, and narrative coherence. Our hierarchical framework takes a step forward and provides new insights into fully automated movie generation. The code and project website are available at: https://github.com/showlab/MovieAgent and https://weijiawu.github.io/MovieAgent.
URLs: https://github.com/showlab/MovieAgent, https://weijiawu.github.io/MovieAgent.
Authors: Weicheng He, Changshun Wu, Chih-Hong Cheng, Xiaowei Huang, Saddek Bensalem
Abstract: Object detection systems must reliably perceive objects of interest without being overly confident to ensure safe decision-making in dynamic environments. Filtering techniques based on out-of-distribution (OoD) detection are commonly added as an extra safeguard to filter hallucinations caused by overconfidence in novel objects. Nevertheless, evaluating YOLO-family detectors and their filters under existing OoD benchmarks often leads to unsatisfactory performance. This paper studies the underlying reasons for performance bottlenecks and proposes a methodology to improve performance fundamentally. Our first contribution is a calibration of all existing evaluation results: Although images in existing OoD benchmark datasets are claimed not to have objects within in-distribution (ID) classes (i.e., categories defined in the training dataset), around 13% of objects detected by the object detector are actually ID objects. Dually, the ID dataset containing OoD objects can also negatively impact the decision boundary of filters. These ultimately lead to a significantly imprecise performance estimation. Our second contribution is to consider the task of hallucination reduction as a joint pipeline of detectors and filters. By developing a methodology to carefully synthesize an OoD dataset that semantically resembles the objects to be detected, and using the crafted OoD dataset in the fine-tuning of YOLO detectors to suppress the objectness score, we achieve a 88% reduction in overall hallucination error with a combined fine-tuned detection and filtering system on the self-driving benchmark BDD-100K. Our code and dataset are available at: https://gricad-gitlab.univ-grenoble-alpes.fr/dnn-safety/m-hood.
URLs: https://gricad-gitlab.univ-grenoble-alpes.fr/dnn-safety/m-hood.
Authors: Xing Xie, Jiawei Liu, Ziyue Lin, Huijie Fan, Zhi Han, Yandong Tang, Liangqiong Qu
Abstract: We present Autoregressive Representation Alignment (ARRA), a new training framework that unlocks global-coherent text-to-image generation in autoregressive LLMs without architectural changes. Unlike prior work that requires complex architectural redesigns, ARRA aligns LLM hidden states with visual representations from external visual foundational models via a global visual alignment loss and a hybrid token,
Authors: Nils Philipp Walter, Jilles Vreeken, Jonas Fischer
Abstract: Neural networks are part of daily-life decision-making, including in high-stakes settings where understanding and transparency are key. Saliency maps have been developed to gain understanding into which input features neural networks use for a specific prediction. Although widely employed, these methods often result in overly general saliency maps that fail to identify the specific information that triggered the classification. In this work, we suggest a framework that allows to incorporate attributions across classes to arrive at saliency maps that actually capture the class-relevant information. On established benchmarks for attribution methods, including the grid-pointing game and randomization-based sanity checks, we show that our framework heavily boosts the performance of standard saliency map approaches. It is, by design, agnostic to model architectures and attribution methods and now allows to identify the distinguishing and shared features used for a model prediction.
Authors: Johan Edstedt, Georg B\"okman, M{\aa}rten Wadenb\"ack, Michael Felsberg
Abstract: Keypoints are what enable Structure-from-Motion (SfM) systems to scale to thousands of images. However, designing a keypoint detection objective is a non-trivial task, as SfM is non-differentiable. Typically, an auxiliary objective involving a descriptor is optimized. This however induces a dependency on the descriptor, which is undesirable. In this paper we propose a fully self-supervised and descriptor-free objective for keypoint detection, through reinforcement learning. To ensure training does not degenerate, we leverage a balanced top-K sampling strategy. While this already produces competitive models, we find that two qualitatively different types of detectors emerge, which are only able to detect light and dark keypoints respectively. To remedy this, we train a third detector, DaD, that optimizes the Kullback-Leibler divergence of the pointwise maximum of both light and dark detectors. Our approach significantly improve upon SotA across a range of benchmarks. Code and model weights are publicly available at https:github.com/parskatt/dad
Authors: Christoph Karg, Sebastian Stricker, Lisa Hutschenreiter, Bogdan Savchynskyy, Dagmar Kainmueller
Abstract: In this work we present a novel approach for unsupervised multi-graph matching, which applies to problems for which a Gaussian distribution of keypoint features can be assumed. We leverage cycle consistency as loss for self-supervised learning, and determine Gaussian parameters through Bayesian Optimization, yielding a highly efficient approach that scales to large datasets. Our fully unsupervised approach enables us to reach the accuracy of state-of-the-art supervised methodology for the use case of annotating cell nuclei in 3D microscopy images of the worm C. elegans. To this end, our approach yields the first unsupervised atlas of C. elegans, i.e. a model of the joint distribution of all of its cell nuclei, without the need for any ground truth cell annotation. This advancement enables highly efficient annotation of cell nuclei in large microscopy datasets of C. elegans. Beyond C. elegans, our approach offers fully unsupervised construction of cell-level atlases for any model organism with a stereotyped cell lineage, and thus bears the potential to catalyze respective comparative developmental studies in a range of further species.
Authors: Carl Olsson, Yaroslava Lochman, Johan Malmport, Christopher Zach
Abstract: Rotation averaging is a key subproblem in applications of computer vision and robotics. Many methods for solving this problem exist, and there are also several theoretical results analyzing difficulty and optimality. However, one aspect that most of these have in common is a focus on the isotropic setting, where the intrinsic uncertainties in the measurements are not fully incorporated into the resulting optimization task. Recent empirical results suggest that moving to an anisotropic framework, where these uncertainties are explicitly included, can result in an improvement of solution quality. However, global optimization for rotation averaging has remained a challenge in this scenario. In this paper we show how anisotropic costs can be incorporated in certifiably optimal rotation averaging. We also demonstrate how existing solvers, designed for isotropic situations, fail in the anisotropic setting. Finally, we propose a stronger relaxation and show empirically that it is able to recover global optima in all tested datasets and leads to a more accurate reconstruction in all but one of the scenes.
Authors: Jiang Lin, Zili Yi
Abstract: Video style transfer aims to alter the style of a video while preserving its content. Previous methods often struggle with content leakage and style misalignment, particularly when using image-driven approaches that aim to transfer precise styles. In this work, we introduce Trajectory Reset Attention Control (TRAC), a novel method that allows for high-quality style transfer while preserving content integrity. TRAC operates by resetting the denoising trajectory and enforcing attention control, thus enhancing content consistency while significantly reducing the computational costs against inversion-based methods. Additionally, a concept termed Style Medium is introduced to bridge the gap between content and style, enabling a more precise and harmonious transfer of stylistic elements. Building upon these concepts, we present a tuning-free framework that offers a stable, flexible, and efficient solution for both image and video style transfer. Experimental results demonstrate that our proposed framework accommodates a wide range of stylized outputs, from precise content preservation to the production of visually striking results with vibrant and expressive styles.
Authors: Fanqing Meng, Lingxiao Du, Zongkai Liu, Zhixiang Zhou, Quanfeng Lu, Daocheng Fu, Botian Shi, Wenhai Wang, Junjun He, Kaipeng Zhang, Ping Luo, Yu Qiao, Qiaosheng Zhang, Wenqi Shao
Abstract: We present MM-Eureka, a multimodal reasoning model that successfully extends large-scale rule-based reinforcement learning (RL) to multimodal reasoning. While rule-based RL has shown remarkable success in improving LLMs' reasoning abilities in text domains, its application to multimodal settings has remained challenging. Our work reproduces key characteristics of text-based RL systems like DeepSeek-R1 in the multimodal space, including steady increases in accuracy reward and response length, and the emergence of reflection behaviors. We demonstrate that both instruction-tuned and pre-trained models can develop strong multimodal reasoning capabilities through rule-based RL without supervised fine-tuning, showing superior data efficiency compared to alternative approaches. We open-source our complete pipeline to foster further research in this area. We release all our codes, models, data, etc. at https://github.com/ModalMinds/MM-EUREKA
Authors: Kangan Qian, Jinyu Miao, Ziang Luo, Zheng Fu, and Jinchen Li, Yining Shi, Yunlong Wang, Kun Jiang, Mengmeng Yang, Diange Yang
Abstract: Accurate and reliable spatial and motion information plays a pivotal role in autonomous driving systems. However, object-level perception models struggle with handling open scenario categories and lack precise intrinsic geometry. On the other hand, occupancy-based class-agnostic methods excel in representing scenes but fail to ensure physics consistency and ignore the importance of interactions between traffic participants, hindering the model's ability to learn accurate and reliable motion. In this paper, we introduce a novel occupancy-instance modeling framework for class-agnostic motion prediction tasks, named LEGO-Motion, which incorporates instance features into Bird's Eye View (BEV) space. Our model comprises (1) a BEV encoder, (2) an Interaction-Augmented Instance Encoder, and (3) an Instance-Enhanced BEV Encoder, improving both interaction relationships and physics consistency within the model, thereby ensuring a more accurate and robust understanding of the environment. Extensive experiments on the nuScenes dataset demonstrate that our method achieves state-of-the-art performance, outperforming existing approaches. Furthermore, the effectiveness of our framework is validated on the advanced FMCW LiDAR benchmark, showcasing its practical applicability and generalization capabilities. The code will be made publicly available to facilitate further research.
Authors: Qizhi Zheng, Zhongze Luo, Meiyan Guo, Xinzhu Wang, Renqimuge Wu, Qiu Meng, Guanghui Dong
Abstract: Accurate and real-time object detection is crucial for anomaly behavior detection, especially in scenarios constrained by hardware limitations, where balancing accuracy and speed is essential for enhancing detection performance. This study proposes a model called HGO-YOLO, which integrates the HGNetv2 architecture into YOLOv8. This combination expands the receptive field and captures a wider range of features while simplifying model complexity through GhostConv. We introduced a lightweight detection head, OptiConvDetect, which utilizes parameter sharing to construct the detection head effectively. Evaluation results show that the proposed algorithm achieves a mAP@0.5 of 87.4% and a recall rate of 81.1%, with a model size of only 4.6 MB and a frame rate of 56 FPS on the CPU. HGO-YOLO not only improves accuracy by 3.0% but also reduces computational load by 51.69% (from 8.9 GFLOPs to 4.3 GFLOPs), while increasing the frame rate by a factor of 1.7. Additionally, real-time tests were conducted on Raspberry Pi4 and NVIDIA platforms. These results indicate that the HGO-YOLO model demonstrates superior performance in anomaly behavior detection.
Authors: R. Spencer Hallyburton, David Hunt, Yiwei He, Judy He, Miroslav Pajic
Abstract: Attacks on sensing and perception threaten the safe deployment of autonomous vehicles (AVs). Security-aware sensor fusion helps mitigate threats but requires accurate field of view (FOV) estimation which has not been evaluated autonomy. To address this gap, we adapt classical computer graphics algorithms to develop the first autonomy-relevant FOV estimators and create the first datasets with ground truth FOV labels. Unfortunately, we find that these approaches are themselves highly vulnerable to attacks on sensing. To improve robustness of FOV estimation against attacks, we propose a learning-based segmentation model that captures FOV features, integrates Monte Carlo dropout (MCD) for uncertainty quantification, and performs anomaly detection on confidence maps. We illustrate through comprehensive evaluations attack resistance and strong generalization across environments. Architecture trade studies demonstrate the model is feasible for real-time deployment in multiple applications.
Authors: Ruidong Chen, Honglin Guo, Lanjun Wang, Chenyu Zhang, Weizhi Nie, An-An Liu
Abstract: Recent advances in text-to-image diffusion models enable photorealistic image generation, but they also risk producing malicious content, such as NSFW images. To mitigate risk, concept erasure methods are studied to facilitate the model to unlearn specific concepts. However, current studies struggle to fully erase malicious concepts implicitly embedded in prompts (e.g., metaphorical expressions or adversarial prompts) while preserving the model's normal generation capability. To address this challenge, our study proposes TRCE, using a two-stage concept erasure strategy to achieve an effective trade-off between reliable erasure and knowledge preservation. Firstly, TRCE starts by erasing the malicious semantics implicitly embedded in textual prompts. By identifying a critical mapping objective(i.e., the [EoT] embedding), we optimize the cross-attention layers to map malicious prompts to contextually similar prompts but with safe concepts. This step prevents the model from being overly influenced by malicious semantics during the denoising process. Following this, considering the deterministic properties of the sampling trajectory of the diffusion model, TRCE further steers the early denoising prediction toward the safe direction and away from the unsafe one through contrastive learning, thus further avoiding the generation of malicious content. Finally, we conduct comprehensive evaluations of TRCE on multiple malicious concept erasure benchmarks, and the results demonstrate its effectiveness in erasing malicious concepts while better preserving the model's original generation ability. The code is available at: http://github.com/ddgoodgood/TRCE. CAUTION: This paper includes model-generated content that may contain offensive material.
Authors: Boeun Kim, Hea In Jeong, JungHoon Sung, Yihua Cheng, Jeongmin Lee, Ju Yong Chang, Sang-Il Choi, Younggeun Choi, Saim Shin, Jungho Kim, Hyung Jin Chang
Abstract: This paper introduces Motion Personalization, a new task that generates personalized motions aligned with text descriptions using several basic motions containing Persona. To support this novel task, we introduce a new large-scale motion dataset called PerMo (PersonaMotion), which captures the unique personas of multiple actors. We also propose a multi-modal finetuning method of a pretrained motion diffusion model called PersonaBooth. PersonaBooth addresses two main challenges: i) A significant distribution gap between the persona-focused PerMo dataset and the pretraining datasets, which lack persona-specific data, and ii) the difficulty of capturing a consistent persona from the motions vary in content (action type). To tackle the dataset distribution gap, we introduce a persona token to accept new persona features and perform multi-modal adaptation for both text and visuals during finetuning. To capture a consistent persona, we incorporate a contrastive learning technique to enhance intra-cohesion among samples with the same persona. Furthermore, we introduce a context-aware fusion mechanism to maximize the integration of persona cues from multiple input motions. PersonaBooth outperforms state-of-the-art motion style transfer methods, establishing a new benchmark for motion personalization.
Authors: Ouxiang Li, Yuan Wang, Xinting Hu, Houcheng Jiang, Tao Liang, Yanbin Hao, Guojun Ma, Fuli Feng
Abstract: Erasing concepts from large-scale text-to-image (T2I) diffusion models has become increasingly crucial due to the growing concerns over copyright infringement, offensive content, and privacy violations. However, existing methods either require costly fine-tuning or degrade image quality for non-target concepts (i.e., prior) due to inherent optimization limitations. In this paper, we introduce SPEED, a model editing-based concept erasure approach that leverages null-space constraints for scalable, precise, and efficient erasure. Specifically, SPEED incorporates Influence-based Prior Filtering (IPF) to retain the most affected non-target concepts during erasing, Directed Prior Augmentation (DPA) to expand prior coverage while maintaining semantic consistency, and Invariant Equality Constraints (IEC) to regularize model editing by explicitly preserving key invariants during the T2I generation process. Extensive evaluations across multiple concept erasure tasks demonstrate that SPEED consistently outperforms existing methods in prior preservation while achieving efficient and high-fidelity concept erasure, successfully removing 100 concepts within just 5 seconds. Our code and models are available at: https://github.com/Ouxiang-Li/SPEED.
Authors: Kexin Di, Xiuxing Li, Yuyang Han, Ziyu Li, Qing Li, Xia Wu
Abstract: Few-shot image classification has become a popular research topic for its wide application in real-world scenarios, however the problem of supervision collapse induced by single image-level annotation remains a major challenge. Existing methods aim to tackle this problem by locating and aligning relevant local features. However, the high intra-class variability in real-world images poses significant challenges in locating semantically relevant local regions under few-shot settings. Drawing inspiration from the human's complementary learning system, which excels at rapidly capturing and integrating semantic features from limited examples, we propose the generalization-optimized Systems Consolidation Adaptive Memory Dual-Network, SCAM-Net. This approach simulates the systems consolidation of complementary learning system with an adaptive memory module, which successfully addresses the difficulty of identifying meaningful features in few-shot scenarios. Specifically, we construct a Hippocampus-Neocortex dual-network that consolidates structured representation of each category, the structured representation is then stored and adaptively regulated following the generalization optimization principle in a long-term memory inside Neocortex. Extensive experiments on benchmark datasets show that the proposed model has achieved state-of-the-art performance.
Authors: Wenqiang Zu, Shenghao Xie, Hao Chen, Yiming Liang, Lei Ma
Abstract: Foundation models pretrained on large-scale natural images have been widely used to adapt to medical image analysis through finetuning. This is largely attributed to pretrained representations capturing universal, robust, and generalizable features, which can be reutilized by downstream tasks. However, these representations are later found to gradually vanish during finetuning, accompanied by a degradation of foundation model's original abilities, e.g., generalizability. In this paper, we argue that pretrained representations can be well preserved while still effectively adapting to downstream tasks. We study this by proposing a new finetuning method RepSim, which minimizes the distance between pretrained and finetuned representations via constraining learnable orthogonal manifold based on similarity invariance. Compared to standard finetuning methods, e.g., full finetuning, our method improves representation similarity by over 30% while maintaining competitive accuracy, and reduces sharpness by 42% across five medical image classification datasets. The code will be released.
Authors: Yan Tai, Luhao Zhu, Zhiqiang Chen, Ynan Ding, Yiying Dong, Xiaohong Liu, Guodong Guo
Abstract: Multimodal Large Language Models (MLLMs) demonstrate robust zero-shot capabilities across diverse vision-language tasks after training on mega-scale datasets. However, dense prediction tasks, such as semantic segmentation and keypoint detection, pose significant challenges for MLLMs when represented solely as text outputs. Simultaneously, current MLLMs utilizing latent embeddings for visual task decoding generally demonstrate limited adaptability to both multi-task learning and multi-granularity scenarios. In this work, we present REF-VLM, an end-to-end framework for unified training of various visual decoding tasks. To address complex visual decoding scenarios, we introduce the Triplet-Based Referring Paradigm (TRP), which explicitly decouples three critical dimensions in visual decoding tasks through a triplet structure: concepts, decoding types, and targets. TRP employs symbolic delimiters to enforce structured representation learning, enhancing the parsability and interpretability of model outputs. Additionally, we construct Visual-Task Instruction Following Dataset (VTInstruct), a large-scale multi-task dataset containing over 100 million multimodal dialogue samples across 25 task types. Beyond text inputs and outputs, VT-Instruct incorporates various visual prompts such as point, box, scribble, and mask, and generates outputs composed of text and visual units like box, keypoint, depth and mask. The combination of different visual prompts and visual units generates a wide variety of task types, expanding the applicability of REF-VLM significantly. Both qualitative and quantitative experiments demonstrate that our REF-VLM outperforms other MLLMs across a variety of standard benchmarks. The code, dataset, and demo available at https://github.com/MacavityT/REF-VLM.
Authors: Shaobin Zhuang, Yiwei Guo, Yanbo Ding, Kunchang Li, Xinyuan Chen, Yaohui Wang, Fangyikang Wang, Ying Zhang, Chen Li, Yali Wang
Abstract: Diffusion models have driven the advancement of vision generation over the past years. However, it is often difficult to apply these large models in downstream tasks, due to massive fine-tuning cost. Recently, Low-Rank Adaptation (LoRA) has been applied for efficient tuning of diffusion models. Unfortunately, the capabilities of LoRA-tuned diffusion models are limited, since the same LoRA is used for different timesteps of the diffusion process. To tackle this problem, we introduce a general and concise TimeStep Master (TSM) paradigm with two key fine-tuning stages. In the fostering stage (1-stage), we apply different LoRAs to fine-tune the diffusion model at different timestep intervals. This results in different TimeStep LoRA experts that can effectively capture different noise levels. In the assembling stage (2-stage), we design a novel asymmetrical mixture of TimeStep LoRA experts, via core-context collaboration of experts at multi-scale intervals. For each timestep, we leverage TimeStep LoRA expert within the smallest interval as the core expert without gating, and use experts within the bigger intervals as the context experts with time-dependent gating. Consequently, our TSM can effectively model the noise level via the expert in the finest interval, and adaptively integrate contexts from the experts of other scales, boosting the versatility of diffusion models. To show the effectiveness of our TSM paradigm, we conduct extensive experiments on three typical and popular LoRA-related tasks of diffusion models, including domain adaptation, post-pretraining, and model distillation. Our TSM achieves the state-of-the-art results on all these tasks, throughout various model structures (UNet, DiT and MM-DiT) and visual data modalities (Image, Video), showing its remarkable generalization capacity.
Authors: Minwen Liao, Hao Bo Dong, Xinyi Wang, Ziyang Yan, Yihua Shao
Abstract: Low-light enhancement has wide applications in autonomous driving, 3D reconstruction, remote sensing, surveillance, and so on, which can significantly improve information utilization. However, most existing methods lack generalization and are limited to specific tasks such as image recovery. To address these issues, we propose \textbf{Gated-Mechanism Mixture-of-Experts (GM-MoE)}, the first framework to introduce a mixture-of-experts network for low-light image enhancement. GM-MoE comprises a dynamic gated weight conditioning network and three sub-expert networks, each specializing in a distinct enhancement task. Combining a self-designed gated mechanism that dynamically adjusts the weights of the sub-expert networks for different data domains. Additionally, we integrate local and global feature fusion within sub-expert networks to enhance image quality by capturing multi-scale features. Experimental results demonstrate that the GM-MoE achieves superior generalization with respect to 25 compared approaches, reaching state-of-the-art performance on PSNR on 5 benchmarks and SSIM on 4 benchmarks, respectively.
Authors: Mingzhen Sun, Weining Wang, Gen Li, Jiawei Liu, Jiahui Sun, Wanquan Feng, Shanshan Lao, SiYu Zhou, Qian He, Jing Liu
Abstract: The task of video generation requires synthesizing visually realistic and temporally coherent video frames. Existing methods primarily use asynchronous auto-regressive models or synchronous diffusion models to address this challenge. However, asynchronous auto-regressive models often suffer from inconsistencies between training and inference, leading to issues such as error accumulation, while synchronous diffusion models are limited by their reliance on rigid sequence length. To address these issues, we introduce Auto-Regressive Diffusion (AR-Diffusion), a novel model that combines the strengths of auto-regressive and diffusion models for flexible, asynchronous video generation. Specifically, our approach leverages diffusion to gradually corrupt video frames in both training and inference, reducing the discrepancy between these phases. Inspired by auto-regressive generation, we incorporate a non-decreasing constraint on the corruption timesteps of individual frames, ensuring that earlier frames remain clearer than subsequent ones. This setup, together with temporal causal attention, enables flexible generation of videos with varying lengths while preserving temporal coherence. In addition, we design two specialized timestep schedulers: the FoPP scheduler for balanced timestep sampling during training, and the AD scheduler for flexible timestep differences during inference, supporting both synchronous and asynchronous generation. Extensive experiments demonstrate the superiority of our proposed method, which achieves competitive and state-of-the-art results across four challenging benchmarks.
Authors: Tijs Konijn, Imaan Bijl, Lu Cao, Fons Verbeek
Abstract: Due to the climate change, hay fever becomes a pressing healthcare problem with an increasing number of affected population, prolonged period of affect and severer symptoms. A precise pollen classification could help monitor the trend of allergic pollen in the air throughout the year and guide preventive strategies launched by municipalities. Most of the pollen classification works use 2D microscopy image or 2D projection derived from 3D image datasets. In this paper, we aim at using whole stack of 3D images for the classification and evaluating the classification performance with different deep learning models. The 3D image dataset used in this paper is from Urticaceae family, particularly the genera Urtica and Parietaria, which are morphologically similar yet differ significantly in allergenic potential. The pre-trained ResNet3D model, using optimal layer selection and extended epochs, achieved the best performance with an F1-score of 98.3%.
Authors: Riccardo Mazzieri, Jacopo Pegoraro, Michele Rossi
Abstract: The adoption of Millimeter-Wave (mmWave) radar devices for human sensing, particularly gait recognition, has recently gathered significant attention due to their efficiency, resilience to environmental conditions, and privacy-preserving nature. In this work, we tackle the challenging problem of Open-set Gait Recognition (OSGR) from sparse mmWave radar point clouds. Unlike most existing research, which assumes a closed-set scenario, our work considers the more realistic open-set case, where unknown subjects might be present at inference time, and should be correctly recognized by the system. Point clouds are well-suited for edge computing applications with resource constraints, but are more significantly affected by noise and random fluctuations than other representations, like the more common micro-Doppler signature. This is the first work addressing open-set gait recognition with sparse point cloud data. To do so, we propose a novel neural network architecture that combines supervised classification with unsupervised reconstruction of the point clouds, creating a robust, rich, and highly regularized latent space of gait features. To detect unknown subjects at inference time, we introduce a probabilistic novelty detection algorithm that leverages the structured latent space and offers a tunable trade-off between inference speed and prediction accuracy. Along with this paper, we release mmGait10, an original human gait dataset featuring over five hours of measurements from ten subjects, under varied walking modalities. Extensive experimental results show that our solution attains F1-Score improvements by 24% over state-of-the-art methods, on average, and across multiple openness levels.
Authors: Lucas Farndale, Paul Henderson, Edward W Roberts, Ke Yuan
Abstract: Self-supervised representation learning methods often fail to learn subtle or complex features, which can be dominated by simpler patterns which are much easier to learn. This limitation is particularly problematic in applications to science and engineering, as complex features can be critical for discovery and analysis. To address this, we introduce Split Component Embedding Registration (SpliCER), a novel architecture which splits the image into sections and distils information from each section to guide the model to learn more subtle and complex features without compromising on simpler features. SpliCER is compatible with any self-supervised loss function and can be integrated into existing methods without modification. The primary contributions of this work are as follows: i) we demonstrate that existing self-supervised methods can learn shortcut solutions when simple and complex features are both present; ii) we introduce a novel self-supervised training method, SpliCER, to overcome the limitations of existing methods, and achieve significant downstream performance improvements; iii) we demonstrate the effectiveness of SpliCER in cutting-edge medical and geospatial imaging settings. SpliCER offers a powerful new tool for representation learning, enabling models to uncover complex features which could be overlooked by other methods.
Authors: Lo-Wei Tai, Ching-En Li, Cheng-Lin Chen, Chih-Jung Tsai, Hwann-Tzong Chen, Tyng-Luh Liu
Abstract: Principal Component Analysis (PCA), a classical dimensionality reduction technique, and 2D Gaussian representation, an adaptation of 3D Gaussian Splatting for image representation, offer distinct approaches to modeling visual data. We present EigenGS, a novel method that bridges these paradigms through an efficient transformation pipeline connecting eigenspace and image-space Gaussian representations. Our approach enables instant initialization of Gaussian parameters for new images without requiring per-image optimization from scratch, dramatically accelerating convergence. EigenGS introduces a frequency-aware learning mechanism that encourages Gaussians to adapt to different scales, effectively modeling varied spatial frequencies and preventing artifacts in high-resolution reconstruction. Extensive experiments demonstrate that EigenGS not only achieves superior reconstruction quality compared to direct 2D Gaussian fitting but also reduces necessary parameter count and training time. The results highlight EigenGS's effectiveness and generalization ability across images with varying resolutions and diverse categories, making Gaussian-based image representation both high-quality and viable for real-time applications.
Authors: Meng Zheng, Jiajin Zhang, Benjamin Planche, Zhongpai Gao, Terrence Chen, Ziyan Wu
Abstract: Image-Text Retrieval (ITR) finds broad applications in healthcare, aiding clinicians and radiologists by automatically retrieving relevant patient cases in the database given the query image and/or report, for more efficient clinical diagnosis and treatment, especially for rare diseases. However conventional ITR systems typically only rely on global image or text representations for measuring patient image/report similarities, which overlook local distinctiveness across patient cases. This often results in suboptimal retrieval performance. In this paper, we propose an Anatomical Location-Conditioned Image-Text Retrieval (ALC-ITR) framework, which, given a query image and the associated suspicious anatomical region(s), aims to retrieve similar patient cases exhibiting the same disease or symptoms in the same anatomical region. To perform location-conditioned multimodal retrieval, we learn a medical Relevance-Region-Aligned Vision Language (RRA-VL) model with semantic global-level and region-/word-level alignment to produce generalizable, well-aligned multi-modal representations. Additionally, we perform location-conditioned contrastive learning to further utilize cross-pair region-level contrastiveness for improved multi-modal retrieval. We show that our proposed RRA-VL achieves state-of-the-art localization performance in phase-grounding tasks, and satisfying multi-modal retrieval performance with or without location conditioning. Finally, we thoroughly investigate the generalizability and explainability of our proposed ALC-ITR system in providing explanations and preliminary diagnosis reports given retrieved patient cases (conditioned on anatomical regions), with proper off-the-shelf LLM prompts.
Authors: Ao Wang, Lihao Liu, Hui Chen, Zijia Lin, Jungong Han, Guiguang Ding
Abstract: Object detection and segmentation are widely employed in computer vision applications, yet conventional models like YOLO series, while efficient and accurate, are limited by predefined categories, hindering adaptability in open scenarios. Recent open-set methods leverage text prompts, visual cues, or prompt-free paradigm to overcome this, but often compromise between performance and efficiency due to high computational demands or deployment complexity. In this work, we introduce YOLOE, which integrates detection and segmentation across diverse open prompt mechanisms within a single highly efficient model, achieving real-time seeing anything. For text prompts, we propose Re-parameterizable Region-Text Alignment (RepRTA) strategy. It refines pretrained textual embeddings via a re-parameterizable lightweight auxiliary network and enhances visual-textual alignment with zero inference and transferring overhead. For visual prompts, we present Semantic-Activated Visual Prompt Encoder (SAVPE). It employs decoupled semantic and activation branches to bring improved visual embedding and accuracy with minimal complexity. For prompt-free scenario, we introduce Lazy Region-Prompt Contrast (LRPC) strategy. It utilizes a built-in large vocabulary and specialized embedding to identify all objects, avoiding costly language model dependency. Extensive experiments show YOLOE's exceptional zero-shot performance and transferability with high inference efficiency and low training cost. Notably, on LVIS, with 3$\times$ less training cost and 1.4$\times$ inference speedup, YOLOE-v8-S surpasses YOLO-Worldv2-S by 3.5 AP. When transferring to COCO, YOLOE-v8-L achieves 0.6 AP$^b$ and 0.4 AP$^m$ gains over closed-set YOLOv8-L with nearly 4$\times$ less training time. Code and models are available at https://github.com/THU-MIG/yoloe.
Authors: K. Mawas, M. Maboudi, M. Gerke
Abstract: Given the substantial growth in the use of additive manufacturing in construction (AMC), it is necessary to ensure the quality of printed specimens which can be much more complex than conventionally manufactured parts. This study explores the various aspects of geometry and surface quality control for 3D concrete printing (3DCP), with a particular emphasis on deposition-based methods, namely extrusion and shotcrete 3D printing (SC3DP). A comprehensive overview of existing quality control (QC) methods and strategies is provided and preceded by an in-depth discussion. Four categories of data capture technologies are investigated and their advantages and limitations in the context of AMC are discussed. Additionally, the effects of environmental conditions and objects' properties on data capture are also analyzed. The study extends to automated data capture planning methods for different sensors. Furthermore, various quality control strategies are explored across different stages of the fabrication cycle of the printed object including: (i) During printing, (ii) Layer-wise, (iii) Preassembly, and (iv) Assembly. In addition to reviewing the methods already applied in AMC, we also address various research gaps and future trends and highlight potential methodologies from adjacent domains that could be transferred to AMC.
Authors: Jiahui Zhang, Fangneng Zhan, Ling Shao, Shijian Lu
Abstract: Anchor-based 3D Gaussian splatting (3D-GS) exploits anchor features in 3D Gaussian prediction, which has achieved impressive 3D rendering quality with reduced Gaussian redundancy. On the other hand, it often encounters the dilemma among anchor features, model size, and rendering quality - large anchor features lead to large 3D models and high-quality rendering whereas reducing anchor features degrades Gaussian attribute prediction which leads to clear artifacts in the rendered textures and geometries. We design SOGS, an anchor-based 3D-GS technique that introduces second-order anchors to achieve superior rendering quality and reduced anchor features and model size simultaneously. Specifically, SOGS incorporates covariance-based second-order statistics and correlation across feature dimensions to augment features within each anchor, compensating for the reduced feature size and improving rendering quality effectively. In addition, it introduces a selective gradient loss to enhance the optimization of scene textures and scene geometries, leading to high-quality rendering with small anchor features. Extensive experiments over multiple widely adopted benchmarks show that SOGS achieves superior rendering quality in novel view synthesis with clearly reduced model size.
Authors: Jiacheng Ruan, Wenzhen Yuan, Xian Gao, Ye Guo, Daoxin Zhang, Zhe Xu, Yao Hu, Ting Liu, Yuzhuo Fu
Abstract: Although large visual-language models (LVLMs) have demonstrated strong performance in multimodal tasks, errors may occasionally arise due to biases during the reasoning process. Recently, reward models (RMs) have become increasingly pivotal in the reasoning process. Specifically, process RMs evaluate each reasoning step, outcome RMs focus on the assessment of reasoning results, and critique RMs perform error analysis on the entire reasoning process, followed by corrections. However, existing benchmarks for vision-language RMs (VLRMs) typically assess only a single aspect of their capabilities (e.g., distinguishing between two answers), thus limiting the all-round evaluation and restricting the development of RMs in the visual-language domain. To address this gap, we propose a comprehensive and challenging benchmark, dubbed as VLRMBench, encompassing 12,634 questions. VLRMBench is constructed based on three distinct types of datasets, covering mathematical reasoning, hallucination understanding, and multi-image understanding. We design 12 tasks across three major categories, focusing on evaluating VLRMs in the aspects of process understanding, outcome judgment, and critique generation. Extensive experiments are conducted on 21 open-source models and 5 advanced closed-source models, highlighting the challenges posed by VLRMBench. For instance, in the `Forecasting Future', a binary classification task, the advanced GPT-4o achieves only a 76.0% accuracy. Additionally, we perform comprehensive analytical studies, offering valuable insights for the future development of VLRMs. We anticipate that VLRMBench will serve as a pivotal benchmark in advancing VLRMs. Code and datasets will be available at https://github.com/JCruan519/VLRMBench.
Authors: Zongzheng Zhang, Xinrun Li, Sizhe Zou, Guoxuan Chi, Siqi Li, Xuchong Qiu, Guoliang Wang, Guantian Zheng, Leichen Wang, Hang Zhao, Hao Zhao
Abstract: Lane topology extraction involves detecting lanes and traffic elements and determining their relationships, a key perception task for mapless autonomous driving. This task requires complex reasoning, such as determining whether it is possible to turn left into a specific lane. To address this challenge, we introduce neuro-symbolic methods powered by vision-language foundation models (VLMs). Existing approaches have notable limitations: (1) Dense visual prompting with VLMs can achieve strong performance but is costly in terms of both financial resources and carbon footprint, making it impractical for robotics applications. (2) Neuro-symbolic reasoning methods for 3D scene understanding fail to integrate visual inputs when synthesizing programs, making them ineffective in handling complex corner cases. To this end, we propose a fast-slow neuro-symbolic lane topology extraction algorithm, named Chameleon, which alternates between a fast system that directly reasons over detected instances using synthesized programs and a slow system that utilizes a VLM with a chain-of-thought design to handle corner cases. Chameleon leverages the strengths of both approaches, providing an affordable solution while maintaining high performance. We evaluate the method on the OpenLane-V2 dataset, showing consistent improvements across various baseline detectors. Our code, data, and models are publicly available at https://github.com/XR-Lee/neural-symbolic
Authors: Bangyan Li, Wenxuan Huang, Yunhang Shen, Yeqiang Wang, Shaohui Lin, Jingzhong Lin, Ling You, Yinqi Zhang, Ke Li, Xing Sun, Yuling Sun
Abstract: Recently, multimodal large models (MLLMs) have demonstrated exceptional capabilities in visual understanding and reasoning across various vision-language tasks. However, MLLMs usually perform poorly in zero-shot medical disease recognition, as they do not fully exploit the captured features and available medical knowledge. To address this challenge, we propose LLaVA-RadZ, a simple yet effective framework for zero-shot medical disease recognition. Specifically, we design an end-to-end training strategy, termed Decoding-Side Feature Alignment Training (DFAT) to take advantage of the characteristics of the MLLM decoder architecture and incorporate modality-specific tokens tailored for different modalities, which effectively utilizes image and text representations and facilitates robust cross-modal alignment. Additionally, we introduce a Domain Knowledge Anchoring Module (DKAM) to exploit the intrinsic medical knowledge of large models, which mitigates the category semantic gap in image-text alignment. DKAM improves category-level alignment, allowing for accurate disease recognition. Extensive experiments on multiple benchmarks demonstrate that our LLaVA-RadZ significantly outperforms traditional MLLMs in zero-shot disease recognition and exhibits the state-of-the-art performance compared to the well-established and highly-optimized CLIP-based approaches.
Authors: Guiwei Zhang, Tianyu Zhang, Mohan Zhou, Yalong Bai, Biye Li
Abstract: We propose V2Flow, a novel tokenizer that produces discrete visual tokens capable of high-fidelity reconstruction, while ensuring structural and latent distribution alignment with the vocabulary space of large language models (LLMs). Leveraging this tight visual-vocabulary coupling, V2Flow enables autoregressive visual generation on top of existing LLMs. Our approach formulates visual tokenization as a flow-matching problem, aiming to learn a mapping from a standard normal prior to the continuous image distribution, conditioned on token sequences embedded within the LLMs vocabulary space. The effectiveness of V2Flow stems from two core designs. First, we propose a Visual Vocabulary resampler, which compresses visual data into compact token sequences, with each represented as a soft categorical distribution over LLM's vocabulary. This allows seamless integration of visual tokens into existing LLMs for autoregressive visual generation. Second, we present a masked autoregressive Rectified-Flow decoder, employing a masked transformer encoder-decoder to refine visual tokens into contextually enriched embeddings. These embeddings then condition a dedicated velocity field for precise reconstruction. Additionally, an autoregressive rectified-flow sampling strategy is incorporated, ensuring flexible sequence lengths while preserving competitive reconstruction quality. Extensive experiments show that V2Flow outperforms mainstream VQ-based tokenizers and facilitates autoregressive visual generation on top of existing. https://github.com/zhangguiwei610/V2Flow
Authors: Calvin Yeung, Tomohiro Suzuki, Ryota Tanaka, Zhuoer Yin, Keisuke Fujii
Abstract: Human pose estimation is a critical task in computer vision and sports biomechanics, with applications spanning sports science, rehabilitation, and biomechanical research. While significant progress has been made in monocular 3D pose estimation, current datasets often fail to capture the complex, high-acceleration movements typical of competitive sports. In this work, we introduce AthletePose3D, a novel dataset designed to address this gap. AthletePose3D includes 12 types of sports motions across various disciplines, with approximately 1.3 million frames and 165 thousand individual postures, specifically capturing high-speed, high-acceleration athletic movements. We evaluate state-of-the-art (SOTA) monocular 2D and 3D pose estimation models on the dataset, revealing that models trained on conventional datasets perform poorly on athletic motions. However, fine-tuning these models on AthletePose3D notably reduces the SOTA model mean per joint position error (MPJPE) from 214mm to 65mm-a reduction of over 69%. We also validate the kinematic accuracy of monocular pose estimations through waveform analysis, highlighting strong correlations in joint angle estimations but limitations in velocity estimation. Our work provides a comprehensive evaluation of monocular pose estimation models in the context of sports, contributing valuable insights for advancing monocular pose estimation techniques in high-performance sports environments. The dataset, code, and model checkpoints are available at: https://github.com/calvinyeungck/AthletePose3D
Authors: Shiu-hong Kao, Yu-Wing Tai, Chi-Keung Tang
Abstract: Reasoning segmentation is a challenging vision-language task that aims to output the segmentation mask with respect to a complex, implicit, and even non-visual query text. Previous works incorporated multimodal Large Language Models (MLLMs) with segmentation models to approach the difficult problem. However, their segmentation quality often falls short in complex cases, particularly when dealing with out-of-domain objects with intricate structures, blurry boundaries, occlusions, or high similarity with surroundings. In this paper, we introduce ThinkFirst, a training-free reasoning segmentation framework that leverages GPT's chain of thought to address these challenging cases. Our approach allows GPT-4o or other powerful MLLMs to generate a detailed, chain-of-thought description of an image. This summarized description is then passed to a language-instructed segmentation assistant to aid the segmentation process. Our framework allows users to easily interact with the segmentation agent using multimodal inputs, such as easy text and image scribbles, for successive refinement or communication. We evaluate the performance of ThinkFirst on diverse objects. Extensive experiments show that, this zero-shot-CoT approach significantly improves the vanilla reasoning segmentation agent, both qualitatively and quantitatively, while being less sensitive or critical to user-supplied prompts after Thinking First.
Authors: Jie Hu, Shizun Wang, Xinchao Wang
Abstract: Recent advancements in 2D-to-3D perception have significantly improved the understanding of 3D scenes from 2D images. However, existing methods face critical challenges, including limited generalization across scenes, suboptimal perception accuracy, and slow reconstruction speeds. To address these limitations, we propose Perception-Efficient 3D Reconstruction (PE3R), a novel framework designed to enhance both accuracy and efficiency. PE3R employs a feed-forward architecture to enable rapid 3D semantic field reconstruction. The framework demonstrates robust zero-shot generalization across diverse scenes and objects while significantly improving reconstruction speed. Extensive experiments on 2D-to-3D open-vocabulary segmentation and 3D reconstruction validate the effectiveness and versatility of PE3R. The framework achieves a minimum 9-fold speedup in 3D semantic field reconstruction, along with substantial gains in perception accuracy and reconstruction precision, setting new benchmarks in the field. The code is publicly available at: https://github.com/hujiecpp/PE3R.
Authors: Weize Li, Yunhao Du, Qixiang Yin, Zhicheng Zhao, Fei Su, Daqi Liu
Abstract: Referring Multi-Object Tracking (RMOT) aims to localize target trajectories specified by natural language expressions in videos. Existing RMOT methods mainly follow two paradigms, namely, one-stage strategies and two-stage ones. The former jointly trains tracking with referring but suffers from substantial computational overhead. Although the latter improves computational efficiency, its CLIP-inspired dual-tower architecture restricts compatibility with other visual/text backbones and is not future-proof. To overcome these limitations, we propose CPAny, a novel encoder-decoder framework for two-stage RMOT, which introduces two core components: (1) a Contextual Visual Semantic Abstractor (CVSA) performs context-aware aggregation on visual backbone features and projects them into a unified semantic space; (2) a Parallel Semantic Summarizer (PSS) decodes the visual and linguistic features at the semantic level in parallel and generates referring scores. By replacing the inherent feature alignment of encoders with a self-constructed unified semantic space, CPAny achieves flexible compatibility with arbitrary emerging visual / text encoders. Meanwhile, CPAny aggregates contextual information by encoding only once and processes multiple expressions in parallel, significantly reducing computational redundancy. Extensive experiments on the Refer-KITTI and Refer-KITTI-V2 datasets show that CPAny outperforms SOTA methods across diverse encoder combinations, with a particular 7.77\% HOTA improvement on Refer-KITTI-V2. Code will be available soon.
Authors: Takeru Inoue, Ryusuke Miyamoto
Abstract: Instance shadow detection is the task of detecting pairs of shadows and objects, where existing methods first detect shadows and objects independently, then associate them. This paper introduces FastInstShadow, a method that enhances detection accuracy through a query-based architecture featuring an association transformer decoder with two dual-path transformer decoders to assess relationships between shadows and objects during detection. Experimental results using the SOBA dataset showed that the proposed method outperforms all existing methods across all criteria. This method makes real-time processing feasible for moderate-resolution images with better accuracy than SSISv2, the most accurate existing method. Our code is available at https://github.com/wlotkr/FastInstShadow.
Authors: Zhongwei Chen, Zhao-Xu Yang, Hai-Jun Rong, Jiawei Lang
Abstract: Traditional UAV-view Geo-Localization (UVGL) supervised paradigms are constrained by the strict reliance on paired data for positive sample selection, which limits their ability to learn cross-view domain-invariant representations from unpaired data. Moreover, it is necessary to reconstruct the pairing relationship with expensive re-labeling costs for scenario-specific training when deploying in a new domain, which fails to meet the practical demands of open-environment applications. To address this issue, we propose a novel cross-domain invariance knowledge transfer network (CDIKTNet), which comprises a cross-domain invariance sub-network and a cross-domain transfer sub-network to realize a closed-loop framework of invariance feature learning and knowledge transfer. The cross-domain invariance sub-network is utilized to construct an essentially shared feature space across domains by learning structural invariance and spatial invariance in cross-view features. Meanwhile, the cross-domain transfer sub-network uses these invariant features as anchors and employs a dual-path contrastive memory learning mechanism to mine latent cross-domain correlation patterns in unpaired data. Extensive experiments demonstrate that our method achieves state-of-the-art performance under fully supervised conditions. More importantly, with merely 2\% paired data, our method exhibits performance comparable to existing supervised paradigms and possesses the ability to transfer directly to qualify for applications in the other scenarios completely without any prior pairing relationship.
Authors: Zhangquan Chen, Xufang Luo, Dongsheng Li
Abstract: Visual understanding is inherently intention-driven - humans selectively focus on different regions of a scene based on their goals. Recent advances in large multimodal models (LMMs) enable flexible expression of such intentions through natural language, allowing queries to guide visual reasoning processes. Frameworks like Visual Chain-of-Thought have demonstrated the benefit of incorporating explicit reasoning steps, where the model predicts a focus region before answering a query. However, existing approaches rely heavily on supervised training with annotated intermediate bounding boxes, which severely limits scalability due to the combinatorial explosion of intention-region pairs. To overcome this limitation, we propose VisRL, the first framework that applies reinforcement learning (RL) to the problem of intention-driven visual perception. VisRL optimizes the entire visual reasoning process using only reward signals. By treating intermediate focus selection as a internal decision optimized through trial-and-error, our method eliminates the need for costly region annotations while aligning more closely with how humans learn to perceive the world. Extensive experiments across multiple benchmarks show that VisRL consistently outperforms strong baselines, demonstrating both its effectiveness and its strong generalization across different LMMs. Our code is available at this [URL](https://github.com/zhangquanchen/VisRL).
Authors: Cl\'ement Chadebec, Onur Tasar, Sanjeev Sreetharan, Benjamin Aubin
Abstract: In this paper, we introduce Latent Bridge Matching (LBM), a new, versatile and scalable method that relies on Bridge Matching in a latent space to achieve fast image-to-image translation. We show that the method can reach state-of-the-art results for various image-to-image tasks using only a single inference step. In addition to its efficiency, we also demonstrate the versatility of the method across different image translation tasks such as object removal, normal and depth estimation, and object relighting. We also derive a conditional framework of LBM and demonstrate its effectiveness by tackling the tasks of controllable image relighting and shadow generation. We provide an open-source implementation of the method at https://github.com/gojasper/LBM.
Authors: Thibaut Loiseau, Guillaume Bourmaud, Vincent Lepetit
Abstract: Pre-training techniques have greatly advanced computer vision, with CroCo's cross-view completion approach yielding impressive results in tasks like 3D reconstruction and pose regression. However, this method requires substantial overlap between training pairs, limiting its effectiveness. We introduce Alligat0R, a novel pre-training approach that reformulates cross-view learning as a co-visibility segmentation task. Our method predicts whether each pixel in one image is co-visible in the second image, occluded, or outside the field of view (FOV), enabling the use of image pairs with any degree of overlap and providing interpretable predictions. To support this, we present Cub3, a large-scale dataset with 2.5 million image pairs and dense co-visibility annotations derived from the nuScenes dataset. This dataset includes diverse scenarios with varying degrees of overlap. The experiments show that Alligat0R significantly outperforms CroCo in relative pose regression, especially in scenarios with limited overlap. Alligat0R and Cub3 will be made publicly available.
Authors: Jen-tse Huang, Jiantong Qin, Jianping Zhang, Youliang Yuan, Wenxuan Wang, Jieyu Zhao
Abstract: This research investigates both explicit and implicit social biases exhibited by Vision-Language Models (VLMs). The key distinction between these bias types lies in the level of awareness: explicit bias refers to conscious, intentional biases, while implicit bias operates subconsciously. To analyze explicit bias, we directly pose questions to VLMs related to gender and racial differences: (1) Multiple-choice questions based on a given image (e.g., "What is the education level of the person in the image?") (2) Yes-No comparisons using two images (e.g., "Is the person in the first image more educated than the person in the second image?") For implicit bias, we design tasks where VLMs assist users but reveal biases through their responses: (1) Image description tasks: Models are asked to describe individuals in images, and we analyze disparities in textual cues across demographic groups. (2) Form completion tasks: Models draft a personal information collection form with 20 attributes, and we examine correlations among selected attributes for potential biases. We evaluate Gemini-1.5, GPT-4V, GPT-4o, LLaMA-3.2-Vision and LLaVA-v1.6. Our code and data are publicly available at https://github.com/uscnlp-lime/VisBias.
Authors: Dunant Cusipuma, David Ortega, Victor Flores-Benites, Arturo Deza
Abstract: As multimodal foundational models start being deployed experimentally in Self-Driving cars, a reasonable question we ask ourselves is how similar to humans do these systems respond in certain driving situations -- especially those that are out-of-distribution? To study this, we create the Robusto-1 dataset that uses dashcam video data from Peru, a country with one of the worst (aggressive) drivers in the world, a high traffic index, and a high ratio of bizarre to non-bizarre street objects likely never seen in training. In particular, to preliminarly test at a cognitive level how well Foundational Visual Language Models (VLMs) compare to Humans in Driving, we move away from bounding boxes, segmentation maps, occupancy maps or trajectory estimation to multi-modal Visual Question Answering (VQA) comparing both humans and machines through a popular method in systems neuroscience known as Representational Similarity Analysis (RSA). Depending on the type of questions we ask and the answers these systems give, we will show in what cases do VLMs and Humans converge or diverge allowing us to probe on their cognitive alignment. We find that the degree of alignment varies significantly depending on the type of questions asked to each type of system (Humans vs VLMs), highlighting a gap in their alignment.
Authors: Junwei Luo, Yingying Zhang, Xue Yang, Kang Wu, Qi Zhu, Lei Liang, Jingdong Chen, Yansheng Li
Abstract: Efficient vision-language understanding of large Remote Sensing Images (RSIs) is meaningful but challenging. Current Large Vision-Language Models (LVLMs) typically employ limited pre-defined grids to process images, leading to information loss when handling gigapixel RSIs. Conversely, using unlimited grids significantly increases computational costs. To preserve image details while reducing computational complexity, we propose a text-guided token pruning method with Dynamic Image Pyramid (DIP) integration. Our method introduces: (i) a Region Focus Module (RFM) that leverages text-aware region localization capability to identify critical vision tokens, and (ii) a coarse-to-fine image tile selection and vision token pruning strategy based on DIP, which is guided by RFM outputs and avoids directly processing the entire large imagery. Additionally, existing benchmarks for evaluating LVLMs' perception ability on large RSI suffer from limited question diversity and constrained image sizes. We construct a new benchmark named LRS-VQA, which contains 7,333 QA pairs across 8 categories, with image length up to 27,328 pixels. Our method outperforms existing high-resolution strategies on four datasets using the same data. Moreover, compared to existing token reduction methods, our approach demonstrates higher efficiency under high-resolution settings. Dataset and code are in https://github.com/VisionXLab/LRS-VQA.
Authors: Bardia Safaei, Faizan Siddiqui, Jiacong Xu, Vishal M. Patel, Shao-Yuan Lo
Abstract: Visual instruction tuning (VIT) for large vision-language models (LVLMs) requires training on expansive datasets of image-instruction pairs, which can be costly. Recent efforts in VIT data selection aim to select a small subset of high-quality image-instruction pairs, reducing VIT runtime while maintaining performance comparable to full-scale training. However, a major challenge often overlooked is that generating instructions from unlabeled images for VIT is highly expensive. Most existing VIT datasets rely heavily on human annotations or paid services like the GPT API, which limits users with constrained resources from creating VIT datasets for custom applications. To address this, we introduce Pre-Instruction Data Selection (PreSel), a more practical data selection paradigm that directly selects the most beneficial unlabeled images and generates instructions only for the selected images. PreSel first estimates the relative importance of each vision task within VIT datasets to derive task-wise sampling budgets. It then clusters image features within each task, selecting the most representative images with the budget. This approach reduces computational overhead for both instruction generation during VIT data formation and LVLM fine-tuning. By generating instructions for only 15% of the images, PreSel achieves performance comparable to full-data VIT on the LLaVA-1.5 and Vision-Flan datasets. The link to our project page: https://bardisafa.github.io/PreSel
Authors: Youjun Zhao, Jiaying Lin, Rynson W. H. Lau
Abstract: Open-vocabulary 3D object detection (OV-3DOD) aims at localizing and classifying novel objects beyond closed sets. The recent success of vision-language models (VLMs) has demonstrated their remarkable capabilities to understand open vocabularies. Existing works that leverage VLMs for 3D object detection (3DOD) generally resort to representations that lose the rich scene context required for 3D perception. To address this problem, we propose in this paper a hierarchical framework, named HCMA, to simultaneously learn local object and global scene information for OV-3DOD. Specifically, we first design a Hierarchical Data Integration (HDI) approach to obtain coarse-to-fine 3D-image-text data, which is fed into a VLM to extract object-centric knowledge. To facilitate the association of feature hierarchies, we then propose an Interactive Cross-Modal Alignment (ICMA) strategy to establish effective intra-level and inter-level feature connections. To better align features across different levels, we further propose an Object-Focusing Context Adjustment (OFCA) module to refine multi-level features by emphasizing object-related features. Extensive experiments demonstrate that the proposed method outperforms SOTA methods on the existing OV-3DOD benchmarks. It also achieves promising OV-3DOD results even without any 3D annotations.
Authors: Yuhong Zhang, Guanlin Wu, Ling-Hao Chen, Zhuokai Zhao, Jing Lin, Xiaoke Jiang, Jiamin Wu, Zhuoheng Li, Hao Frank Yang, Haoqian Wang, Lei Zhang
Abstract: In this paper, we present a novel framework designed to reconstruct long-sequence 3D human motion in the world coordinates from in-the-wild videos with multiple shot transitions. Such long-sequence in-the-wild motions are highly valuable to applications such as motion generation and motion understanding, but are of great challenge to be recovered due to abrupt shot transitions, partial occlusions, and dynamic backgrounds presented in such videos. Existing methods primarily focus on single-shot videos, where continuity is maintained within a single camera view, or simplify multi-shot alignment in camera space only. In this work, we tackle the challenges by integrating an enhanced camera pose estimation with Human Motion Recovery (HMR) by incorporating a shot transition detector and a robust alignment module for accurate pose and orientation continuity across shots. By leveraging a custom motion integrator, we effectively mitigate the problem of foot sliding and ensure temporal consistency in human pose. Extensive evaluations on our created multi-shot dataset from public 3D human datasets demonstrate the robustness of our method in reconstructing realistic human motion in world coordinates.
Authors: Zeyinzi Jiang, Zhen Han, Chaojie Mao, Jingfeng Zhang, Yulin Pan, Yu Liu
Abstract: Diffusion Transformer has demonstrated powerful capability and scalability in generating high-quality images and videos. Further pursuing the unification of generation and editing tasks has yielded significant progress in the domain of image content creation. However, due to the intrinsic demands for consistency across both temporal and spatial dynamics, achieving a unified approach for video synthesis remains challenging. We introduce VACE, which enables users to perform Video tasks within an All-in-one framework for Creation and Editing. These tasks include reference-to-video generation, video-to-video editing, and masked video-to-video editing. Specifically, we effectively integrate the requirements of various tasks by organizing video task inputs, such as editing, reference, and masking, into a unified interface referred to as the Video Condition Unit (VCU). Furthermore, by utilizing a Context Adapter structure, we inject different task concepts into the model using formalized representations of temporal and spatial dimensions, allowing it to handle arbitrary video synthesis tasks flexibly. Extensive experiments demonstrate that the unified model of VACE achieves performance on par with task-specific models across various subtasks. Simultaneously, it enables diverse applications through versatile task combinations. Project page: https://ali-vilab.github.io/VACE-Page/.
Authors: Yuxin Jiang, Liming Jiang, Shuai Yang, Jia-Wei Liu, Ivor Tsang, Mike Zheng Shou
Abstract: We present Style Matching Score (SMS), a novel optimization method for image stylization with diffusion models. Balancing effective style transfer with content preservation is a long-standing challenge. Unlike existing efforts, our method reframes image stylization as a style distribution matching problem. The target style distribution is estimated from off-the-shelf style-dependent LoRAs via carefully designed score functions. To preserve content information adaptively, we propose Progressive Spectrum Regularization, which operates in the frequency domain to guide stylization progressively from low-frequency layouts to high-frequency details. In addition, we devise a Semantic-Aware Gradient Refinement technique that leverages relevance maps derived from diffusion semantic priors to selectively stylize semantically important regions. The proposed optimization formulation extends stylization from pixel space to parameter space, readily applicable to lightweight feedforward generators for efficient one-step stylization. SMS effectively balances style alignment and content preservation, outperforming state-of-the-art approaches, verified by extensive experiments.
Authors: Yujie Wei, Shiwei Zhang, Hangjie Yuan, Biao Gong, Longxiang Tang, Xiang Wang, Haonan Qiu, Hengjia Li, Shuai Tan, Yingya Zhang, Hongming Shan
Abstract: Relational video customization refers to the creation of personalized videos that depict user-specified relations between two subjects, a crucial task for comprehending real-world visual content. While existing methods can personalize subject appearances and motions, they still struggle with complex relational video customization, where precise relational modeling and high generalization across subject categories are essential. The primary challenge arises from the intricate spatial arrangements, layout variations, and nuanced temporal dynamics inherent in relations; consequently, current models tend to overemphasize irrelevant visual details rather than capturing meaningful interactions. To address these challenges, we propose DreamRelation, a novel approach that personalizes relations through a small set of exemplar videos, leveraging two key components: Relational Decoupling Learning and Relational Dynamics Enhancement. First, in Relational Decoupling Learning, we disentangle relations from subject appearances using relation LoRA triplet and hybrid mask training strategy, ensuring better generalization across diverse relationships. Furthermore, we determine the optimal design of relation LoRA triplet by analyzing the distinct roles of the query, key, and value features within MM-DiT's attention mechanism, making DreamRelation the first relational video generation framework with explainable components. Second, in Relational Dynamics Enhancement, we introduce space-time relational contrastive loss, which prioritizes relational dynamics while minimizing the reliance on detailed subject appearances. Extensive experiments demonstrate that DreamRelation outperforms state-of-the-art methods in relational video customization. Code and models will be made publicly available.
Authors: Sedrick Keh, Jean Mercat, Samir Yitzhak Gadre, Kushal Arora, Igor Vasiljevic, Benjamin Burchfiel, Shuran Song, Russ Tedrake, Thomas Kollar, Ludwig Schmidt, Achal Dave
Abstract: Pre-trained LLMs that are further trained with image data perform well on vision-language tasks. While adding images during a second training phase effectively unlocks this capability, it is unclear how much of a gain or loss this two-step pipeline gives over VLMs which integrate images earlier into the training process. To investigate this, we train models spanning various datasets, scales, image-text ratios, and amount of pre-training done before introducing vision tokens. We then fine-tune these models and evaluate their downstream performance on a suite of vision-language and text-only tasks. We find that pre-training with a mixture of image and text data allows models to perform better on vision-language tasks while maintaining strong performance on text-only evaluations. On an average of 6 diverse tasks, we find that for a 1B model, introducing visual tokens 80% of the way through pre-training results in a 2% average improvement over introducing visual tokens to a fully pre-trained model.
Authors: Ying Xu, Marius Pedersen, Kiran Raja
Abstract: The rapid development of deep learning and generative AI technologies has profoundly transformed the digital contact landscape, creating realistic Deepfake that poses substantial challenges to public trust and digital media integrity. This paper introduces a novel Deepfake detention framework, Volume of Differences (VoD), designed to enhance detection accuracy by exploiting temporal and spatial inconsistencies between consecutive video frames. VoD employs a progressive learning approach that captures differences across multiple axes through the use of consecutive frame differences (CFD) and a network with stepwise expansions. We evaluate our approach with intra-dataset and cross-dataset testing scenarios on various well-known Deepfake datasets. Our findings demonstrate that VoD excels with the data it has been trained on and shows strong adaptability to novel, unseen data. Additionally, comprehensive ablation studies examine various configurations of segment length, sampling steps, and intervals, offering valuable insights for optimizing the framework. The code for our VoD framework is available at https://github.com/xuyingzhongguo/VoD.
Authors: Bo Jiang, Shaoyu Chen, Qian Zhang, Wenyu Liu, Xinggang Wang
Abstract: OpenAI o1 and DeepSeek R1 achieve or even surpass human expert-level performance in complex domains like mathematics and science, with reinforcement learning (RL) and reasoning playing a crucial role. In autonomous driving, recent end-to-end models have greatly improved planning performance but still struggle with long-tailed problems due to limited common sense and reasoning abilities. Some studies integrate vision-language models (VLMs) into autonomous driving, but they typically rely on pre-trained models with simple supervised fine-tuning (SFT) on driving data, without further exploration of training strategies or optimizations specifically tailored for planning. In this paper, we propose AlphaDrive, a RL and reasoning framework for VLMs in autonomous driving. AlphaDrive introduces four GRPO-based RL rewards tailored for planning and employs a two-stage planning reasoning training strategy that combines SFT with RL. As a result, AlphaDrive significantly improves both planning performance and training efficiency compared to using only SFT or without reasoning. Moreover, we are also excited to discover that, following RL training, AlphaDrive exhibits some emergent multimodal planning capabilities, which is critical for improving driving safety and efficiency. To the best of our knowledge, AlphaDrive is the first to integrate GRPO-based RL with planning reasoning into autonomous driving. Code will be released to facilitate future research.
Authors: Zanting Ye, Xiaolong Niu, Xuanbin Wu, Wantong Lu, Lijun Lu
Abstract: Organ segmentation in Positron Emission Tomography (PET) plays a vital role in cancer quantification. Low-dose PET (LDPET) provides a safer alternative by reducing radiation exposure. However, the inherent noise and blurred boundaries make organ segmentation more challenging. Additionally, existing PET organ segmentation methods rely on co-registered Computed Tomography (CT) annotations, overlooking the problem of modality mismatch. In this study, we propose LDOS, a novel CT-free ultra-LDPET organ segmentation pipeline. Inspired by Masked Autoencoders (MAE), we reinterpret LDPET as a naturally masked version of Full-Dose PET (FDPET). LDOS adopts a simple yet effective architecture: a shared encoder extracts generalized features, while task-specific decoders independently refine outputs for denoising and segmentation. By integrating CT-derived organ annotations into the denoising process, LDOS improves anatomical boundary recognition and alleviates the PET/CT misalignments. Experiments demonstrate that LDOS achieves state-of-the-art performance with mean Dice scores of 73.11% (18F-FDG) and 73.97% (68Ga-FAPI) across 18 organs in 5% dose PET. Our code is publicly available.
Authors: Selcuk Yazar
Abstract: Illumination estimation remains a pivotal challenge in image processing, particularly for robotics, where robust environmental perception is essential under varying lighting conditions. Traditional approaches, such as RGB histograms and GIST descriptors, often fail in complex scenarios due to their sensitivity to illumination changes. This study introduces a novel method utilizing the Wasserstein distance, rooted in optimal transport theory, to estimate illuminant and light direction in images. Experiments on diverse images indoor scenes, black-and-white photographs, and night images demonstrate the method's efficacy in detecting dominant light sources and estimating their directions, outperforming traditional statistical methods in complex lighting environments. The approach shows promise for applications in light source localization, image quality assessment, and object detection enhancement. Future research may explore adaptive thresholding and integrate gradient analysis to enhance accuracy, offering a scalable solution for real-world illumination challenges in robotics and beyond.
Authors: A. Quadir, M. Tanveer
Abstract: In recent years, the random vector functional link (RVFL) network has gained significant popularity in hyperspectral image (HSI) classification due to its simplicity, speed, and strong generalization performance. However, despite these advantages, RVFL models face several limitations, particularly in handling non-linear relationships and complex data structures. The random initialization of input-to-hidden weights can lead to instability, and the model struggles with determining the optimal number of hidden nodes, affecting its performance on more challenging datasets. To address these issues, we propose a novel randomized based restricted kernel machine ($R^2KM$) model that combines the strehyperngths of RVFL and restricted kernel machines (RKM). $R^2KM$ introduces a layered structure that represents kernel methods using both visible and hidden variables, analogous to the energy function in restricted Boltzmann machines (RBM). This structure enables $R^2KM$ to capture complex data interactions and non-linear relationships more effectively, improving both interpretability and model robustness. A key contribution of $R^2KM$ is the introduction of a novel conjugate feature duality based on the Fenchel-Young inequality, which expresses the problem in terms of conjugate dual variables and provides an upper bound on the objective function. This duality enhances the model's flexibility and scalability, offering a more efficient and flexible solution for complex data analysis tasks. Extensive experiments on hyperspectral image datasets and real-world data from the UCI and KEEL repositories show that $R^2KM$ outperforms baseline models, demonstrating its effectiveness in classification and regression tasks.
Authors: Mostafa Ahmed Mostafa Ahmed, Mohamed Khaled Mohamed Elsayed, Radwa Waheed Ezzat Abdelmohsen
Abstract: The automotive industry is increasingly reliant on software to manage complex vehicle functionalities, making efficient and secure firmware updates essential. Traditional firmware update methods, requiring physical connections through On-Board Diagnostics (OBD) ports, are inconvenient, costly, and time-consuming. Firmware Over-the-Air (FOTA) technology offers a revolutionary solution by enabling wireless updates, reducing operational costs, and enhancing the user experience. This project aims to design and implement an advanced FOTA system tailored for modern vehicles, incorporating the AUTOSAR architecture for scalability and standardization, and utilizing delta updating to minimize firmware update sizes, thereby improving bandwidth efficiency and reducing flashing times. To ensure security, the system integrates the UDS 0x27 protocol for authentication and data integrity during the update process. Communication between Electronic Control Units (ECUs) is achieved using the CAN protocol, while the ESP8266 module and the master ECU communicate via SPI for data transfer. The system's architecture includes key components such as a bootloader, boot manager, and bootloader updater to facilitate seamless firmware updates. The functionality of the system is demonstrated through two applications: a blinking LED and a Lane Keeping Assist (LKA) system, showcasing its versatility in handling critical automotive features. This project represents a significant step forward in automotive technology, offering a user-centric, efficient, and secure solution for automotive firmware management.
Authors: Yiqing Guo, Nagur Cherukuru, Eric Lehmann, Xiubin Qi, Mark Doubelld, S. L. Kesav Unnithan, Ming Feng
Abstract: Sea surface temperature (SST) is a fundamental physical parameter characterising the thermal state of sea surface. The Thermal Infrared Sensor (TIRS) onboard Landsat-8, with its 100-meter spatial resolution, offers a unique opportunity to uncover fine-scale coastal SST patterns that would otherwise be overlooked by coarser-resolution thermal sensors. In this study, we first develop an operational approach for SST retrieval from the TIRS sensor, and subsequently propose a novel algorithm for establishing daily SST climatology which serves as the baseline to detect anomalous SST events. We applied the proposed methods to temperate coastal waters in South Australia for the ten-year period from 2014 to 2023. For ground validation purposes, a buoy was deployed off the coast of Port Lincoln, South Australia, to record in-situ time-series SST. The spatiotemporal patterns of SST in the study area were analysed based on the ten years of satellite-derived SST imagery. The daily baseline climatology of SST with 100 m resolution was constructed, which allowed for the detection and analysis of anomalous SST events during the study period of 2014-2023. Our results suggest the following: (1) the satellite-derived SST data, generated with the proposed algorithm, aligned well with the in-situ measured SST values; (2) the semi-enclosed, shallow regions of Upper Spencer Gulf and Upper St Vincent Gulf showed higher temperatures during summer and cooler temperatures during winter than waters closer to the open ocean, resulting in a higher seasonal variation in SST; (3) the near-shore shallow areas in Spencer Gulf and St Vincent Gulf, and regions surrounding Kangaroo Island, were identified to have a higher probability of SST anomalies compared to the rest of the study area; and (4) anomalous SST events were more likely to happen during the warm months than the cool months.
Authors: Sefik Serengil, Alper Ozpinar
Abstract: This paper explores the use of partially homomorphic encryption (PHE) for encrypted vector similarity search, with a focus on facial recognition and broader applications like reverse image search, recommendation engines, and large language models (LLMs). While fully homomorphic encryption (FHE) exists, we demonstrate that encrypted cosine similarity can be computed using PHE, offering a more practical alternative. Since PHE does not directly support cosine similarity, we propose a method that normalizes vectors in advance, enabling dot product calculations as a proxy. We also apply min-max normalization to handle negative dimension values. Experiments on the Labeled Faces in the Wild (LFW) dataset use DeepFace's FaceNet128d, FaceNet512d, and VGG-Face (4096d) models in a two-tower setup. Pre-encrypted embeddings are stored in one tower, while an edge device captures images, computes embeddings, and performs encrypted-plaintext dot products via additively homomorphic encryption. We implement this with LightPHE, evaluating Paillier, Damgard-Jurik, and Okamoto-Uchiyama schemes, excluding others due to performance or decryption complexity. Tests at 80-bit and 112-bit security (NIST-secure until 2030) compare PHE against FHE (via TenSEAL), analyzing encryption, decryption, operation time, cosine similarity loss, key/ciphertext sizes. Results show PHE is less computationally intensive, faster, and produces smaller ciphertexts/keys, making it well-suited for memory-constrained environments and real-world privacy-preserving encrypted similarity search.
Authors: Carson Sobolewski, Zhenjiang Mao, Kshitij Vejre, Ivan Ruchkin
Abstract: Vision-based autonomous racing relies on accurate perception for robust control. However, image distribution changes caused by sensor noise, adverse weather, and dynamic lighting can degrade perception, leading to suboptimal control decisions. Existing approaches, including domain adaptation and adversarial training, improve robustness but struggle to generalize to unseen corruptions while introducing computational overhead. To address this challenge, we propose a real-time image repair module that restores corrupted images before they are used by the controller. Our method leverages generative adversarial models, specifically CycleGAN and pix2pix, for image repair. CycleGAN enables unpaired image-to-image translation to adapt to novel corruptions, while pix2pix exploits paired image data when available to improve the quality. To ensure alignment with control performance, we introduce a control-focused loss function that prioritizes perceptual consistency in repaired images. We evaluated our method in a simulated autonomous racing environment with various visual corruptions. The results show that our approach significantly improves performance compared to baselines, mitigating distribution shift and enhancing controller reliability.
Authors: Danielle L. Ferreira, Ahana Gangopadhyay, Hsi-Ming Chang, Ravi Soni, Gopal Avinash
Abstract: Accurate segmentation of anatomical structures in ultrasound (US) images, particularly small ones, is challenging due to noise and variability in imaging conditions (e.g., probe position, patient anatomy, tissue characteristics and pathology). To address this, we introduce Segment Anything Small (SAS), a simple yet effective scale- and texture-aware data augmentation technique designed to enhance the performance of deep learning models for segmenting small anatomical structures in ultrasound images. SAS employs a dual transformation strategy: (1) simulating diverse organ scales by resizing and embedding organ thumbnails into a black background, and (2) injecting noise into regions of interest to simulate varying tissue textures. These transformations generate realistic and diverse training data without introducing hallucinations or artifacts, improving the model's robustness to noise and variability. We fine-tuned a promptable foundation model on a controlled organ-specific medical imaging dataset and evaluated its performance on one internal and five external datasets. Experimental results demonstrate significant improvements in segmentation performance, with Dice score gains of up to 0.35 and an average improvement of 0.16 [95% CI 0.132,0.188]. Additionally, our iterative point prompts provide precise control and adaptive refinement, achieving performance comparable to bounding box prompts with just two points. SAS enhances model robustness and generalizability across diverse anatomical structures and imaging conditions, particularly for small structures, without compromising the accuracy of larger ones. By offering a computationally efficient solution that eliminates the need for extensive human labeling efforts, SAS emerges as a powerful tool for advancing medical image analysis, particularly in resource-constrained settings.
Authors: Yao Du, Jiaxin Zhuang, Xiaoyu Zheng, Jing Cong, Limei Guo, Chao He, Lin Luo, Xiaomeng Li
Abstract: Histopathology image analysis is fundamental to digital pathology, with hematoxylin and eosin (H&E) staining as the gold standard for diagnostic and prognostic assessments. While H&E imaging effectively highlights cellular and tissue structures, it lacks sensitivity to birefringence and tissue anisotropy, which are crucial for assessing collagen organization, fiber alignment, and microstructural alterations--key indicators of tumor progression, fibrosis, and other pathological conditions. To bridge this gap, we propose PolarHE, a dual modality fusion framework that integrates H&E with polarization imaging, leveraging the polarization ability to enhance tissue characterization. Our approach employs a feature decomposition strategy to disentangle common and modality specific features, ensuring effective multimodal representation learning. Through comprehensive validation, our approach significantly outperforms previous methods, achieving an accuracy of 86.70% on the Chaoyang dataset and 89.06% on the MHIST dataset. Moreover, polarization property visualization reveals distinct optical signatures of pathological tissues, highlighting its diagnostic potential. t-SNE visualizations further confirm our model effectively captures both shared and unique modality features, reinforcing the complementary nature of polarization imaging. These results demonstrate that polarization imaging is a powerful and underutilized modality in computational pathology, enriching feature representation and improving diagnostic accuracy. PolarHE establishes a promising direction for multimodal learning, paving the way for more interpretable and generalizable pathology models. Our code will be released after paper acceptance.
Authors: Franklin Mingzhe Li, Kaitlyn Ng, Bin Zhu, Patrick Carrington
Abstract: Following recipes while cooking is an important but difficult task for visually impaired individuals. We developed OSCAR (Object Status Context Awareness for Recipes), a novel approach that provides recipe progress tracking and context-aware feedback on the completion of cooking tasks through tracking object statuses. OSCAR leverages both Large-Language Models (LLMs) and Vision-Language Models (VLMs) to manipulate recipe steps, extract object status information, align visual frames with object status, and provide cooking progress tracking log. We evaluated OSCAR's recipe following functionality using 173 YouTube cooking videos and 12 real-world non-visual cooking videos to demonstrate OSCAR's capability to track cooking steps and provide contextual guidance. Our results highlight the effectiveness of using object status to improve performance compared to baseline by over 20% across different VLMs, and we present factors that impact prediction performance. Furthermore, we contribute a dataset of real-world non-visual cooking videos with step annotations as an evaluation benchmark.
Authors: Krish Didwania, Ishaan Gakhar, Prakhar Arya, Sanskriti Labroo
Abstract: Contrast enhancement, a key aspect of image-to-image translation (I2IT), improves visual quality by adjusting intensity differences between pixels. However, many existing methods struggle to preserve fine-grained details, often leading to the loss of low-level features. This paper introduces LapLoss, a novel approach designed for I2IT contrast enhancement, based on the Laplacian pyramid-centric networks, forming the core of our proposed methodology. The proposed approach employs a multiple discriminator architecture, each operating at a different resolution to capture high-level features, in addition to maintaining low-level details and textures under mixed lighting conditions. The proposed methodology computes the loss at multiple scales, balancing reconstruction accuracy and perceptual quality to enhance overall image generation. The distinct blend of the loss calculation at each level of the pyramid, combined with the architecture of the Laplacian pyramid enables LapLoss to exceed contemporary contrast enhancement techniques. This framework achieves state-of-the-art results, consistently performing well across different lighting conditions in the SICE dataset.
Authors: Qi Zhang, Shunan Zhang, Ziqi Zhao, Kun Wang, Jun Xu, Jianqi Sun
Abstract: Osteoporotic vertebral compression fractures (VCFs) are prevalent in the elderly population, typically assessed on computed tomography (CT) scans by evaluating vertebral height loss. This assessment helps determine the fracture's impact on spinal stability and the need for surgical intervention. However, clinical data indicate that many VCFs exhibit irregular compression, complicating accurate diagnosis. While deep learning methods have shown promise in aiding VCFs screening, they often lack interpretability and sufficient sensitivity, limiting their clinical applicability. To address these challenges, we introduce a novel vertebra synthesis-height loss quantification-VCFs grading framework. Our proposed model, HealthiVert-GAN, utilizes a coarse-to-fine synthesis network designed to generate pseudo-healthy vertebral images that simulate the pre-fracture state of fractured vertebrae. This model integrates three auxiliary modules that leverage the morphology and height information of adjacent healthy vertebrae to ensure anatomical consistency. Additionally, we introduce the Relative Height Loss of Vertebrae (RHLV) as a quantification metric, which divides each vertebra into three sections to measure height loss between pre-fracture and post-fracture states, followed by fracture severity classification using a Support Vector Machine (SVM). Our approach achieves state-of-the-art classification performance on both the Verse2019 dataset and our private dataset, and it provides cross-sectional distribution maps of vertebral height loss. This practical tool enhances diagnostic sensitivity in clinical settings and assisting in surgical decision-making. Our code is available: https://github.com/zhibaishouheilab/HealthiVert-GAN.
Authors: Zixuan Liu, Aaron Honjaya, Yuekai Xu, Yi Zhang, Hefu Pan, Xin Wang, Linda G Shapiro, Sheng Wang, Ruikang K Wang
Abstract: Retinal vessel segmentation is critical for diagnosing ocular conditions, yet current deep learning methods are limited by modality-specific challenges and significant distribution shifts across imaging devices, resolutions, and anatomical regions. In this paper, we propose GrInAdapt, a novel framework for source-free multi-target domain adaptation that leverages multi-view images to refine segmentation labels and enhance model generalizability for optical coherence tomography angiography (OCTA) of the fundus of the eye. GrInAdapt follows an intuitive three-step approach: (i) grounding images to a common anchor space via registration, (ii) integrating predictions from multiple views to achieve improved label consensus, and (iii) adapting the source model to diverse target domains. Furthermore, GrInAdapt is flexible enough to incorporate auxiliary modalities such as color fundus photography, to provide complementary cues for robust vessel segmentation. Extensive experiments on a multi-device, multi-site, and multi-modal retinal dataset demonstrate that GrInAdapt significantly outperforms existing domain adaptation methods, achieving higher segmentation accuracy and robustness across multiple domains. These results highlight the potential of GrInAdapt to advance automated retinal vessel analysis and support robust clinical decision-making.
Authors: Xudong Lu, Yinghao Chen, Renshou Wu, Haohao Gao, Xi Chen, Xue Yang, Xiangyu Zhao, Aojun Zhou, Fangyuan Li, Yafei Wen, Xiaoxin Chen, Shuai Ren, Hongsheng Li
Abstract: Recent advancements in Multimodal Large Language Models (MLLMs) have enabled their deployment on mobile devices. However, challenges persist in maintaining strong language capabilities and ensuring hardware compatibility, both of which are crucial for user experience and practical deployment efficiency. In our deployment process, we observe that existing MLLMs often face performance degradation on pure language tasks, and the current NPU platforms on smartphones do not support the MoE architecture, which is commonly used to preserve pure language capabilities during multimodal training. To address these issues, we systematically analyze methods to maintain pure language capabilities during the training of MLLMs, focusing on both training data and model architecture aspects. Based on these analyses, we propose GenieBlue, an efficient MLLM structural design that integrates both linguistic and multimodal capabilities for LLMs on mobile devices. GenieBlue freezes the original LLM parameters during MLLM training to maintain pure language capabilities. It acquires multimodal capabilities by duplicating specific transformer blocks for full fine-tuning and integrating lightweight LoRA modules. This approach preserves language capabilities while achieving comparable multimodal performance through extensive training. Deployed on smartphone NPUs, GenieBlue demonstrates efficiency and practicality for applications on mobile devices.
Authors: Masaru Yajima, Kei Ota, Asako Kanezaki, Rei Kawakami
Abstract: Achieving zero-shot peg insertion, where inserting an arbitrary peg into an unseen hole without task-specific training, remains a fundamental challenge in robotics. This task demands a highly generalizable perception system capable of detecting potential holes, selecting the correct mating hole from multiple candidates, estimating its precise pose, and executing insertion despite uncertainties. While learning-based methods have been applied to peg insertion, they often fail to generalize beyond the specific peg-hole pairs encountered during training. Recent advancements in Vision-Language Models (VLMs) offer a promising alternative, leveraging large-scale datasets to enable robust generalization across diverse tasks. Inspired by their success, we introduce a novel zero-shot peg insertion framework that utilizes a VLM to identify mating holes and estimate their poses without prior knowledge of their geometry. Extensive experiments demonstrate that our method achieves 90.2% accuracy, significantly outperforming baselines in identifying the correct mating hole across a wide range of previously unseen peg-hole pairs, including 3D-printed objects, toy puzzles, and industrial connectors. Furthermore, we validate the effectiveness of our approach in a real-world connector insertion task on a backpanel of a PC, where our system successfully detects holes, identifies the correct mating hole, estimates its pose, and completes the insertion with a success rate of 88.3%. These results highlight the potential of VLM-driven zero-shot reasoning for enabling robust and generalizable robotic assembly.
Authors: Md Sadman Sakib, Yu Sun
Abstract: Modern robotic systems, deployed across domains from industrial automation to domestic assistance, face a critical challenge: executing tasks with precision and adaptability in dynamic, unpredictable environments. To address this, we propose STAR (Smart Task Adaptation and Recovery), a novel framework that synergizes Foundation Models (FMs) with dynamically expanding Knowledge Graphs (KGs) to enable resilient task planning and autonomous failure recovery. While FMs offer remarkable generalization and contextual reasoning, their limitations, including computational inefficiency, hallucinations, and output inconsistencies hinder reliable deployment. STAR mitigates these issues by embedding learned knowledge into structured, reusable KGs, which streamline information retrieval, reduce redundant FM computations, and provide precise, scenario-specific insights. The framework leverages FM-driven reasoning to diagnose failures, generate context-aware recovery strategies, and execute corrective actions without human intervention or system restarts. Unlike conventional approaches that rely on rigid protocols, STAR dynamically expands its KG with experiential knowledge, ensuring continuous adaptation to novel scenarios. To evaluate the effectiveness of this approach, we developed a comprehensive dataset that includes various robotic tasks and failure scenarios. Through extensive experimentation, STAR demonstrated an 86% task planning accuracy and 78% recovery success rate, showing significant improvements over baseline methods. The framework's ability to continuously learn from experience while maintaining structured knowledge representation makes it particularly suitable for long-term deployment in real-world applications.
Authors: Xiang Lan, Feng Wu, Kai He, Qinghao Zhao, Shenda Hong, Mengling Feng
Abstract: While recent multimodal large language models (MLLMs) have advanced automated ECG interpretation, they still face two key limitations: (1) insufficient multimodal synergy between time series signals and visual ECG representations, and (2) limited explainability in linking diagnoses to granular waveform evidence. We introduce GEM, the first MLLM unifying ECG time series, 12-lead ECG images and text for grounded and clinician-aligned ECG interpretation. GEM enables feature-grounded analysis, evidence-driven reasoning, and a clinician-like diagnostic process through three core innovations: a dual-encoder framework extracting complementary time series and image features, cross-modal alignment for effective multimodal understanding, and knowledge-guided instruction generation for generating high-granularity grounding data (ECG-Grounding) linking diagnoses to measurable parameters ($e.g.$, QRS/PR Intervals). Additionally, we propose the Grounded ECG Understanding task, a clinically motivated benchmark designed to comprehensively assess the MLLM's capability in grounded ECG understanding. Experimental results on both existing and our proposed benchmarks show GEM significantly improves predictive performance (CSN $7.4\% \uparrow$), explainability ($22.7\% \uparrow$), and grounding ($24.8\% \uparrow$), making it more suitable for real-world clinical applications. GitHub repository: https://github.com/lanxiang1017/GEM.git
Authors: Qi Zhang, Xiuyuan Chen, Ziyi He, Lianming Wu, Kun Wang, Jianqi Sun, Hongxing Shen
Abstract: Cervical spondylosis, a complex and prevalent condition, demands precise and efficient diagnostic techniques for accurate assessment. While MRI offers detailed visualization of cervical spine anatomy, manual interpretation remains labor-intensive and prone to error. To address this, we developed an innovative AI-assisted Expert-based Diagnosis System that automates both segmentation and diagnosis of cervical spondylosis using MRI. Leveraging a dataset of 960 cervical MRI images from patients with cervical disc herniation, our system features a pathology-guided segmentation model capable of accurately segmenting key cervical anatomical structures. The segmentation is followed by an expert-based diagnostic framework that automates the calculation of critical clinical indicators. Our segmentation model achieved an impressive average Dice coefficient exceeding 0.90 across four cervical spinal anatomies and demonstrated enhanced accuracy in herniation areas. Diagnostic evaluation further showcased the system precision, with a mean absolute error (MAE) of 2.44 degree for the C2-C7 Cobb angle and 3.60 precentage for the Maximum Spinal Cord Compression (MSCC) coefficient. In addition, our method delivered high accuracy, precision, recall, and F1 scores in herniation localization, K-line status assessment, and T2 hyperintensity detection. Comparative analysis demonstrates that our system outperforms existing methods, establishing a new benchmark for segmentation and diagnostic tasks for cervical spondylosis.
Authors: Kai Yang, Zijian Bai, Yang Xiao, Xinyu Li, Xiaohan Shi
Abstract: 3D reconstruction garners increasing attention alongside the advancement of high-level image applications, where dense stereo matching (DSM) serves as a pivotal technique. Previous studies often rely on publicly available datasets for training, focusing on modifying network architectures or incorporating specialized modules to extract domain-invariant features and thus improve model robustness. In contrast, inspired by single-frame structured-light phase-shifting encoding, this study introduces RGB-Speckle, a cross-scene 3D reconstruction framework based on an active stereo camera system, designed to enhance robustness. Specifically, we propose a novel phase pre-normalization encoding-decoding method: first, we randomly perturb phase-shift maps and embed them into the three RGB channels to generate color speckle patterns; subsequently, the camera captures phase-encoded images modulated by objects as input to a stereo matching network. This technique effectively mitigates external interference and ensures consistent input data for RGB-Speckle, thereby bolstering cross-domain 3D reconstruction stability. To validate the proposed method, we conduct complex experiments: (1) construct a color speckle dataset for complex scenarios based on the proposed encoding scheme; (2) evaluate the impact of the phase pre-normalization encoding-decoding technique on 3D reconstruction accuracy; and (3) further investigate its robustness across diverse conditions. Experimental results demonstrate that the proposed RGB-Speckle model offers significant advantages in cross-domain and cross-scene 3D reconstruction tasks, enhancing model generalization and reinforcing robustness in challenging environments, thus providing a novel solution for robust 3D reconstruction research.
Authors: Haotong Yang, Qingyuan Zheng, Yunjian Gao, Yongkun Yang, Yangbo He, Zhouchen Lin, Muhan Zhang
Abstract: With the rapid advancement of text-conditioned Video Generation Models (VGMs), the quality of generated videos has significantly improved, bringing these models closer to functioning as ``*world simulators*'' and making real-world-level video generation more accessible and cost-effective. However, the generated videos often contain factual inaccuracies and lack understanding of fundamental physical laws. While some previous studies have highlighted this issue in limited domains through manual analysis, a comprehensive solution has not yet been established, primarily due to the absence of a generalized, automated approach for modeling and assessing the causal reasoning of these models across diverse scenarios. To address this gap, we propose VACT: an **automated** framework for modeling, evaluating, and measuring the causal understanding of VGMs in real-world scenarios. By combining causal analysis techniques with a carefully designed large language model assistant, our system can assess the causal behavior of models in various contexts without human annotation, which offers strong generalization and scalability. Additionally, we introduce multi-level causal evaluation metrics to provide a detailed analysis of the causal performance of VGMs. As a demonstration, we use our framework to benchmark several prevailing VGMs, offering insight into their causal reasoning capabilities. Our work lays the foundation for systematically addressing the causal understanding deficiencies in VGMs and contributes to advancing their reliability and real-world applicability.
Authors: Youngjoon Jeong, Junha Chun, Soonwoo Cha, Taesup Kim
Abstract: A world model is essential for an agent to predict the future and plan in domains such as autonomous driving and robotics. To achieve this, recent advancements have focused on video generation, which has gained significant attention due to the impressive success of diffusion models. However, these models require substantial computational resources. To address these challenges, we propose a world model leveraging object-centric representation space using slot attention, guided by language instructions. Our model perceives the current state as an object-centric representation and predicts future states in this representation space conditioned on natural language instructions. This approach results in a more compact and computationally efficient model compared to diffusion-based generative alternatives. Furthermore, it flexibly predicts future states based on language instructions, and offers a significant advantage in manipulation tasks where object recognition is crucial. In this paper, we demonstrate that our latent predictive world model surpasses generative world models in visuo-linguo-motor control tasks, achieving superior sample and computation efficiency. We also investigate the generalization performance of the proposed method and explore various strategies for predicting actions using object-centric representations.
Authors: YingLiang Ma, Sandra Howell, Aldo Rinaldi, Tarv Dhanjal, Kawal S. Rhode
Abstract: Objective: Interventional devices, catheters and insertable imaging devices such as transesophageal echo (TOE) probes are routinely used in minimally invasive cardiovascular procedures. Detecting their positions and orientations in X-ray fluoroscopic images is important for many clinical applications. Method: In this paper, a novel attention mechanism was designed to guide a convolution neural network (CNN) model to the areas of wires in X-ray images, as nearly all interventional devices and catheters used in cardiovascular procedures contain wires. The attention mechanism includes multi-scale Gaussian derivative filters and a dot-product-based attention layer. By utilizing the proposed attention mechanism, a lightweight foundation model can be created to detect multiple objects simultaneously with higher precision and real-time speed. Results: The proposed model was trained and tested on a total of 12,438 X-ray images. An accuracy of 0.88 was achieved for detecting an echo probe and 0.87 for detecting an artificial valve at 58 FPS. The accuracy was measured by intersection-over-union (IoU). We also achieved a 99.8% success rate in detecting a 10-electrode catheter and a 97.8% success rate in detecting an ablation catheter. Conclusion: Our detection foundation model can simultaneously detect and identify both interventional devices and flexible catheters in real-time X-ray fluoroscopic images. Significance: The proposed model employs a novel attention mechanism to achieve high-performance object detection, making it suitable for various clinical applications and robotic-assisted surgeries. Codes are available at https://github.com/YingLiangMa/AttWire.
Authors: Yanjun Chen, Yirong Sun, Xinghao Chen, Jian Wang, Xiaoyu Shen, Wenjie Li, Wei Zhang
Abstract: Chain-of-Thought (CoT) reasoning has proven effective in natural language tasks but remains underexplored in multimodal alignment. This study investigates its integration into 3D vision-language learning by embedding structured reasoning into alignment training. We introduce the 3D-CoT Benchmark, a dataset with hierarchical CoT annotations covering shape recognition, functional inference, and causal reasoning. Through controlled experiments, we compare CoT-structured and standard textual annotations across large reasoning models (LRMs) and large language models (LLMs). Our evaluation employs a dual-layer framework assessing both intermediate reasoning and final inference quality. Extensive experiments demonstrate that CoT significantly improves 3D semantic grounding, with LRMs leveraging CoT more effectively than LLMs. Furthermore, we highlight that annotation structure influences performance-explicit reasoning markers aid LLMs, while unmarked CoT better aligns with LRM inference patterns. Our analyses suggest that CoT is crucial for enhancing multimodal reasoning, with implications beyond 3D tasks.
Authors: Md Ohiduzzaman Ovi, Maliha Sanjana, Fahad Fahad, Mahjabin Runa, Zarin Tasnim Rothy, Tanmoy Sarkar Pias, A. M. Tayeful Islam, Rumman Ahmed Prodhan
Abstract: Pediatric dental segmentation is critical in dental diagnostics, presenting unique challenges due to variations in dental structures and the lower number of pediatric X-ray images. This study proposes a custom SegUNet model with a VGG19 backbone, designed explicitly for pediatric dental segmentation and applied to the Children's Dental Panoramic Radiographs dataset. The SegUNet architecture with a VGG19 backbone has been employed on this dataset for the first time, achieving state-of-the-art performance. The model reached an accuracy of 97.53%, a dice coefficient of 92.49%, and an intersection over union (IOU) of 91.46%, setting a new benchmark for this dataset. These results demonstrate the effectiveness of the VGG19 backbone in enhancing feature extraction and improving segmentation precision. Comprehensive evaluations across metrics, including precision, recall, and specificity, indicate the robustness of this approach. The model's ability to generalize across diverse dental structures makes it a valuable tool for clinical applications in pediatric dental care. It offers a reliable and efficient solution for automated dental diagnostics.
Authors: Gaurav Patel, Qiang Qiu
Abstract: Machine Unlearning has recently garnered significant attention, aiming to selectively remove knowledge associated with specific data while preserving the model's performance on the remaining data. A fundamental challenge in this process is balancing effective unlearning with knowledge retention, as naive optimization of these competing objectives can lead to conflicting gradients, hindering convergence and degrading overall performance. To address this issue, we propose Learning to Unlearn while Retaining, aimed to mitigate gradient conflicts between unlearning and retention objectives. Our approach strategically avoids conflicts through an implicit gradient regularization mechanism that emerges naturally within the proposed framework. This prevents conflicting gradients between unlearning and retention, leading to effective unlearning while preserving the model's utility. We validate our approach across both discriminative and generative tasks, demonstrating its effectiveness in achieving unlearning without compromising performance on remaining data. Our results highlight the advantages of avoiding such gradient conflicts, outperforming existing methods that fail to account for these interactions.
Authors: Guofeng Zhang, Ruyi Zha, Hao He, Yixun Liang, Alan Yuille, Hongdong Li, Yuanhao Cai
Abstract: Sparse-view 3D CT reconstruction aims to recover volumetric structures from a limited number of 2D X-ray projections. Existing feedforward methods are constrained by the limited capacity of CNN-based architectures and the scarcity of large-scale training datasets. In this paper, we propose an X-ray Large Reconstruction Model (X-LRM) for extremely sparse-view (<10 views) CT reconstruction. X-LRM consists of two key components: X-former and X-triplane. Our X-former can handle an arbitrary number of input views using an MLP-based image tokenizer and a Transformer-based encoder. The output tokens are then upsampled into our X-triplane representation, which models the 3D radiodensity as an implicit neural field. To support the training of X-LRM, we introduce Torso-16K, a large-scale dataset comprising over 16K volume-projection pairs of various torso organs. Extensive experiments demonstrate that X-LRM outperforms the state-of-the-art method by 1.5 dB and achieves 27x faster speed and better flexibility. Furthermore, the downstream evaluation of lung segmentation tasks also suggests the practical value of our approach. Our code, pre-trained models, and dataset will be released at https://github.com/caiyuanhao1998/X-LRM
Authors: Md Yousuf Harun, Christopher Kanan
Abstract: To adapt to real-world data streams, continual learning (CL) systems must rapidly learn new concepts while preserving and utilizing prior knowledge. When it comes to adding new information to continually-trained deep neural networks (DNNs), classifier weights for newly encountered categories are typically initialized randomly, leading to high initial training loss (spikes) and instability. Consequently, achieving optimal convergence and accuracy requires prolonged training, increasing computational costs. Inspired by Neural Collapse (NC), we propose a weight initialization strategy to improve learning efficiency in CL. In DNNs trained with mean-squared-error, NC gives rise to a Least-Square (LS) classifier in the last layer, whose weights can be analytically derived from learned features. We leverage this LS formulation to initialize classifier weights in a data-driven manner, aligning them with the feature distribution rather than using random initialization. Our method mitigates initial loss spikes and accelerates adaptation to new tasks. We evaluate our approach in large-scale CL settings, demonstrating faster adaptation and improved CL performance.
Authors: Zihao Peng, Xijun Wang, Shengbo Chen, Hong Rao, Cong Shen
Abstract: Diffusion models are powerful generative models that can produce highly realistic samples for various tasks. Typically, these models are constructed using centralized, independently and identically distributed (IID) training data. However, in practical scenarios, data is often distributed across multiple clients and frequently manifests non-IID characteristics. Federated Learning (FL) can leverage this distributed data to train diffusion models, but the performance of existing FL methods is unsatisfactory in non-IID scenarios. To address this, we propose FedDDPM-Federated Learning with Denoising Diffusion Probabilistic Models, which leverages the data generative capability of diffusion models to facilitate model training. In particular, the server uses well-trained local diffusion models uploaded by each client before FL training to generate auxiliary data that can approximately represent the global data distribution. Following each round of model aggregation, the server further optimizes the global model using the auxiliary dataset to alleviate the impact of heterogeneous data on model performance. We provide a rigorous convergence analysis of FedDDPM and propose an enhanced algorithm, FedDDPM+, to reduce training overheads. FedDDPM+ detects instances of slow model learning and performs a one-shot correction using the auxiliary dataset. Experimental results validate that our proposed algorithms outperform the state-of-the-art FL algorithms on the MNIST, CIFAR10 and CIFAR100 datasets.
Authors: Yu Jin, Jingming Liu, Zhexu Luo, Yifei Peng, Ziang Qin, Wang-Zhou Dai, Yao-Xiang Ding, Kun Zhou
Abstract: Visual generative abductive learning studies jointly training symbol-grounded neural visual generator and inducing logic rules from data, such that after learning, the visual generation process is guided by the induced logic rules. A major challenge for this task is to reduce the time cost of logic abduction during learning, an essential step when the logic symbol set is large and the logic rule to induce is complicated. To address this challenge, we propose a pre-training method for obtaining meta-rule selection policy for the recently proposed visual generative learning approach AbdGen [Peng et al., 2023], aiming at significantly reducing the candidate meta-rule set and pruning the search space. The selection model is built based on the embedding representation of both symbol grounding of cases and meta-rules, which can be effectively integrated with both neural model and logic reasoning system. The pre-training process is done on pure symbol data, not involving symbol grounding learning of raw visual inputs, making the entire learning process low-cost. An additional interesting observation is that the selection policy can rectify symbol grounding errors unseen during pre-training, which is resulted from the memorization ability of attention mechanism and the relative stability of symbolic patterns. Experimental results show that our method is able to effectively address the meta-rule selection problem for visual abduction, boosting the efficiency of visual generative abductive learning. Code is available at https://github.com/future-item/metarule-select.
Authors: Fei Tang, Yongliang Shen, Hang Zhang, Siqi Chen, Guiyang Hou, Wenqi Zhang, Wenqiao Zhang, Kaitao Song, Weiming Lu, Yueting Zhuang
Abstract: Humans can flexibly switch between different modes of thinking based on task complexity: from rapid intuitive judgments to in-depth analytical understanding. However, current Graphical User Interface (GUI) grounding systems which locate interface elements based on natural language instructions rely solely on immediate prediction without reasoning, struggling to understand complex interface layouts with nested structures and hierarchical relationships, limiting their effectiveness on complex interfaces. Inspired by human dual-system cognition, we present Focus, a novel GUI grounding framework that combines fast prediction with systematic analysis. The framework dynamically switches between rapid and deliberate processing through an adaptive system switching based on task complexity, optimizing both efficiency and accuracy. Focus decomposes grounding into progressive stages: interface summarization, visual focused analysis, and precise coordinate prediction. This structured decomposition enables systematic understanding of both interface layouts and visual relationships. Extensive experiments show that Focus achieves state-of-the-art performance using only 300K of the training data with a 2B parameter model compared to existing approaches. Focus demonstrates superior performance particularly in complex GUI scenarios, achieving 77.4% average accuracy on ScreenSpot and 13.3% on the more challenging ScreenSpot-Pro. Our analysis reveals the effectiveness of this dual-system approach while demonstrating its potential for improving complex GUI interaction scenarios.
Authors: Yanling Wang, Yihan Zhao, Xiaodong Chen, Shasha Guo, Lixin Liu, Haoyang Li, Yong Xiao, Jing Zhang, Qi Li, Ke Xu
Abstract: Large vision-language models (LVLMs) have demonstrated remarkable achievements, yet the generation of non-factual responses remains prevalent in fact-seeking question answering (QA). Current multimodal fact-seeking benchmarks primarily focus on comparing model outputs to ground truth answers, providing limited insights into the performance of modality-specific modules. To bridge this gap, we introduce VisualSimpleQA, a multimodal fact-seeking benchmark with two key features. First, it enables streamlined and decoupled evaluation of LVLMs in visual and linguistic modalities. Second, it incorporates well-defined difficulty criteria to guide human annotation and facilitates the extraction of a challenging subset, VisualSimpleQA-hard. Experiments on 15 LVLMs show that even state-of-the-art models such as GPT-4o achieve merely 60%+ correctness in multimodal fact-seeking QA on VisualSimpleQA and 30%+ on VisualSimpleQA-hard. Furthermore, the decoupled evaluation across these models highlights substantial opportunities for improvement in both visual and linguistic modules. The dataset is available at https://huggingface.co/datasets/WYLing/VisualSimpleQA.
URLs: https://huggingface.co/datasets/WYLing/VisualSimpleQA.
Authors: Haoqiang Kang, Enna Sachdeva, Piyush Gupta, Sangjae Bae, Kwonjoon Lee
Abstract: Vision-Language Models (VLMs) have recently shown promising advancements in sequential decision-making tasks through task-specific fine-tuning. However, common fine-tuning methods, such as Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) techniques like Proximal Policy Optimization (PPO), present notable limitations: SFT assumes Independent and Identically Distributed (IID) data, while PPO focuses on maximizing cumulative rewards. These limitations often restrict solution diversity and hinder generalization in multi-step reasoning tasks. To address these challenges, we introduce a novel framework, GFlowVLM, a framework that fine-tune VLMs using Generative Flow Networks (GFlowNets) to promote generation of diverse solutions for complex reasoning tasks. GFlowVLM models the environment as a non-Markovian decision process, allowing it to capture long-term dependencies essential for real-world applications. It takes observations and task descriptions as inputs to prompt chain-of-thought (CoT) reasoning which subsequently guides action selection. We use task based rewards to fine-tune VLM with GFlowNets. This approach enables VLMs to outperform prior fine-tuning methods, including SFT and RL. Empirical results demonstrate the effectiveness of GFlowVLM on complex tasks such as card games (NumberLine, BlackJack) and embodied planning tasks (ALFWorld), showing enhanced training efficiency, solution diversity, and stronger generalization capabilities across both in-distribution and out-of-distribution scenarios.
Authors: Jialin Lu, Junjie Shan, Ziqi Zhao, Ka-Ho Chow
Abstract: As object detection becomes integral to many safety-critical applications, understanding its vulnerabilities is essential. Backdoor attacks, in particular, pose a serious threat by implanting hidden triggers in victim models, which adversaries can later exploit to induce malicious behaviors during inference. However, current understanding is limited to single-target attacks, where adversaries must define a fixed malicious behavior (target) before training, making inference-time adaptability impossible. Given the large output space of object detection (including object existence prediction, bounding box estimation, and classification), the feasibility of flexible, inference-time model control remains unexplored. This paper introduces AnywhereDoor, a multi-target backdoor attack for object detection. Once implanted, AnywhereDoor allows adversaries to make objects disappear, fabricate new ones, or mislabel them, either across all object classes or specific ones, offering an unprecedented degree of control. This flexibility is enabled by three key innovations: (i) objective disentanglement to scale the number of supported targets; (ii) trigger mosaicking to ensure robustness even against region-based detectors; and (iii) strategic batching to address object-level data imbalances that hinder manipulation. Extensive experiments demonstrate that AnywhereDoor grants attackers a high degree of control, improving attack success rates by 26% compared to adaptations of existing methods for such flexible control.
Authors: Jiaxin Ai, Pengfei Zhou, Zhaopan Xu, Ming Li, Fanrui Zhang, Zizhen Li, Jianwen Sun, Yukang Feng, Baojin Huang, Zhongyuan Wang, Kaipeng Zhang
Abstract: As multi-modal large language models (MLLMs) frequently exhibit errors when solving scientific problems, evaluating the validity of their reasoning processes is critical for ensuring reliability and uncovering fine-grained model weaknesses. Since human evaluation is laborious and costly, prompting MLLMs as automated process judges has become a common practice. However, the reliability of these model-based judges remains uncertain. To address this, we introduce ProJudgeBench, the first comprehensive benchmark specifically designed for evaluating abilities of MLLM-based process judges. ProJudgeBench comprises 2,400 test cases and 50,118 step-level labels, spanning four scientific disciplines with diverse difficulty levels and multi-modal content. In ProJudgeBench, each step is meticulously annotated by human experts for correctness, error type, and explanation, enabling a systematic evaluation of judges' capabilities to detect, classify and diagnose errors. Evaluation on ProJudgeBench reveals a significant performance gap between open-source and proprietary models. To bridge this gap, we further propose ProJudge-173k, a large-scale instruction-tuning dataset, and a Dynamic Dual-Phase fine-tuning strategy that encourages models to explicitly reason through problem-solving before assessing solutions. Both contributions significantly enhance the process evaluation capabilities of open-source models. All the resources will be released to foster future research of reliable multi-modal process evaluation.
Authors: Jiangdong Cai, Haotian Jiang, Zhenrong Shen, Yonghao Li, Honglin Xiong, Lichi Zhang, Qian Wang
Abstract: The deployment of computer-aided diagnosis systems for cervical cancer screening using whole slide images (WSIs) faces critical challenges due to domain shifts caused by staining variations across different scanners and imaging environments. While existing stain augmentation methods improve patch-level robustness, they fail to scale to WSIs due to two key limitations: (1) inconsistent stain patterns when extending patch operations to gigapixel slides, and (2) prohibitive computational/storage costs from offline processing of augmented WSIs.To address this, we propose Latent Style Augmentation (LSA), a framework that performs efficient, online stain augmentation directly on WSI-level latent features. We first introduce WSAug, a WSI-level stain augmentation method ensuring consistent stain across patches within a WSI. Using offline-augmented WSIs by WSAug, we design and train Stain Transformer, which can simulate targeted style in the latent space, efficiently enhancing the robustness of the WSI-level classifier. We validate our method on a multi-scanner WSI dataset for cervical cancer diagnosis. Despite being trained on data from a single scanner, our approach achieves significant performance improvements on out-of-distribution data from other scanners. Code will be available at https://github.com/caijd2000/LSA.
Authors: Yaxuan Li, Han Jiang, Yifei Ma, Shihua Qin, Fangxu Xing
Abstract: Dynamic Magnetic Resonance Imaging (MRI) of the vocal tract has become an increasingly adopted imaging modality for speech motor studies. Beyond image signals, systematic data loss, noise pollution, and audio file corruption can occur due to the unpredictability of the MRI acquisition environment. In such cases, generating audio from images is critical for data recovery in both clinical and research applications. However, this remains challenging due to hardware constraints, acoustic interference, and data corruption. Existing solutions, such as denoising and multi-stage synthesis methods, face limitations in audio fidelity and generalizability. To address these challenges, we propose a Knowledge Enhanced Conditional Variational Autoencoder (KE-CVAE), a novel two-step "knowledge enhancement + variational inference" framework for generating speech audio signals from cine dynamic MRI sequences. This approach introduces two key innovations: (1) integration of unlabeled MRI data for knowledge enhancement, and (2) a variational inference architecture to improve generative modeling capacity. To the best of our knowledge, this is one of the first attempts at synthesizing speech audio directly from dynamic MRI video sequences. The proposed method was trained and evaluated on an open-source dynamic vocal tract MRI dataset recorded during speech. Experimental results demonstrate its effectiveness in generating natural speech waveforms while addressing MRI-specific acoustic challenges, outperforming conventional deep learning-based synthesis approaches.
Authors: AgiBot-World-Contributors, Qingwen Bu, Jisong Cai, Li Chen, Xiuqi Cui, Yan Ding, Siyuan Feng, Shenyuan Gao, Xindong He, Xu Huang, Shu Jiang, Yuxin Jiang, Cheng Jing, Hongyang Li, Jialu Li, Chiming Liu, Yi Liu, Yuxiang Lu, Jianlan Luo, Ping Luo, Yao Mu, Yuehan Niu, Yixuan Pan, Jiangmiao Pang, Yu Qiao, Guanghui Ren, Cheng Ruan, Jiaqi Shan, Yongjian Shen, Chengshi Shi, Mingkang Shi, Modi Shi, Chonghao Sima, Jianheng Song, Huijie Wang, Wenhao Wang, Dafeng Wei, Chengen Xie, Guo Xu, Junchi Yan, Cunbiao Yang, Lei Yang, Shukai Yang, Maoqing Yao, Jia Zeng, Chi Zhang, Qinglin Zhang, Bin Zhao, Chengyue Zhao, Jiaqi Zhao, Jianchao Zhu
Abstract: We explore how scalable robot data can address real-world challenges for generalized robotic manipulation. Introducing AgiBot World, a large-scale platform comprising over 1 million trajectories across 217 tasks in five deployment scenarios, we achieve an order-of-magnitude increase in data scale compared to existing datasets. Accelerated by a standardized collection pipeline with human-in-the-loop verification, AgiBot World guarantees high-quality and diverse data distribution. It is extensible from grippers to dexterous hands and visuo-tactile sensors for fine-grained skill acquisition. Building on top of data, we introduce Genie Operator-1 (GO-1), a novel generalist policy that leverages latent action representations to maximize data utilization, demonstrating predictable performance scaling with increased data volume. Policies pre-trained on our dataset achieve an average performance improvement of 30% over those trained on Open X-Embodiment, both in in-domain and out-of-distribution scenarios. GO-1 exhibits exceptional capability in real-world dexterous and long-horizon tasks, achieving over 60% success rate on complex tasks and outperforming prior RDT approach by 32%. By open-sourcing the dataset, tools, and models, we aim to democratize access to large-scale, high-quality robot data, advancing the pursuit of scalable and general-purpose intelligence.
Authors: Sheng Song, Yiting Chen, Duo Xu, Songhan Ge, Yunqian Huang, Junni Shi, Man Chen, Hongbo Chen, Rui Zheng
Abstract: Freehand 3D ultrasound enables volumetric imaging by tracking a conventional ultrasound probe during freehand scanning, offering enriched spatial information that improves clinical diagnosis. However, the quality of reconstructed volumes is often compromised by tracking system noise and irregular probe movements, leading to artifacts in the final reconstruction. To address these challenges, we propose ImplicitCell, a novel framework that integrates Implicit Neural Representation (INR) with an ultrasound resolution cell model for joint optimization of volume reconstruction and pose refinement. Three distinct datasets are used for comprehensive validation, including phantom, common carotid artery, and carotid atherosclerosis. Experimental results demonstrate that ImplicitCell significantly reduces reconstruction artifacts and improves volume quality compared to existing methods, particularly in challenging scenarios with noisy tracking data. These improvements enhance the clinical utility of freehand 3D ultrasound by providing more reliable and precise diagnostic information.
Authors: Xavier Thomas, Deepti Ghadiyaram
Abstract: Domain Generalization aims to develop models that can generalize to novel and unseen data distributions. In this work, we study how model architectures and pre-training objectives impact feature richness and propose a method to effectively leverage them for domain generalization. Specifically, given a pre-trained feature space, we first discover latent domain structures, referred to as pseudo-domains, that capture domain-specific variations in an unsupervised manner. Next, we augment existing classifiers with these complementary pseudo-domain representations making them more amenable to diverse unseen test domains. We analyze how different pre-training feature spaces differ in the domain-specific variances they capture. Our empirical studies reveal that features from diffusion models excel at separating domains in the absence of explicit domain labels and capture nuanced domain-specific information. On 5 datasets, we show that our very simple framework improves generalization to unseen domains by a maximum test accuracy improvement of over 4% compared to the standard baseline Empirical Risk Minimization (ERM). Crucially, our method outperforms most algorithms that access domain labels during training.
Authors: Junhao Cao, Nicolas Folastre, Gozde Oney, Edgar Rauch, Stavros Nicolopoulos, Partha Pratim Das, Arnaud Demorti\`ere
Abstract: This study presents a novel integration of unsupervised learning and decision-making strategies for the advanced analysis of 4D-STEM datasets, with a focus on non-negative matrix factorization (NMF) as the primary clustering method. Our approach introduces a systematic framework to determine the optimal number of components (k) required for robust and interpretable orientation mapping. By leveraging the K-Component Loss method and Image Quality Assessment (IQA) metrics, we effectively balance reconstruction fidelity and model complexity. Additionally, we highlight the critical role of dataset preprocessing in improving clustering stability and accuracy. Furthermore, our spatial weight matrix analysis provides insights into overlapping regions within the dataset by employing threshold-based visualization, facilitating a detailed understanding of cluster interactions. The results demonstrate the potential of combining NMF with advanced IQA metrics and preprocessing techniques for reliable orientation mapping and structural analysis in 4D-STEM datasets, paving the way for future applications in multi-dimensional material characterization.
Authors: Cheng Huang, Weizheng Xie, Tsengdar J. Lee, Jui-Kai Wang, Karanjit Kooner, Jia Zhang
Abstract: Structural changes in main retinal blood vessels serve as critical biomarkers for the onset and progression of glaucoma. Identifying these vessels is vital for vascular modeling yet highly challenging. This paper proposes X-GAN, a generative AI-powered unsupervised segmentation model designed for extracting main blood vessels from Optical Coherence Tomography Angiography (OCTA) images. The process begins with the Space Colonization Algorithm (SCA) to rapidly generate a skeleton of vessels, featuring their radii. By synergistically integrating generative adversarial networks (GANs) with biostatistical modeling of vessel radii, X-GAN enables a fast reconstruction of both 2D and 3D representations of the vessels. Based on this reconstruction, X-GAN achieves nearly 100\% segmentation accuracy without relying on labeled data or high-performance computing resources. Also, to address the Issue, data scarity, we introduce GSS-RetVein, a high-definition mixed 2D and 3D glaucoma retinal dataset. GSS-RetVein provides a rigorous benchmark due to its exceptionally clear capillary structures, introducing controlled noise for testing model robustness. Its 2D images feature sharp capillary boundaries, while its 3D component enhances vascular reconstruction and blood flow prediction, supporting glaucoma progression simulations. Experimental results confirm GSS-RetVein's superiority in evaluating main vessel segmentation compared to existing datasets. Code and dataset are here: https://github.com/VikiXie/SatMar8.
Authors: Tianyi Zhang, Weiming Zhi, Joshua Mangelson, Matthew Johnson-Roberson
Abstract: This paper tackles the problem of generating representations of underwater 3D terrain. Off-the-shelf generative models, trained on Internet-scale data but not on specialized underwater images, exhibit downgraded realism, as images of the seafloor are relatively uncommon. To this end, we introduce DreamSea, a generative model to generate hyper-realistic underwater scenes. DreamSea is trained on real-world image databases collected from underwater robot surveys. Images from these surveys contain massive real seafloor observations and covering large areas, but are prone to noise and artifacts from the real world. We extract 3D geometry and semantics from the data with visual foundation models, and train a diffusion model that generates realistic seafloor images in RGBD channels, conditioned on novel fractal distribution-based latent embeddings. We then fuse the generated images into a 3D map, building a 3DGS model supervised by 2D diffusion priors which allows photorealistic novel view rendering. DreamSea is rigorously evaluated, demonstrating the ability to robustly generate large-scale underwater scenes that are consistent, diverse, and photorealistic. Our work drives impact in multiple domains, spanning filming, gaming, and robot simulation.
Authors: Lidia Al-Zogbi, Deepak Raina, Vinciya Pandian, Thorsten Fleiter, Axel Krieger
Abstract: Femoral artery access is essential for numerous clinical procedures, including diagnostic angiography, therapeutic catheterization, and emergency interventions. Despite its critical role, successful vascular access remains challenging due to anatomical variability, overlying adipose tissue, and the need for precise ultrasound (US) guidance. Errors in needle placement can lead to severe complications, restricting the procedure to highly skilled clinicians in controlled hospital settings. While robotic systems have shown promise in addressing these challenges through autonomous scanning and vessel reconstruction, clinical translation remains limited due to reliance on simplified phantom models that fail to capture human anatomical complexity. In this work, we present a method for autonomous robotic US scanning of bifurcated femoral arteries, and validate it on five vascular phantoms created from real patient computed tomography (CT) data. Additionally, we introduce a video-based deep learning US segmentation network tailored for vascular imaging, enabling improved 3D arterial reconstruction. The proposed network achieves a Dice score of 89.21% and an Intersection over Union of 80.54% on a newly developed vascular dataset. The quality of the reconstructed artery centerline is evaluated against ground truth CT data, demonstrating an average L2 deviation of 0.91+/-0.70 mm, with an average Hausdorff distance of 4.36+/-1.11mm. This study is the first to validate an autonomous robotic system for US scanning of the femoral artery on a diverse set of patient-specific phantoms, introducing a more advanced framework for evaluating robotic performance in vascular imaging and intervention.
Authors: Gexin Huang, Ruinan Jin, Yucheng Tang, Can Zhao, Tatsuya Harada, Xiaoxiao Li, Gu Lin
Abstract: Accurately visualizing and editing tumor progression in medical imaging is crucial for diagnosis, treatment planning, and clinical communication. To address the challenges of subjectivity and limited precision in existing methods, we propose SkEditTumor, a sketch-based diffusion model for controllable tumor progression editing. By leveraging sketches as structural priors, our method enables precise modifications of tumor regions while maintaining structural integrity and visual realism. We evaluate SkEditTumor on four public datasets - BraTS, LiTS, KiTS, and MSD-Pancreas - covering diverse organs and imaging modalities. Experimental results demonstrate that our method outperforms state-of-the-art baselines, achieving superior image fidelity and segmentation accuracy. Our contributions include a novel integration of sketches with diffusion models for medical image editing, fine-grained control over tumor progression visualization, and extensive validation across multiple datasets, setting a new benchmark in the field.
Authors: Murtaza Dalal
Abstract: How can we build generalist robot systems? Scale may not be enough due to the significant multimodality of robotics tasks, lack of easily accessible data and the challenges of deploying on physical hardware. Meanwhile, most deployed robotic systems today are inherently modular and can leverage the independent generalization capabilities of each module to perform well. Therefore, this thesis seeks to tackle the task of building generalist robot agents by integrating these components into one: combining modularity with large-scale learning for general purpose robot control. The first question we consider is: how can we build modularity and hierarchy into learning systems? Our key insight is that rather than having the agent learn hierarchy and low-level control end-to-end, we can enforce modularity via planning to enable more efficient and capable robot learners. Next, we come to the role of scale in building generalist robot systems. To scale, neural networks require vast amounts of diverse data, expressive architectures to fit the data and a source of supervision to generate the data. We leverage a powerful supervision source: classical planning, which can generalize, but is expensive to run and requires access to privileged information to perform well in practice. We use these planners to supervise large-scale policy learning in simulation to produce generalist agents. Finally, we consider how to unify modularity with large-scale policy learning to build real-world robot systems capable of performing zero-shot manipulation. We do so by tightly integrating key ingredients of modular high and mid-level planning, learned local control, procedural scene generation and large-scale policy learning for sim2real transfer. We demonstrate that this recipe can produce a single, generalist agent that can solve challenging long-horizon manipulation tasks in the real world.
Authors: Yuchen Mao, Hongwei Li, Yinyi Lai, Giorgos Papanastasiou, Peng Qi, Yunjie Yang, Chengjia Wang
Abstract: Large-scale vision models like SAM have extensive visual knowledge, yet their general nature and computational demands limit their use in specialized tasks like medical image segmentation. In contrast, task-specific models such as U-Net++ often underperform due to sparse labeled data. This study introduces a strategic knowledge mining method that leverages SAM's broad understanding to boost the performance of small, locally hosted deep learning models. In our approach, we trained a U-Net++ model on a limited labeled dataset and extend its capabilities by converting SAM's output infered on unlabeled images into prompts. This process not only harnesses SAM's generalized visual knowledge but also iteratively improves SAM's prediction to cater specialized medical segmentation tasks via U-Net++. The mined knowledge, serving as "pseudo labels", enriches the training dataset, enabling the fine-tuning of the local network. Applied to the Kvasir SEG and COVID-QU-Ex datasets which consist of gastrointestinal polyp and lung X-ray images respectively, our proposed method consistently enhanced the segmentation performance on Dice by 3% and 1% respectively over the baseline U-Net++ model, when the same amount of labelled data were used during training (75% and 50% of labelled data). Remarkably, our proposed method surpassed the baseline U-Net++ model even when the latter was trained exclusively on labeled data (100% of labelled data). These results underscore the potential of knowledge mining to overcome data limitations in specialized models by leveraging the broad, albeit general, knowledge of large-scale models like SAM, all while maintaining operational efficiency essential for clinical applications.
Authors: S M A Sharif, Rizwan Ali Naqvi, Woong-Kee Loh
Abstract: Medical image denoising is considered among the most challenging vision tasks. Despite the real-world implications, existing denoising methods have notable drawbacks as they often generate visual artifacts when applied to heterogeneous medical images. This study addresses the limitation of the contemporary denoising methods with an artificial intelligence (AI)-driven two-stage learning strategy. The proposed method learns to estimate the residual noise from the noisy images. Later, it incorporates a novel noise attention mechanism to correlate estimated residual noise with noisy inputs to perform denoising in a course-to-refine manner. This study also proposes to leverage a multi-modal learning strategy to generalize the denoising among medical image modalities and multiple noise patterns for widespread applications. The practicability of the proposed method has been evaluated with dense experiments. The experimental results demonstrated that the proposed method achieved state-of-the-art performance by significantly outperforming the existing medical image denoising methods in quantitative and qualitative comparisons. Overall, it illustrates a performance gain of 7.64 in Peak Signal-to-Noise Ratio (PSNR), 0.1021 in Structural Similarity Index (SSIM), 0.80 in DeltaE ($\Delta E$), 0.1855 in Visual Information Fidelity Pixel-wise (VIFP), and 18.54 in Mean Squared Error (MSE) metrics.
Authors: Somayeh Farahani, Marjaneh Hejazi, Antonio Di Ieva, Emad Fatemizadeh, Sidong Liu
Abstract: Accurate, noninvasive glioma characterization is crucial for effective clinical management. Traditional methods, dependent on invasive tissue sampling, often fail to capture the spatial heterogeneity of the tumor. While deep learning has improved segmentation and molecular profiling, few approaches simultaneously integrate tumor morphology and molecular features. Foundation deep learning models, which learn robust, task-agnostic representations from large-scale datasets, hold great promise but remain underutilized in glioma imaging biomarkers. We propose the Multi-Task SWIN-UNETR (MTS-UNET) model, a novel foundation-based framework built on the BrainSegFounder model, pretrained on large-scale neuroimaging data. MTS-UNET simultaneously performs glioma segmentation, histological grading, and molecular subtyping (IDH mutation and 1p/19q co-deletion). It incorporates two key modules: Tumor-Aware Feature Encoding (TAFE) for multi-scale, tumor-focused feature extraction and Cross-Modality Differential (CMD) for highlighting subtle T2-FLAIR mismatch signals associated with IDH mutation. The model was trained and validated on a diverse, multi-center cohort of 2,249 glioma patients from seven public datasets. MTS-UNET achieved a mean Dice score of 84% for segmentation, along with AUCs of 90.58% for IDH mutation, 69.22% for 1p/19q co-deletion prediction, and 87.54% for grading, significantly outperforming baseline models (p<=0.05). Ablation studies validated the essential contributions of the TAFE and CMD modules and demonstrated the robustness of the framework. The foundation-based MTS-UNET model effectively integrates tumor segmentation with multi-level classification, exhibiting strong generalizability across diverse MRI datasets. This framework shows significant potential for advancing noninvasive, personalized glioma management by improving predictive accuracy and interpretability.
Authors: Yilong Wang, Edward Johns
Abstract: We introduce One-Shot Dual-Arm Imitation Learning (ODIL), which enables dual-arm robots to learn precise and coordinated everyday tasks from just a single demonstration of the task. ODIL uses a new three-stage visual servoing (3-VS) method for precise alignment between the end-effector and target object, after which replay of the demonstration trajectory is sufficient to perform the task. This is achieved without requiring prior task or object knowledge, or additional data collection and training following the single demonstration. Furthermore, we propose a new dual-arm coordination paradigm for learning dual-arm tasks from a single demonstration. ODIL was tested on a real-world dual-arm robot, demonstrating state-of-the-art performance across six precise and coordinated tasks in both 4-DoF and 6-DoF settings, and showing robustness in the presence of distractor objects and partial occlusions. Videos are available at: https://www.robot-learning.uk/one-shot-dual-arm.
Authors: Shufang Zhang, Tao Jiang, Jiazheng Wu, Ziyu Meng, Ziyang Zhang, Shan An
Abstract: 3D point cloud mapping plays a essential role in localization and autonomous navigation. However, dynamic objects often leave residual traces during the map construction process, which undermine the performance of subsequent tasks. Therefore, dynamic object removal has become a critical challenge in point cloud based map construction within dynamic scenarios. Existing approaches, however, often incur significant computational overhead, making it difficult to meet the real-time processing requirements. To address this issue, we introduce the Height Interval Filtering (HIF) method. This approach constructs pillar-based height interval representations to probabilistically model the vertical dimension, with interval probabilities updated through Bayesian inference. It ensures real-time performance while achieving high accuracy and improving robustness in complex environments. Additionally, we propose a low-height preservation strategy that enhances the detection of unknown spaces, reducing misclassification in areas blocked by obstacles (occluded regions). Experiments on public datasets demonstrate that HIF delivers a 7.7 times improvement in time efficiency with comparable accuracy to existing SOTA methods. The code will be publicly available.
Authors: Weidong Guo, Hantao Zhang, Shouhong Wan, Bingbing Zou, Wanqin Wang, Chenyang Qiu, Peiquan Jin
Abstract: Lesion synthesis methods have made significant progress in generating large-scale synthetic datasets. However, existing approaches predominantly focus on texture synthesis and often fail to accurately model masks for anatomically complex lesions. Additionally, these methods typically lack precise control over the synthesis process. For example, perirectal lymph nodes, which range in diameter from 1 mm to 10 mm, exhibit irregular and intricate contours that are challenging for current techniques to replicate faithfully. To address these limitations, we introduce CAFusion, a novel approach for synthesizing perirectal lymph nodes. By leveraging Signed Distance Functions (SDF), CAFusion generates highly realistic 3D anatomical structures. Furthermore, it offers flexible control over both anatomical and textural features by decoupling the generation of morphological attributes (such as shape, size, and position) from textural characteristics, including signal intensity. Experimental results demonstrate that our synthetic data substantially improve segmentation performance, achieving a 6.45% increase in the Dice coefficient. In the visual Turing test, experienced radiologists found it challenging to distinguish between synthetic and real lesions, highlighting the high degree of realism and anatomical accuracy achieved by our approach. These findings validate the effectiveness of our method in generating high-quality synthetic lesions for advancing medical image processing applications.
Authors: Junyan Lin, Feng Gap, Lin Qi, Junyu Dong, Qian Du, Xinbo Gao
Abstract: Hyperspectral image (HSI) and LiDAR data joint classification is a challenging task. Existing multi-source remote sensing data classification methods often rely on human-designed frameworks for feature extraction, which heavily depend on expert knowledge. To address these limitations, we propose a novel Dynamic Cross-Modal Feature Interaction Network (DCMNet), the first framework leveraging a dynamic routing mechanism for HSI and LiDAR classification. Specifically, our approach introduces three feature interaction blocks: Bilinear Spatial Attention Block (BSAB), Bilinear Channel Attention Block (BCAB), and Integration Convolutional Block (ICB). These blocks are designed to effectively enhance spatial, spectral, and discriminative feature interactions. A multi-layer routing space with routing gates is designed to determine optimal computational paths, enabling data-dependent feature fusion. Additionally, bilinear attention mechanisms are employed to enhance feature interactions in spatial and channel representations. Extensive experiments on three public HSI and LiDAR datasets demonstrate the superiority of DCMNet over state-of-the-art methods. Our code will be available at https://github.com/oucailab/DCMNet.
Authors: Juncheng Wang, Chao Xu, Cheng Yu, Lei Shang, Zhe Hu, Shujun Wang, Liefeng Bo
Abstract: Video-to-audio generation is essential for synthesizing realistic audio tracks that synchronize effectively with silent videos. Following the perspective of extracting essential signals from videos that can precisely control the mature text-to-audio generative diffusion models, this paper presents how to balance the representation of mel-spectrograms in terms of completeness and complexity through a new approach called Mel Quantization-Continuum Decomposition (Mel-QCD). We decompose the mel-spectrogram into three distinct types of signals, employing quantization or continuity to them, we can effectively predict them from video by a devised video-to-all (V2X) predictor. Then, the predicted signals are recomposed and fed into a ControlNet, along with a textual inversion design, to control the audio generation process. Our proposed Mel-QCD method demonstrates state-of-the-art performance across eight metrics, evaluating dimensions such as quality, synchronization, and semantic consistency. Our codes and demos will be released at \href{Website}{https://wjc2830.github.io/MelQCD/}.
Authors: Wenzhuo Xu, Zhipeng Wei, Xiongtao Sun, Deyue Zhang, Dongdong Yang, Quanchen Zou, Xiangzheng Zhang
Abstract: Recently, Multimodal Large Language Models (MLLMs) have demonstrated their superior ability in understanding multimodal contents. However, they remain vulnerable to jailbreak attacks, which exploit weaknesses in their safety alignment to generate harmful responses. Previous studies categorize jailbreaks as successful or failed based on whether responses contain malicious content. However, given the stochastic nature of MLLM responses, this binary classification of an input's ability to jailbreak MLLMs is inappropriate. Derived from this viewpoint, we introduce jailbreak probability to quantify the jailbreak potential of an input, which represents the likelihood that MLLMs generated a malicious response when prompted with this input. We approximate this probability through multiple queries to MLLMs. After modeling the relationship between input hidden states and their corresponding jailbreak probability using Jailbreak Probability Prediction Network (JPPN), we use continuous jailbreak probability for optimization. Specifically, we propose Jailbreak-Probability-based Attack (JPA) that optimizes adversarial perturbations on inputs to maximize jailbreak probability. To counteract attacks, we also propose two defensive methods: Jailbreak-Probability-based Finetuning (JPF) and Jailbreak-Probability-based Defensive Noise (JPDN), which minimizes jailbreak probability in the MLLM parameters and input space, respectively. Extensive experiments show that (1) JPA yields improvements (up to 28.38\%) under both white and black box settings compared to previous methods with small perturbation bounds and few iterations. (2) JPF and JPDN significantly reduce jailbreaks by at most over 60\%. Both of the above results demonstrate the significance of introducing jailbreak probability to make nuanced distinctions among input jailbreak abilities.
Authors: Yongwoo Kim, Sungmin Cha, Donghyun Kim
Abstract: Machine unlearning is a process to remove specific data points from a trained model while maintaining the performance on retain data, addressing privacy or legal requirements. Despite its importance, existing unlearning evaluations tend to focus on logit-based metrics (i.e., accuracy) under small-scale scenarios. We observe that this could lead to a false sense of security in unlearning approaches under real-world scenarios. In this paper, we conduct a new comprehensive evaluation that employs representation-based evaluations of the unlearned model under large-scale scenarios to verify whether the unlearning approaches genuinely eliminate the targeted forget data from the model's representation perspective. Our analysis reveals that current state-of-the-art unlearning approaches either completely degrade the representational quality of the unlearned model or merely modify the classifier (i.e., the last layer), thereby achieving superior logit-based evaluation metrics while maintaining significant representational similarity to the original model. Furthermore, we introduce a novel unlearning evaluation setup from a transfer learning perspective, in which the forget set classes exhibit semantic similarity to downstream task classes, necessitating that feature representations diverge significantly from those of the original model. Our comprehensive benchmark not only addresses a critical gap between theoretical machine unlearning and practical scenarios, but also establishes a foundation to inspire future research directions in developing genuinely effective unlearning methodologies.
Authors: Shihao Hou, Xinyi Shang, Shreyank N Gowda, Yang Lu, Chao Wu, Yan Yan, Hanzi Wang
Abstract: Effectively handling the co-occurrence of non-IID data and long-tailed distributions remains a critical challenge in federated learning. While fine-tuning vision-language models (VLMs) like CLIP has shown to be promising in addressing non-IID data challenges, this approach leads to severe degradation of tail classes in federated long-tailed scenarios. Under the composite effects of strong non-IID data distribution and long-tailed class imbalances, VLM fine-tuning may even fail to yield any improvement. To address this issue, we propose Class-Aware Prompt Learning for Federated Long-tailed Learning (CAPT), a novel framework that leverages a pre-trained VLM to effectively handle both data heterogeneity and long-tailed distributions. CAPT introduces a dual-prompt mechanism that synergizes general and class-aware prompts, enabling the framework to capture global trends while preserving class-specific knowledge. To better aggregate and share knowledge across clients, we introduce a heterogeneity-aware client clustering strategy that groups clients based on their data distributions, enabling efficient collaboration and knowledge sharing. Extensive experiments on various long-tailed datasets with different levels of data heterogeneity demonstrate that CAPT significantly improves tail class performance without compromising overall accuracy, outperforming state-of-the-art methods in federated long-tailed learning scenarios.
Authors: Zhi Qin, Qianhui Gui, Mouxiao Bian, Rui Wang, Hong Ge, Dandan Yao, Ziying Sun, Yuan Zhao, Yu Zhang, Hui Shi, Dongdong Wang, Chenxin Song, Shenghong Ju, Lihao Liu, Junjun He, Jie Xu, Yuan-Cheng Wang
Abstract: Medical imaging quality control (QC) is essential for accurate diagnosis, yet traditional QC methods remain labor-intensive and subjective. To address this challenge, in this study, we establish a standardized dataset and evaluation framework for medical imaging QC, systematically assessing large language models (LLMs) in image quality assessment and report standardization. Specifically, we first constructed and anonymized a dataset of 161 chest X-ray (CXR) radiographs and 219 CT reports for evaluation. Then, multiple LLMs, including Gemini 2.0-Flash, GPT-4o, and DeepSeek-R1, were evaluated based on recall, precision, and F1 score to detect technical errors and inconsistencies. Experimental results show that Gemini 2.0-Flash achieved a Macro F1 score of 90 in CXR tasks, demonstrating strong generalization but limited fine-grained performance. DeepSeek-R1 excelled in CT report auditing with a 62.23\% recall rate, outperforming other models. However, its distilled variants performed poorly, while InternLM2.5-7B-chat exhibited the highest additional discovery rate, indicating broader but less precise error detection. These findings highlight the potential of LLMs in medical imaging QC, with DeepSeek-R1 and Gemini 2.0-Flash demonstrating superior performance.
Authors: Ruidan Xing, Runyi Huang, Qing Xu, Lei He
Abstract: End-to-end autonomous driving solutions, which process multi-modal sensory data to directly generate refined control commands, have become a dominant paradigm in autonomous driving research. However, these approaches predominantly depend on single-vehicle data collection for model training and optimization, resulting in significant challenges such as high data acquisition and annotation costs, the scarcity of critical driving scenarios, and fragmented datasets that impede model generalization. To mitigate these limitations, we introduce RS2V-L, a novel framework for reconstructing and synthesizing vehicle-mounted LiDAR data from roadside sensor observations. Specifically, our method transforms roadside LiDAR point clouds into the vehicle-mounted LiDAR coordinate system by leveraging the target vehicle's relative pose. Subsequently, high-fidelity vehicle-mounted LiDAR data is synthesized through virtual LiDAR modeling, point cloud classification, and resampling techniques. To the best of our knowledge, this is the first approach to reconstruct vehicle-mounted LiDAR data from roadside sensor inputs. Extensive experimental evaluations demonstrate that incorporating the generated data into model training-complementing the KITTI dataset-enhances 3D object detection accuracy by over \text{30\%} while improving the efficiency of end-to-end autonomous driving data generation by more than an order of magnitude. These findings strongly validate the effectiveness of the proposed method and underscore its potential in reducing dependence on costly vehicle-mounted data collection while improving the robustness of autonomous driving models.
Authors: Xiaoyan Kui, Zijie Fan, Zexin Ji, Qinsong Li, Chengtao Liu, Weixin Si, Beiji Zou
Abstract: Magnetic resonance imaging (MRI) reconstruction is a fundamental task aimed at recovering high-quality images from undersampled or low-quality MRI data. This process enhances diagnostic accuracy and optimizes clinical applications. In recent years, deep learning-based MRI reconstruction has made significant progress. Advancements include single-modality feature extraction using different network architectures, the integration of multimodal information, and the adoption of unsupervised or semi-supervised learning strategies. However, despite extensive research, MRI reconstruction remains a challenging problem that has yet to be fully resolved. This survey provides a systematic review of MRI reconstruction methods, covering key aspects such as data acquisition and preprocessing, publicly available datasets, single and multi-modal reconstruction models, training strategies, and evaluation metrics based on image reconstruction and downstream tasks. Additionally, we analyze the major challenges in this field and explore potential future directions.
Authors: Valentin von Bornhaupt, Johannes Gr\"un, and Justus Bisten, Tobias Bauer, Theodor R\"uber, Thomas Schultz
Abstract: Whole-brain tractography in diffusion MRI is often followed by a parcellation in which each streamline is classified as belonging to a specific white matter bundle, or discarded as a false positive. Efficient parcellation is important both in large-scale studies, which have to process huge amounts of data, and in the clinic, where computational resources are often limited. TractCloud is a state-of-the-art approach that aims to maximize accuracy with a local-global representation. We demonstrate that the local context does not contribute to the accuracy of that approach, and is even detrimental when dealing with pathological cases. Based on this observation, we propose PETParc, a new method for Parallel Efficient Tractography Parcellation. PETParc is a transformer-based architecture in which the whole-brain tractogram is randomly partitioned into sub-tractograms whose streamlines are classified in parallel, while serving as global context for each other. This leads to a speedup of up to two orders of magnitude relative to TractCloud, and permits inference even on clinical workstations without a GPU. PETParc accounts for the lack of streamline orientation either via a novel flip-invariant embedding, or by simply using flips as part of data augmentation. Despite the speedup, results are often even better than those of prior methods. The code and pretrained model will be made public upon acceptance.
Authors: Yanis Basso-Bert, Anca Molnos, Romain Lemaire, William Guicquero, Antoine Dupret
Abstract: Binary Neural Networks (BNNs) are a promising approach to enable Artificial Neural Network (ANN) implementation on ultra-low power edge devices. Such devices may compute data in highly dynamic environments, in which the classes targeted for inference can evolve or even novel classes may arise, requiring continual learning. Class Incremental Learning (CIL) is a common type of continual learning for classification problems, that has been scarcely addressed in the context of BNNs. Furthermore, most of existing BNNs models are not fully binary, as they require several real-valued network layers, at the input, the output, and for batch normalization. This paper goes a step further, enabling class incremental learning in Fully-Binarized NNs (FBNNs) through four main contributions. We firstly revisit the FBNN design and its training procedure that is suitable to CIL. Secondly, we explore loss balancing, a method to trade-off the performance of past and current classes. Thirdly, we propose a semi-supervised method to pre-train the feature extractor of the FBNN for transferable representations. Fourthly, two conventional CIL methods, \ie, Latent and Native replay, are thoroughly compared. These contributions are exemplified first on the CIFAR100 dataset, before being scaled up to address the CORE50 continual learning benchmark. The final results based on our 3Mb FBNN on CORE50 exhibit at par and better performance than conventional real-valued larger NN models.
Authors: Hanzhi Chen, Boyang Sun, Anran Zhang, Marc Pollefeys, Stefan Leutenegger
Abstract: Future robots are envisioned as versatile systems capable of performing a variety of household tasks. The big question remains, how can we bridge the embodiment gap while minimizing physical robot learning, which fundamentally does not scale well. We argue that learning from in-the-wild human videos offers a promising solution for robotic manipulation tasks, as vast amounts of relevant data already exist on the internet. In this work, we present VidBot, a framework enabling zero-shot robotic manipulation using learned 3D affordance from in-the-wild monocular RGB-only human videos. VidBot leverages a pipeline to extract explicit representations from them, namely 3D hand trajectories from videos, combining a depth foundation model with structure-from-motion techniques to reconstruct temporally consistent, metric-scale 3D affordance representations agnostic to embodiments. We introduce a coarse-to-fine affordance learning model that first identifies coarse actions from the pixel space and then generates fine-grained interaction trajectories with a diffusion model, conditioned on coarse actions and guided by test-time constraints for context-aware interaction planning, enabling substantial generalization to novel scenes and embodiments. Extensive experiments demonstrate the efficacy of VidBot, which significantly outperforms counterparts across 13 manipulation tasks in zero-shot settings and can be seamlessly deployed across robot systems in real-world environments. VidBot paves the way for leveraging everyday human videos to make robot learning more scalable.
Authors: Wietske A. P. Bastiaansen, Melek Rousian, Anton H. J. Koning, Wiro J. Niessen, Bernadette S. de Bakker, R\'egine P. M. Steegers-Theunissen, Stefan Klein
Abstract: Early brain development is crucial for lifelong neurodevelopmental health. However, current clinical practice offers limited knowledge of normal embryonic brain anatomy on ultrasound, despite the brain undergoing rapid changes within the time-span of days. To provide detailed insights into normal brain development and identify deviations, we created the 4D Human Embryonic Brain Atlas using a deep learning-based approach for groupwise registration and spatiotemporal atlas generation. Our method introduced a time-dependent initial atlas and penalized deviations from it, ensuring age-specific anatomy was maintained throughout rapid development. The atlas was generated and validated using 831 3D ultrasound images from 402 subjects in the Rotterdam Periconceptional Cohort, acquired between gestational weeks 8 and 12. We evaluated the effectiveness of our approach with an ablation study, which demonstrated that incorporating a time-dependent initial atlas and penalization produced anatomically accurate results. In contrast, omitting these adaptations led to anatomically incorrect atlas. Visual comparisons with an existing ex-vivo embryo atlas further confirmed the anatomical accuracy of our atlas. In conclusion, the proposed method successfully captures the rapid anotomical development of the embryonic brain. The resulting 4D Human Embryonic Brain Atlas provides a unique insights into this crucial early life period and holds the potential for improving the detection, prevention, and treatment of prenatal neurodevelopmental disorders.
Authors: Yan Ren, Shilin Lu, Adams Wai-Kin Kong
Abstract: Recent advances in 3D Gaussian Splatting (3DGS) have revolutionized scene reconstruction, opening new possibilities for 3D steganography by hiding 3D secrets within 3D covers. The key challenge in steganography is ensuring imperceptibility while maintaining high-fidelity reconstruction. However, existing methods often suffer from detectability risks and utilize only suboptimal 3DGS features, limiting their full potential. We propose a novel end-to-end key-secured 3D steganography framework (KeySS) that jointly optimizes a 3DGS model and a key-secured decoder for secret reconstruction. Our approach reveals that Gaussian features contribute unequally to secret hiding. The framework incorporates a key-controllable mechanism enabling multi-secret hiding and unauthorized access prevention, while systematically exploring optimal feature update to balance fidelity and security. To rigorously evaluate steganographic imperceptibility beyond conventional 2D metrics, we introduce 3D-Sinkhorn distance analysis, which quantifies distributional differences between original and steganographic Gaussian parameters in the representation space. Extensive experiments demonstrate that our method achieves state-of-the-art performance in both cover and secret reconstruction while maintaining high security levels, advancing the field of 3D steganography. Code is available at https://github.com/RY-Paper/KeySS
Authors: Zixuan Wang, Chi-Keung Tang, Yu-Wing Tai
Abstract: Film production is an important application for generative audio, where richer context is provided through multiple scenes. In ReelWave, we propose a multi-agent framework for audio generation inspired by the professional movie production process. We first capture semantic and temporal synchronized "on-screen" sound by training a prediction model that predicts three interpretable time-varying audio control signals comprising loudness, pitch, and timbre. These three parameters are subsequently specified as conditions by a cross-attention module. Then, our framework infers "off-screen" sound to complement the generation through cooperative interaction between communicative agents. Each agent takes up specific roles similar to the movie production team and is supervised by an agent called the director. Besides, we investigate when the conditional video consists of multiple scenes, a case frequently seen in videos extracted from movies of considerable length. Consequently, our framework can capture a richer context of audio generation conditioned on video clips extracted from movies.
Authors: Xinyu Nan, Meng He, Zifan Chen, Bin Dong, Lei Tang, Li Zhang
Abstract: The incidence of gastrointestinal cancers remains significantly high, particularly in China, emphasizing the importance of accurate prognostic assessments and effective treatment strategies. Research shows a strong correlation between abdominal muscle and fat tissue composition and patient outcomes. However, existing manual methods for analyzing abdominal tissue composition are time-consuming and costly, limiting clinical research scalability. To address these challenges, we developed an AI-driven tool for automated analysis of abdominal CT scans to effectively identify and segment muscle, subcutaneous fat, and visceral fat. Our tool integrates a multi-view localization model and a high-precision 2D nnUNet-based segmentation model, demonstrating a localization accuracy of 90% and a Dice Score Coefficient of 0.967 for segmentation. Furthermore, it features an interactive interface that allows clinicians to refine the segmentation results, ensuring high-quality outcomes effectively. Our tool offers a standardized method for effectively extracting critical abdominal tissues, potentially enhancing the management and treatment for gastrointestinal cancers. The code is available at https://github.com/NanXinyu/AI-Tool4Abdominal-Seg.git}{https://github.com/NanXinyu/AI-Tool4Abdominal-Seg.git.
URLs: https://github.com/NanXinyu/AI-Tool4Abdominal-Seg.git, https://github.com/NanXinyu/AI-Tool4Abdominal-Seg.git.
Authors: Cristian Perez Jensen, Seyedmorteza Sadat
Abstract: While classifier-free guidance (CFG) is essential for conditional diffusion models, it doubles the number of neural function evaluations (NFEs) per inference step. To mitigate this inefficiency, we introduce adapter guidance distillation (AGD), a novel approach that simulates CFG in a single forward pass. AGD leverages lightweight adapters to approximate CFG, effectively doubling the sampling speed while maintaining or even improving sample quality. Unlike prior guidance distillation methods that tune the entire model, AGD keeps the base model frozen and only trains minimal additional parameters ($\sim$2%) to significantly reduce the resource requirement of the distillation phase. Additionally, this approach preserves the original model weights and enables the adapters to be seamlessly combined with other checkpoints derived from the same base model. We also address a key mismatch between training and inference in existing guidance distillation methods by training on CFG-guided trajectories instead of standard diffusion trajectories. Through extensive experiments, we show that AGD achieves comparable or superior FID to CFG across multiple architectures with only half the NFEs. Notably, our method enables the distillation of large models ($\sim$2.6B parameters) on a single consumer GPU with 24 GB of VRAM, making it more accessible than previous approaches that require multiple high-end GPUs. We will publicly release the implementation of our method.
Authors: Jiarui Wu, Yujin Wang, Lingen Li, Zhang Fan, Tianfan Xue
Abstract: Photo finishing tuning aims to automate the manual tuning process of the photo finishing pipeline, like Adobe Lightroom or Darktable. Previous works either use zeroth-order optimization, which is slow when the set of parameters increases, or rely on a differentiable proxy of the target finishing pipeline, which is hard to train. To overcome these challenges, we propose a novel goal-conditioned reinforcement learning framework for efficiently tuning parameters using a goal image as a condition. Unlike previous approaches, our tuning framework does not rely on any proxy and treats the photo finishing pipeline as a black box. Utilizing a trained reinforcement learning policy, it can efficiently find the desired set of parameters within just 10 queries, while optimization based approaches normally take 200 queries. Furthermore, our architecture utilizes a goal image to guide the iterative tuning of pipeline parameters, allowing for flexible conditioning on pixel-aligned target images, style images, or any other visually representable goals. We conduct detailed experiments on photo finishing tuning and photo stylization tuning tasks, demonstrating the advantages of our method. Project website: https://openimaginglab.github.io/RLPixTuner/.
Authors: Rui Qiao, Zhaoxuan Wu, Jingtan Wang, Pang Wei Koh, Bryan Kian Hsiang Low
Abstract: Machine learning models often have uneven performance among subpopulations (a.k.a., groups) in the data distributions. This poses a significant challenge for the models to generalize when the proportions of the groups shift during deployment. To improve robustness to such shifts, existing approaches have developed strategies that train models or perform hyperparameter tuning using the group-labeled data to minimize the worst-case loss over groups. However, a non-trivial amount of high-quality labels is often required to obtain noticeable improvements. Given the costliness of the labels, we propose to adopt a different paradigm to enhance group label efficiency: utilizing the group-labeled data as a target set to optimize the weights of other group-unlabeled data. We introduce Group-robust Sample Reweighting (GSR), a two-stage approach that first learns the representations from group-unlabeled data, and then tinkers the model by iteratively retraining its last layer on the reweighted data using influence functions. Our GSR is theoretically sound, practically lightweight, and effective in improving the robustness to subpopulation shifts. In particular, GSR outperforms the previous state-of-the-art approaches that require the same amount or even more group labels.
Authors: Yubo Zhao, Qi Wu, Yifan Wang, Yu-Wing Tai, Chi-Keung Tang
Abstract: Large Language Models (LLMs) have demonstrated strong generalizable reasoning and planning capabilities. However, their efficacies in spatial path planning and obstacle-free trajectory generation remain underexplored. Leveraging LLMs for navigation holds significant potential, given LLMs' ability to handle unseen scenarios, support user-agent interactions, and provide global control across complex systems, making them well-suited for agentic planning and humanoid motion generation. As one of the first studies in this domain, we explore the zero-shot navigation and path generation capabilities of LLMs by constructing a dataset and proposing an evaluation protocol. Specifically, we represent paths using anchor points connected by straight lines, enabling movement in various directions. This approach offers greater flexibility and practicality compared to previous methods while remaining simple and intuitive for LLMs. We demonstrate that, when tasks are well-structured in this manner, modern LLMs exhibit substantial planning proficiency in avoiding obstacles while autonomously refining navigation with the generated motion to reach the target. Further, this spatial reasoning ability of a single LLM motion agent interacting in a static environment can be seamlessly generalized in multi-motion agents coordination in dynamic environments. Unlike traditional approaches that rely on single-step planning or local policies, our training-free LLM-based method enables global, dynamic, closed-loop planning, and autonomously resolving collision issues.
Authors: Luis D. Reyes Vargas, Martin J. Menten, Johannes C. Paetzold, Nassir Navab, Mohammad Farid Azampour
Abstract: Skeletonization extracts thin representations from images that compactly encode their geometry and topology. These representations have become an important topological prior for preserving connectivity in curvilinear structures, aiding medical tasks like vessel segmentation. Existing compatible skeletonization algorithms face significant trade-offs: morphology-based approaches are computationally efficient but prone to frequent breakages, while topology-preserving methods require substantial computational resources. We propose a novel framework for training iterative skeletonization algorithms with a learnable component. The framework leverages synthetic data, task-specific augmentation, and a model distillation strategy to learn compact neural networks that produce thin, connected skeletons with a fully differentiable iterative algorithm. Our method demonstrates a 100 times speedup over topology-constrained algorithms while maintaining high accuracy and generalizing effectively to new domains without fine-tuning. Benchmarking and downstream validation in 2D and 3D tasks demonstrate its computational efficiency and real-world applicability
Authors: Ziliang Xiong, Shipeng Liu, Nathaniel Helgesen, Joakim Johnander, Per-Erik Forssen
Abstract: In recent years, there has been increased interest in the design, training, and evaluation of end-to-end autonomous driving (AD) systems. One often overlooked aspect is the uncertainty of planned trajectories predicted by these systems, despite awareness of their own uncertainty being key to achieve safety and robustness. We propose to estimate this uncertainty by adapting loss prediction from the uncertainty quantification literature. To this end, we introduce a novel light-weight module, dubbed CATPlan, that is trained to decode motion and planning embeddings into estimates of the collision loss used to partially supervise end-to-end AD systems. During inference, these estimates are interpreted as collision risk. We evaluate CATPlan on the safety-critical, nerf-based, closed-loop benchmark NeuroNCAP and find that it manages to detect collisions with a $54.8\%$ relative improvement to average precision over a GMM-based baseline in which the predicted trajectory is compared to the forecasted trajectories of other road users. Our findings indicate that the addition of CATPlan can lead to safer end-to-end AD systems and hope that our work will spark increased interest in uncertainty quantification for such systems.
Authors: Chengrui Zhu, Ryoichi Ishikawa, Masataka Kagesawa, Tomohisa Yuzawa, Toru Watsuji, Takeshi Oishi
Abstract: Reconstructing three-dimensional (3D) structures from two-dimensional (2D) X-ray images is a valuable and efficient technique in medical applications that requires less radiation exposure than computed tomography scans. Recent approaches that use implicit neural representations have enabled the synthesis of novel views from sparse X-ray images. However, although image synthesis has improved the accuracy, the accuracy of surface shape estimation remains insufficient. Therefore, we propose a novel approach for reconstructing 3D scenes using a Neural Attenuation Surface (NeAS) that simultaneously captures the surface geometry and attenuation coefficient fields. NeAS incorporates a signed distance function (SDF), which defines the attenuation field and aids in extracting the 3D surface within the scene. We conducted experiments using simulated and authentic X-ray images, and the results demonstrated that NeAS could accurately extract 3D surfaces within a scene using only 2D X-ray images.
Authors: Soumya Banerjee, Vinay Kumar Verma
Abstract: Active learning aims to select optimal samples for labeling, minimizing annotation costs. This paper introduces a unified representation learning framework tailored for active learning with task awareness. It integrates diverse sources, comprising reconstruction, adversarial, self-supervised, knowledge-distillation, and classification losses into a unified VAE-based ADROIT approach. The proposed approach comprises three key components - a unified representation generator (VAE), a state discriminator, and a (proxy) task-learner or classifier. ADROIT learns a latent code using both labeled and unlabeled data, incorporating task-awareness by leveraging labeled data with the proxy classifier. Unlike previous approaches, the proxy classifier additionally employs a self-supervised loss on unlabeled data and utilizes knowledge distillation to align with the target task-learner. The state discriminator distinguishes between labeled and unlabeled data, facilitating the selection of informative unlabeled samples. The dynamic interaction between VAE and the state discriminator creates a competitive environment, with the VAE attempting to deceive the discriminator, while the state discriminator learns to differentiate between labeled and unlabeled inputs. Extensive evaluations on diverse datasets and ablation analysis affirm the effectiveness of the proposed model.
Authors: Chengmeng Li, Junjie Wen, Yan Peng, Yaxin Peng, Feifei Feng, Yichen Zhu
Abstract: Vision-Language-Action (VLA) models excel at robotic tasks by leveraging large-scale 2D vision-language pretraining, but their reliance on RGB images limits spatial reasoning critical for real-world interaction. Retraining these models with 3D data is computationally prohibitive, while discarding existing 2D datasets wastes valuable resources. To bridge this gap, we propose PointVLA, a framework that enhances pre-trained VLAs with point cloud inputs without requiring retraining. Our method freezes the vanilla action expert and injects 3D features via a lightweight modular block. To identify the most effective way of integrating point cloud representations, we conduct a skip-block analysis to pinpoint less useful blocks in the vanilla action expert, ensuring that 3D features are injected only into these blocks--minimizing disruption to pre-trained representations. Extensive experiments demonstrate that PointVLA outperforms state-of-the-art 2D imitation learning methods, such as OpenVLA, Diffusion Policy and DexVLA, across both simulated and real-world robotic tasks. Specifically, we highlight several key advantages of PointVLA enabled by point cloud integration: (1) Few-shot multi-tasking, where PointVLA successfully performs four different tasks using only 20 demonstrations each; (2) Real-vs-photo discrimination, where PointVLA distinguishes real objects from their images, leveraging 3D world knowledge to improve safety and reliability; (3) Height adaptability, Unlike conventional 2D imitation learning methods, PointVLA enables robots to adapt to objects at varying table height that unseen in train data. Furthermore, PointVLA achieves strong performance in long-horizon tasks, such as picking and packing objects from a moving conveyor belt, showcasing its ability to generalize across complex, dynamic environments.
Authors: Tianyu Chen, Yasi Zhang, Zhendong Wang, Ying Nian Wu, Oscar Leong, Mingyuan Zhou
Abstract: Diffusion models have achieved remarkable success in generating high-resolution, realistic images across diverse natural distributions. However, their performance heavily relies on high-quality training data, making it challenging to learn meaningful distributions from corrupted samples. This limitation restricts their applicability in scientific domains where clean data is scarce or costly to obtain. In this work, we introduce denoising score distillation (DSD), a surprisingly effective and novel approach for training high-quality generative models from low-quality data. DSD first pretrains a diffusion model exclusively on noisy, corrupted samples and then distills it into a one-step generator capable of producing refined, clean outputs. While score distillation is traditionally viewed as a method to accelerate diffusion models, we show that it can also significantly enhance sample quality, particularly when starting from a degraded teacher model. Across varying noise levels and datasets, DSD consistently improves generative performancewe summarize our empirical evidence in Fig. 1. Furthermore, we provide theoretical insights showing that, in a linear model setting, DSD identifies the eigenspace of the clean data distributions covariance matrix, implicitly regularizing the generator. This perspective reframes score distillation as not only a tool for efficiency but also a mechanism for improving generative models, particularly in low-quality data settings.
Authors: Shengqin Jiang, Yuan Gao, Bowen Li, Fengna Cheng, Renlong Hang, Qingshan Liu
Abstract: Efficient models for remote sensing object counting are urgently required for applications in scenarios with limited computing resources, such as drones or embedded systems. A straightforward yet powerful technique to achieve this is knowledge distillation, which steers the learning of student networks by leveraging the experience of already-trained teacher networks. However, it faces a pair of challenges: Firstly, due to its two-stage training nature, a longer training period is essential, especially as the training samples increase. Secondly, despite the proficiency of teacher networks in transmitting assimilated knowledge, they tend to overlook the latent insights gained during their learning process. To address these challenges, we introduce an online distillation learning method for remote sensing object counting. It builds an end-to-end training framework that seamlessly integrates two distinct networks into a unified one. It comprises a shared shallow module, a teacher branch, and a student branch. The shared module serving as the foundation for both branches is dedicated to learning some primitive information. The teacher branch utilizes prior knowledge to reduce the difficulty of learning and guides the student branch in online learning. In parallel, the student branch achieves parameter reduction and rapid inference capabilities by means of channel reduction. This design empowers the student branch not only to receive privileged insights from the teacher branch but also to tap into the latent reservoir of knowledge held by the teacher branch during the learning process. Moreover, we propose a relation-in-relation distillation method that allows the student branch to effectively comprehend the evolution of the relationship of intra-layer teacher features among different inter-layer features. Extensive experiments demonstrate the effectiveness of our method.
Authors: Moayed Haji Ali, Andrew Bond, Tolga Birdal, Duygu Ceylan, Levent Karacan, Erkut Erdem, Aykut Erdem
Abstract: We propose $\textbf{VidStyleODE}$, a spatiotemporally continuous disentangled $\textbf{Vid}$eo representation based upon $\textbf{Style}$GAN and Neural-$\textbf{ODE}$s. Effective traversal of the latent space learned by Generative Adversarial Networks (GANs) has been the basis for recent breakthroughs in image editing. However, the applicability of such advancements to the video domain has been hindered by the difficulty of representing and controlling videos in the latent space of GANs. In particular, videos are composed of content (i.e., appearance) and complex motion components that require a special mechanism to disentangle and control. To achieve this, VidStyleODE encodes the video content in a pre-trained StyleGAN $\mathcal{W}_+$ space and benefits from a latent ODE component to summarize the spatiotemporal dynamics of the input video. Our novel continuous video generation process then combines the two to generate high-quality and temporally consistent videos with varying frame rates. We show that our proposed method enables a variety of applications on real videos: text-guided appearance manipulation, motion manipulation, image animation, and video interpolation and extrapolation. Project website: https://cyberiada.github.io/VidStyleODE
Authors: Kaiyi Huang, Chengqi Duan, Kaiyue Sun, Enze Xie, Zhenguo Li, Xihui Liu
Abstract: Despite the impressive advances in text-to-image models, they often struggle to effectively compose complex scenes with multiple objects, displaying various attributes and relationships. To address this challenge, we present T2I-CompBench++, an enhanced benchmark for compositional text-to-image generation. T2I-CompBench++ comprises 8,000 compositional text prompts categorized into four primary groups: attribute binding, object relationships, generative numeracy, and complex compositions. These are further divided into eight sub-categories, including newly introduced ones like 3D-spatial relationships and numeracy. In addition to the benchmark, we propose enhanced evaluation metrics designed to assess these diverse compositional challenges. These include a detection-based metric tailored for evaluating 3D-spatial relationships and numeracy, and an analysis leveraging Multimodal Large Language Models (MLLMs), i.e. GPT-4V, ShareGPT4v as evaluation metrics. Our experiments benchmark 11 text-to-image models, including state-of-the-art models, such as FLUX.1, SD3, DALLE-3, Pixart-${\alpha}$, and SD-XL on T2I-CompBench++. We also conduct comprehensive evaluations to validate the effectiveness of our metrics and explore the potential and limitations of MLLMs.
Authors: Yingzhi Tang, Qijian Zhang, Junhui Hou, Yebin Liu
Abstract: The latest trends in the research field of single-view human reconstruction devote to learning deep implicit functions constrained by explicit body shape priors. Despite the remarkable performance improvements compared with traditional processing pipelines, existing learning approaches still show different aspects of limitations in terms of flexibility, generalizability, robustness, and/or representation capability. To comprehensively address the above issues, in this paper, we investigate an explicit point-based human reconstruction framework called HaP, which adopts point clouds as the intermediate representation of the target geometric structure. Technically, our approach is featured by fully-explicit point cloud estimation, manipulation, generation, and refinement in the 3D geometric space, instead of an implicit learning process that can be ambiguous and less controllable. The overall workflow is carefully organized with dedicated designs of the corresponding specialized learning components as well as processing procedures. Extensive experiments demonstrate that our framework achieves quantitative performance improvements of 20% to 40% over current state-of-the-art methods, and better qualitative results. Our promising results may indicate a paradigm rollback to the fully-explicit and geometry-centric algorithm design, which enables to exploit various powerful point cloud modeling architectures and processing techniques. We will make our code and data publicly available at https://github.com/yztang4/HaP.
Authors: Do Huu Dat, Po Yuan Mao, Tien Hoang Nguyen, Wray Buntine, Mohammed Bennamoun
Abstract: Compositional Zero-Shot Learning (CZSL) has emerged as an essential paradigm in machine learning, aiming to overcome the constraints of traditional zero-shot learning by incorporating compositional thinking into its methodology. Conventional zero-shot learning has difficulty managing unfamiliar combinations of seen and unseen classes because it depends on pre-defined class embeddings. In contrast, Compositional Zero-Shot Learning leverages the inherent hierarchies and structural connections among classes, creating new class representations by combining attributes, components, or other semantic elements. In our paper, we propose a novel framework that for the first time combines the Modern \underline{H}opfield Network with a Mixture \underline{o}f \underline{E}x\underline{p}erts (HOPE) to classify the compositions of previously unseen objects. Specifically, the Modern Hopfield Network creates a memory that stores label prototypes and identifies relevant labels for a given input image. Subsequently, the Mixture of Expert models integrates the image with the appropriate prototype to produce the final composition classification. Our approach achieves SOTA performance on several benchmarks, including MIT-States and UT-Zappos. We also examine how each component contributes to improved generalization.
Authors: Evelyn Mannix, Howard Bondell
Abstract: One of the early weaknesses identified in deep neural networks trained for image classification tasks was their inability to provide low confidence predictions on out-of-distribution (OOD) data that was significantly different from the in-distribution (ID) data used to train them. Representation learning, where neural networks are trained in specific ways that improve their ability to detect OOD examples, has emerged as a promising solution. However, these approaches require long training times and can add additional overhead to detect OOD examples. Recent developments in Vision Transformer (ViT) foundation models$\unicode{x2013}$large networks trained on large and diverse datasets with self-supervised approaches$\unicode{x2013}$also show strong performance in OOD detection, and could address these challenges. This paper presents Mixture of Exemplars (MoLAR), an efficient approach to tackling OOD detection challenges that is designed to maximise the benefit of training a classifier with a high quality, frozen, pretrained foundation model backbone. MoLAR provides strong OOD performance when only comparing the similarity of OOD examples to the exemplars, a small set of images chosen to be representative of the dataset, leading to up to 30 times faster OOD detection inference over other methods that provide best performance when the full ID dataset is used. In some cases, only using these exemplars actually improves performance with MoLAR. Extensive experiments demonstrate the improved OOD detection performance of MoLAR in comparison to comparable approaches in both supervised and semi-supervised settings, and code is available at github.com/emannix/molar-mixture-of-exemplars.
Authors: Alexander Becker, Rodrigo Caye Daudt, Dominik Narnhofer, Torben Peters, Nando Metzger, Jan Dirk Wegner, Konrad Schindler
Abstract: Recent approaches to arbitrary-scale single image super-resolution (ASR) use neural fields to represent continuous signals that can be sampled at arbitrary resolutions. However, point-wise queries of neural fields do not naturally match the point spread function (PSF) of pixels, which may cause aliasing in the super-resolved image. Existing methods attempt to mitigate this by approximating an integral version of the field at each scaling factor, compromising both fidelity and generalization. In this work, we introduce neural heat fields, a novel neural field formulation that inherently models a physically exact PSF. Our formulation enables analytically correct anti-aliasing at any desired output resolution, and -- unlike supersampling -- at no additional cost. Building on this foundation, we propose Thera, an end-to-end ASR method that substantially outperforms existing approaches, while being more parameter-efficient and offering strong theoretical guarantees. The project page is at https://therasr.github.io.
Authors: Moayed Haji-Ali, Guha Balakrishnan, Vicente Ordonez
Abstract: Diffusion models have revolutionized image generation in recent years, yet they are still limited to a few sizes and aspect ratios. We propose ElasticDiffusion, a novel training-free decoding method that enables pretrained text-to-image diffusion models to generate images with various sizes. ElasticDiffusion attempts to decouple the generation trajectory of a pretrained model into local and global signals. The local signal controls low-level pixel information and can be estimated on local patches, while the global signal is used to maintain overall structural consistency and is estimated with a reference image. We test our method on CelebA-HQ (faces) and LAION-COCO (objects/indoor/outdoor scenes). Our experiments and qualitative results show superior image coherence quality across aspect ratios compared to MultiDiffusion and the standard decoding strategy of Stable Diffusion. Project page: https://elasticdiffusion.github.io/
Authors: Qian Wang, Hefei Ling, Yingwei Li, Qihao Liu, Ruoxi Jia, Ning Yu
Abstract: In response to the rapidly evolving nature of adversarial attacks against visual classifiers on a monthly basis, numerous defenses have been proposed to generalize against as many known attacks as possible. However, designing a defense method that generalizes to all types of attacks is not realistic because the environment in which defense systems operate is dynamic and comprises various unique attacks that emerge as time goes on. A well-matched approach to the dynamic environment lies in a defense system that continuously collects adversarial data online to quickly improve itself. Therefore, we put forward a practical defense deployment against a challenging threat model and propose, for the first time, the Continual Adversarial Defense (CAD) framework that adapts to attack sequences under four principles: (1)~continual adaptation to new attacks without catastrophic forgetting, (2)~few-shot adaptation, (3)~memory-efficient adaptation, and (4)~high accuracy on both clean and adversarial data. We explore and integrate cutting-edge continual learning, few-shot learning, and ensemble learning techniques to qualify the principles. Extensive experiments validate the effectiveness of our approach against multiple stages of modern adversarial attacks and demonstrate significant improvements over numerous baseline methods. In particular, CAD is capable of quickly adapting with minimal budget and a low cost of defense failure while maintaining good performance against previous attacks. Our research sheds light on a brand-new paradigm for continual defense adaptation against dynamic and evolving attacks.
Authors: Tonmoy Hossain, Miaomiao Zhang
Abstract: Geometric transformations have been widely used to augment the size of training images. Existing methods often assume a unimodal distribution of the underlying transformations between images, which limits their power when data with multimodal distributions occur. In this paper, we propose a novel model, Multimodal Geometric Augmentation (MGAug), that for the first time generates augmenting transformations in a multimodal latent space of geometric deformations. To achieve this, we first develop a deep network that embeds the learning of latent geometric spaces of diffeomorphic transformations (a.k.a. diffeomorphisms) in a variational autoencoder (VAE). A mixture of multivariate Gaussians is formulated in the tangent space of diffeomorphisms and serves as a prior to approximate the hidden distribution of image transformations. We then augment the original training dataset by deforming images using randomly sampled transformations from the learned multimodal latent space of VAE. To validate the efficiency of our model, we jointly learn the augmentation strategy with two distinct domain-specific tasks: multi-class classification on 2D synthetic datasets and segmentation on real 3D brain magnetic resonance images (MRIs). We also compare MGAug with state-of-the-art transformation-based image augmentation algorithms. Experimental results show that our proposed approach outperforms all baselines by significantly improved prediction accuracy. Our code is publicly available at https://github.com/tonmoy-hossain/MGAug.
Authors: Jun Wang, Hao Ruan, Liangjian Wen, Yong Dai, Mingjie Wang
Abstract: Visual gaze estimation, with its wide-ranging application scenarios, has garnered increasing attention within the research community. Although existing approaches infer gaze solely from image signals, recent advances in visual-language collaboration have demonstrated that the integration of linguistic information can significantly enhance performance across various visual tasks. Leveraging the remarkable transferability of large-scale Contrastive Language-Image Pre-training (CLIP) models, we address the open and urgent question of how to effectively apply linguistic cues to gaze estimation. In this work, we propose GazeCLIP, a novel gaze estimation framework that deeply explores text-face collaboration. Specifically, we introduce a meticulously designed linguistic description generator to produce text signals enriched with coarse directional cues. Furthermore, we present a CLIP-based backbone adept at characterizing text-face pairs for gaze estimation, complemented by a fine-grained multimodal fusion module that models the intricate interrelationships between heterogeneous inputs. Extensive experiments on three challenging datasets demonstrate the superiority of GazeCLIP, which achieves state-of-the-art accuracy. Our findings underscore the potential of using visual-language collaboration to advance gaze estimation and open new avenues for future research in multimodal learning for visual tasks. The implementation code and the pre-trained model will be made publicly available.
Authors: Zheng Chen, Qingan Yan, Huangying Zhan, Changjiang Cai, Xiangyu Xu, Yuzhong Huang, Weihan Wang, Ziyue Feng, Yi Xu, Lantao Liu
Abstract: Identifying spatially complete planar primitives from visual data is a crucial task in computer vision. Prior methods are largely restricted to either 2D segment recovery or simplifying 3D structures, even with extensive plane annotations. We present PlanarNeRF, a novel framework capable of detecting dense 3D planes through online learning. Drawing upon the neural field representation, PlanarNeRF brings three major contributions. First, it enhances 3D plane detection with concurrent appearance and geometry knowledge. Second, a lightweight plane fitting module is proposed to estimate plane parameters. Third, a novel global memory bank structure with an update mechanism is introduced, ensuring consistent cross-frame correspondence. The flexible architecture of PlanarNeRF allows it to function in both 2D-supervised and self-supervised solutions, in each of which it can effectively learn from sparse training signals, significantly improving training efficiency. Through extensive experiments, we demonstrate the effectiveness of PlanarNeRF in various scenarios and remarkable improvement over existing works.
Authors: Chenyang Wang, Junjun Jiang, Xingyu Hu, Xianming Liu, Xiangyang Ji
Abstract: Deep learning systems are prone to catastrophic forgetting when learning from a sequence of tasks, as old data from previous tasks is unavailable when learning a new task. To address this, some methods propose replaying data from previous tasks during new task learning, typically using extra memory to store replay data. However, it is not expected in practice due to memory constraints and data privacy issues. Instead, data-free replay methods invert samples from the classification model. While effective, these methods face inconsistencies between inverted and real training data, overlooked in recent works. To that effect, we propose to measure the data consistency quantitatively by some simplification and assumptions. Using this measurement, we gain insight to develop a novel loss function that reduces inconsistency. Specifically, the loss minimizes the KL divergence between distributions of inverted and real data under a tied multivariate Gaussian assumption, which is simple to implement in continual learning. Additionally, we observe that old class weight norms decrease continually as learning progresses. We analyze the reasons and propose a regularization term to balance class weights, making old class samples more distinguishable. To conclude, we introduce Consistency-enhanced data replay with a Debiased classifier for class incremental learning (CwD). Extensive experiments on CIFAR-100, Tiny-ImageNet, and ImageNet100 show consistently improved performance of CwD compared to previous approaches.
Authors: Yi Huang, Jiancheng Huang, Yifan Liu, Mingfu Yan, Jiaxi Lv, Jianzhuang Liu, Wei Xiong, He Zhang, Liangliang Cao, Shifeng Chen
Abstract: Denoising diffusion models have emerged as a powerful tool for various image generation and editing tasks, facilitating the synthesis of visual content in an unconditional or input-conditional manner. The core idea behind them is learning to reverse the process of gradually adding noise to images, allowing them to generate high-quality samples from a complex distribution. In this survey, we provide an exhaustive overview of existing methods using diffusion models for image editing, covering both theoretical and practical aspects in the field. We delve into a thorough analysis and categorization of these works from multiple perspectives, including learning strategies, user-input conditions, and the array of specific editing tasks that can be accomplished. In addition, we pay special attention to image inpainting and outpainting, and explore both earlier traditional context-driven and current multimodal conditional methods, offering a comprehensive analysis of their methodologies. To further evaluate the performance of text-guided image editing algorithms, we propose a systematic benchmark, EditEval, featuring an innovative metric, LMM Score. Finally, we address current limitations and envision some potential directions for future research. The accompanying repository is released at https://github.com/SiatMMLab/Awesome-Diffusion-Model-Based-Image-Editing-Methods.
URLs: https://github.com/SiatMMLab/Awesome-Diffusion-Model-Based-Image-Editing-Methods.
Authors: Jiaxiang Cheng, Pan Xie, Xin Xia, Jiashi Li, Jie Wu, Yuxi Ren, Huixia Li, Xuefeng Xiao, Min Zheng, Lean Fu
Abstract: Recent advancement in text-to-image models (e.g., Stable Diffusion) and corresponding personalized technologies (e.g., DreamBooth and LoRA) enables individuals to generate high-quality and imaginative images. However, they often suffer from limitations when generating images with resolutions outside of their trained domain. To overcome this limitation, we present the Resolution Adapter (ResAdapter), a domain-consistent adapter designed for diffusion models to generate images with unrestricted resolutions and aspect ratios. Unlike other multi-resolution generation methods that process images of static resolution with complex post-process operations, ResAdapter directly generates images with the dynamical resolution. Especially, after learning a deep understanding of pure resolution priors, ResAdapter trained on the general dataset, generates resolution-free images with personalized diffusion models while preserving their original style domain. Comprehensive experiments demonstrate that ResAdapter with only 0.5M can process images with flexible resolutions for arbitrary diffusion models. More extended experiments demonstrate that ResAdapter is compatible with other modules (e.g., ControlNet, IP-Adapter and LCM-LoRA) for image generation across a broad range of resolutions, and can be integrated into other multi-resolution model (e.g., ElasticDiffusion) for efficiently generating higher-resolution images. Project link is https://res-adapter.github.io
Authors: Evelyn J. Mannix, Liam Hodgkinson, Howard Bondell
Abstract: Interpretable computer vision models explain their classifications through comparing the distances between the local embeddings of an image and a set of prototypes that represent the training data. However, these approaches introduce additional hyper-parameters that need to be tuned to apply to new datasets, scale poorly, and are more computationally intensive to train in comparison to black-box approaches. In this work, we introduce Component Features (ComFe), a highly scalable interpretable-by-design image classification head for pretrained Vision Transformers (ViTs) that can obtain competitive performance in comparison to comparable non-interpretable methods. ComFe is the first interpretable head, that we know of, and unlike other interpretable approaches, can be readily applied to large scale datasets such as ImageNet-1K. Additionally, ComFe provides improved robustness and outperforms previous interpretable approaches on key benchmark datasets$\unicode{x2013}$using a consistent set of hyper-parameters and without finetuning the pretrained ViT backbone. With only global image labels and no segmentation or part annotations, ComFe can identify consistent component features within an image and determine which of these features are informative in making a prediction. Code is available at https://github.com/emannix/comfe-component-features.
Authors: Yuelin Zhang, Kim Yan, Chun Ping Lam, Chengyu Fang, Wenxuan Xie, Yufu Qiu, Raymond Shing-Yan Tang, Shing Shin Cheng
Abstract: Flexible endoscope motion tracking and analysis in mechanical simulators have proven useful for endoscopy training. Common motion tracking methods based on electromagnetic tracker are however limited by their high cost and material susceptibility. In this work, the motion-guided dual-camera vision tracker is proposed to provide robust and accurate tracking of the endoscope tip's 3D position. The tracker addresses several unique challenges of tracking flexible endoscope tip inside a dynamic, life-sized mechanical simulator. To address the appearance variation and keep dual-camera tracking consistency, the cross-camera mutual template strategy (CMT) is proposed by introducing dynamic transient mutual templates. To alleviate large occlusion and light-induced distortion, the Mamba-based motion-guided prediction head (MMH) is presented to aggregate historical motion with visual tracking. The proposed tracker achieves superior performance against state-of-the-art vision trackers, achieving 42% and 72% improvements against the second-best method in average error and maximum error. Further motion analysis involving novice and expert endoscopists also shows that the tip 3D motion provided by the proposed tracker enables more reliable motion analysis and more substantial differentiation between different expertise levels, compared with other trackers. Project page: https://github.com/PieceZhang/MotionDCTrack
Authors: Chaoyang Wang, Xiangtai Li, Henghui Ding, Lu Qi, Jiangning Zhang, Yunhai Tong, Chen Change Loy, Shuicheng Yan
Abstract: In-context segmentation has drawn increasing attention with the advent of vision foundation models. Its goal is to segment objects using given reference images. Most existing approaches adopt metric learning or masked image modeling to build the correlation between visual prompts and input image queries. This work approaches the problem from a fresh perspective - unlocking the capability of the latent diffusion model (LDM) for in-context segmentation and investigating different design choices. Specifically, we examine the problem from three angles: instruction extraction, output alignment, and meta-architectures. We design a two-stage masking strategy to prevent interfering information from leaking into the instructions. In addition, we propose an augmented pseudo-masking target to ensure the model predicts without forgetting the original images. Moreover, we build a new and fair in-context segmentation benchmark that covers both image and video datasets. Experiments validate the effectiveness of our approach, demonstrating comparable or even stronger results than previous specialist or visual foundation models. We hope our work inspires others to rethink the unification of segmentation and generation.
Authors: Alexander Hepburn, Raul Santos-Rodriguez, Javier Portilla
Abstract: Two-alternative forced choice (2AFC) experiments are popular in the visual perception literature to understand how human observers perceive distances within triplets made of a reference image and two distorted versions. Previously, this had been conducted in controlled environments, with triplets sharing images, making it possible to rank the perceived quality and evaluate perceptual distance models against the ranking. Recently, crowd-sourced perceptual datasets have emerged, with no images shared between triplets, making ranking infeasible. Evaluations using this data reduces the judgements on a triplet to a binary decision, namely, whether the distance model agrees with the human decision - which is suboptimal and prone to misleading conclusions. Instead, we statistically model the underlying decision-making process during 2AFC experiments using a binomial distribution. We estimate a smooth and consistent distribution of the judgements on the reference-distorted distance plane, according to each distance model. We estimate the parameter of the local binomial distribution using maximum likelihood, and a global measurement of the expected log-likelihood of the judgements. We calculate meaningful and well-founded metrics, beyond the mere prediction accuracy as percentage agreement and compare to a neural network counterpart, also optimised to maximise likelihood according to a binomial model.
Authors: Lorenzo Agnolucci, Leonardo Galteri, Marco Bertini
Abstract: No-Reference Image Quality Assessment (NR-IQA) focuses on designing methods to measure image quality in alignment with human perception when a high-quality reference image is unavailable. Most state-of-the-art NR-IQA approaches are opinion-aware, i.e. they require human annotations for training. This dependency limits their scalability and broad applicability. To overcome this limitation, we propose QualiCLIP (Quality-aware CLIP), a CLIP-based self-supervised opinion-unaware approach that does not require human opinions. In particular, we introduce a quality-aware image-text alignment strategy to make CLIP generate quality-aware image representations. Starting from pristine images, we synthetically degrade them with increasing levels of intensity. Then, we train CLIP to rank these degraded images based on their similarity to quality-related antonym text prompts. At the same time, we force CLIP to generate consistent representations for images with similar content and the same level of degradation. Our experiments show that the proposed method improves over existing opinion-unaware approaches across multiple datasets with diverse distortion types. Moreover, despite not requiring human annotations, QualiCLIP achieves excellent performance against supervised opinion-aware methods in cross-dataset experiments, thus demonstrating remarkable generalization capabilities. The code and the model are publicly available at https://github.com/miccunifi/QualiCLIP.
Authors: Kartik Narayan, Vibashan VS, Rama Chellappa, Vishal M. Patel
Abstract: In this work, we introduce FaceXFormer, an end-to-end unified transformer model capable of performing ten facial analysis tasks within a single framework. These tasks include face parsing, landmark detection, head pose estimation, attribute prediction, age, gender, and race estimation, facial expression recognition, face recognition, and face visibility. Traditional face analysis approaches rely on task-specific architectures and pre-processing techniques, limiting scalability and integration. In contrast, FaceXFormer employs a transformer-based encoder-decoder architecture, where each task is represented as a learnable token, enabling seamless multi-task processing within a unified model. To enhance efficiency, we introduce FaceX, a lightweight decoder with a novel bi-directional cross-attention mechanism, which jointly processes face and task tokens to learn robust and generalized facial representations. We train FaceXFormer on ten diverse face perception datasets and evaluate it against both specialized and multi-task models across multiple benchmarks, demonstrating state-of-the-art or competitive performance. Additionally, we analyze the impact of various components of FaceXFormer on performance, assess real-world robustness in "in-the-wild" settings, and conduct a computational performance evaluation. To the best of our knowledge, FaceXFormer is the first model capable of handling ten facial analysis tasks while maintaining real-time performance at 33.21 FPS. Code: https://github.com/Kartik-3004/facexformer
Authors: Jiaqi Yue, Chunhui Zhao, Jiancheng Zhao, Biao Huang
Abstract: Generalized zero-shot learning (GZSL) focuses on recognizing seen and unseen classes against domain shift problem where data of unseen classes may be misclassified as seen classes. However, existing GZSL is still limited to seen domains. In the current work, we study cross-domain GZSL (CDGZSL) which addresses GZSL towards unseen domains. Different from existing GZSL methods, CDGZSL constructs a common feature space across domains and acquires the corresponding intrinsic semantics shared among domains to transfer from seen to unseen domains. Considering the information asymmetry problem caused by redundant class semantics annotated with large language models (LLMs), we present Meta Domain Alignment Semantic Refinement (MDASR). Technically, MDASR consists of two parts: Inter-class similarity alignment, which eliminates the non-intrinsic semantics not shared across all domains under the guidance of inter-class feature relationships, and unseen-class meta generation, which preserves intrinsic semantics to maintain connectivity between seen and unseen classes by simulating feature generation. MDASR effectively aligns the redundant semantic space with the common feature space, mitigating the information asymmetry in CDGZSL. The effectiveness of MDASR is demonstrated on two datasets, Office-Home and Mini-DomainNet, and we have shared the LLM-based semantics for these datasets as a benchmark.
Authors: Ahmad Mahmood, Ashmal Vayani, Muzammal Naseer, Salman Khan, Fahad Shahbaz Khan
Abstract: Recent studies have demonstrated the effectiveness of Large Language Models (LLMs) as reasoning modules that can deconstruct complex tasks into more manageable sub-tasks, particularly when applied to visual reasoning tasks for images. In contrast, this paper introduces a Video Understanding and Reasoning Framework (VURF) based on the reasoning power of LLMs. Ours is a novel approach to extend the utility of LLMs in the context of video tasks, leveraging their capacity to generalize from minimal input and output demonstrations within a contextual framework. We harness their contextual learning capabilities by presenting LLMs with pairs of instructions and their corresponding high-level programs to generate executable visual programs for video understanding. To enhance the program's accuracy and robustness, we implement two important strategies. \emph{Firstly,} we employ a feedback-generation approach, powered by GPT-3.5, to rectify errors in programs utilizing unsupported functions. \emph{Secondly}, taking motivation from recent works on self-refinement of LLM outputs, we introduce an iterative procedure for improving the quality of the in-context examples by aligning the initial outputs to the outputs that would have been generated had the LLM not been bound by the structure of the in-context examples. Our results on several video-specific tasks, including visual QA, video anticipation, pose estimation, and multi-video QA, illustrate these enhancements' efficacy in improving the performance of visual programming approaches for video tasks.
Authors: Guang Lin, Zerui Tao, Jianhai Zhang, Toshihisa Tanaka, Qibin Zhao
Abstract: Diffusion models (DMs) based adversarial purification (AP) has shown to be the most powerful alternative to adversarial training (AT). However, these methods neglect the fact that pre-trained diffusion models themselves are not robust to adversarial attacks as well. Additionally, the diffusion process can easily destroy semantic information and generate a high quality image but totally different from the original input image after the reverse process, leading to degraded standard accuracy. To overcome these issues, a natural idea is to harness adversarial training strategy to retrain or fine-tune the pre-trained diffusion model, which is computationally prohibitive. We propose a novel robust reverse process with adversarial guidance, which is independent of given pre-trained DMs and avoids retraining or fine-tuning the DMs. This robust guidance can not only ensure to generate purified examples retaining more semantic content but also mitigate the accuracy-robustness trade-off of DMs for the first time, which also provides DM-based AP an efficient adaptive ability to new attacks. Extensive experiments are conducted on CIFAR-10, CIFAR-100 and ImageNet to demonstrate that our method achieves the state-of-the-art results and exhibits generalization against different attacks.
Authors: Quang-Huy Che, Duc-Tri Le, Minh-Quan Pham, Vinh-Tiep Nguyen, Duc-Khai Lam
Abstract: Semantic segmentation is crucial for autonomous driving, particularly for the tasks of Drivable Area and Lane Segmentation, ensuring safety and navigation. To address the high computational costs of current state-of-the-art (SOTA) models, this paper introduces TwinLiteNetPlus, a model capable of balancing efficiency and accuracy. TwinLiteNetPlus incorporates standard and depth-wise separable dilated convolutions, reducing complexity while maintaining high accuracy. It is available in four configurations, from the robust 1.94 million-parameter TwinLiteNetPlus_{Large} to the ultra-lightweight 34K-parameter TwinLiteNetPlus_{Nano}. Notably, TwinLiteNetPlus_{Large} attains a 92.9% mIoU (mean Intersection over Union) for Drivable Area Segmentation and a 34.2% IoU (Intersection over Union) for Lane Segmentation. These results achieve remarkable performance, surpassing current state-of-the-art models while only requiring 11 times fewer Floating Point Operations (FLOPs) for computation. Rigorously evaluated on various embedded devices, TwinLiteNetPlus demonstrates promising latency and power efficiency, underscoring its potential for real-world autonomous vehicle applications. The code is available on https://github.com/chequanghuy/TwinLiteNetPlus.
Authors: Shuai Xiang, Pieter M. Blok, James Burridge, Haozhou Wang, Wei Guo
Abstract: Object detection has wide applications in agriculture, but domain shifts of diverse environments limit the broader use of the trained models. Existing domain adaptation methods usually require retraining the model for new domains, which is impractical for agricultural applications due to constantly changing environments. In this paper, we propose DODA ($D$iffusion for $O$bject-detection $D$omain Adaptation in $A$griculture), a diffusion-based framework that can adapt the detector to a new domain in just 2 minutes. DODA incorporates external domain embeddings and an improved layout-to-image approach, allowing it to generate high-quality detection data for new domains without additional training. We demonstrate DODA's effectiveness on the Global Wheat Head Detection dataset, where fine-tuning detectors on DODA-generated data yields significant improvements across multiple domains. DODA provides a simple yet powerful solution for agricultural domain adaptation, reducing the barriers for growers to use detection in personalised environments. The code is available at https://github.com/UTokyo-FieldPhenomics-Lab/DODA.
Authors: Chen Yang, Yangfan He, Aaron Xuxiang Tian, Dong Chen, Jianhui Wang, Tianyu Shi, Arsalan Heydarian, Pei Liu
Abstract: In this paper, we introduce a novel approach for autonomous driving trajectory generation by harnessing the complementary strengths of diffusion probabilistic models (a.k.a., diffusion models) and transformers. Our proposed framework, termed the "World-Centric Diffusion Transformer"(WcDT), optimizes the entire trajectory generation process, from feature extraction to model inference. To enhance the scene diversity and stochasticity, the historical trajectory data is first preprocessed into "Agent Move Statement" and encoded into latent space using Denoising Diffusion Probabilistic Models (DDPM) enhanced with Diffusion with Transformer (DiT) blocks. Then, the latent features, historical trajectories, HD map features, and historical traffic signal information are fused with various transformer-based encoders that are used to enhance the interaction of agents with other elements in the traffic scene. The encoded traffic scenes are then decoded by a trajectory decoder to generate multimodal future trajectories. Comprehensive experimental results show that the proposed approach exhibits superior performance in generating both realistic and diverse trajectories, showing its potential for integration into automatic driving simulation systems. Our code is available at \url{https://github.com/yangchen1997/WcDT}.
Authors: Haoyang He, Yuhu Bai, Jiangning Zhang, Qingdong He, Hongxu Chen, Zhenye Gan, Chengjie Wang, Xiangtai Li, Guanzhong Tian, Lei Xie
Abstract: Recent advancements in anomaly detection have seen the efficacy of CNN- and transformer-based approaches. However, CNNs struggle with long-range dependencies, while transformers are burdened by quadratic computational complexity. Mamba-based models, with their superior long-range modeling and linear efficiency, have garnered substantial attention. This study pioneers the application of Mamba to multi-class unsupervised anomaly detection, presenting MambaAD, which consists of a pre-trained encoder and a Mamba decoder featuring (Locality-Enhanced State Space) LSS modules at multi-scales. The proposed LSS module, integrating parallel cascaded (Hybrid State Space) HSS blocks and multi-kernel convolutions operations, effectively captures both long-range and local information. The HSS block, utilizing (Hybrid Scanning) HS encoders, encodes feature maps into five scanning methods and eight directions, thereby strengthening global connections through the (State Space Model) SSM. The use of Hilbert scanning and eight directions significantly improves feature sequence modeling. Comprehensive experiments on six diverse anomaly detection datasets and seven metrics demonstrate state-of-the-art performance, substantiating the method's effectiveness. The code and models are available at https://lewandofskee.github.io/projects/MambaAD.
Authors: Rongjian Xu, Zhilu Zhang, Renlong Wu, Wangmeng Zuo
Abstract: Despite the significant progress in image denoising, it is still challenging to restore fine-scale details while removing noise, especially in extremely low-light environments. Leveraging near-infrared (NIR) images to assist visible RGB image denoising shows the potential to address this issue, becoming a promising technology. Nonetheless, existing works still struggle with taking advantage of NIR information effectively for real-world image denoising, due to the content inconsistency between NIR-RGB images and the scarcity of real-world paired datasets. To alleviate the problem, we propose an efficient Selective Fusion Module (SFM), which can be plug-and-played into the advanced denoising networks to merge the deep NIR-RGB features. Specifically, we sequentially perform the global and local modulation for NIR and RGB features, and then integrate the two modulated features. Furthermore, we present a Real-world NIR-Assisted Image Denoising (Real-NAID) dataset, which covers diverse scenarios as well as various noise levels. Extensive experiments on both synthetic and our real-world datasets demonstrate that the proposed method achieves better results than state-of-the-art ones. The dataset, codes, and pre-trained models will be publicly available at https://github.com/ronjonxu/NAID.
Authors: Junghyup Lee, Jeimin Jeon, Dohyung Kim, Bumsub Ham
Abstract: Quantization-aware training (QAT) simulates a quantization process during training to lower bit-precision of weights/activations. It learns quantized weights indirectly by updating latent weights,i.e., full-precision inputs to a quantizer, using gradient-based optimizers. We claim that coupling a user-defined learning rate (LR) with these optimizers is sub-optimal for QAT. Quantized weights transit discrete levels of a quantizer, only if corresponding latent weights pass transition points, where the quantizer changes discrete states. This suggests that the changes of quantized weights are affected by both the LR for latent weights and their distributions. It is thus difficult to control the degree of changes for quantized weights by scheduling the LR manually. We conjecture that the degree of parameter changes in QAT is related to the number of quantized weights transiting discrete levels. Based on this, we introduce a transition rate (TR) scheduling technique that controls the number of transitions of quantized weights explicitly. Instead of scheduling a LR for latent weights, we schedule a target TR of quantized weights, and update the latent weights with a novel transition-adaptive LR (TALR), enabling considering the degree of changes for the quantized weights during QAT. Experimental results demonstrate the effectiveness of our approach on standard benchmarks.
Authors: Zhangyong Tang, Tianyang Xu, Zhenhua Feng, Xuefeng Zhu, Chunyang Cheng, Xiao-Jun Wu, Josef Kittler
Abstract: RGBT tracking draws increasing attention because its robustness in multi-modal warranting (MMW) scenarios, such as nighttime and adverse weather conditions, where relying on a single sensing modality fails to ensure stable tracking results. However, existing benchmarks predominantly contain videos collected in common scenarios where both RGB and thermal infrared (TIR) information are of sufficient quality. This weakens the representativeness of existing benchmarks in severe imaging conditions, leading to tracking failures in MMW scenarios. To bridge this gap, we present a new benchmark considering the modality validity, MV-RGBT, captured specifically from MMW scenarios where either RGB (extreme illumination) or TIR (thermal truncation) modality is invalid. Hence, it is further divided into two subsets according to the valid modality, offering a new compositional perspective for evaluation and providing valuable insights for future designs. Moreover, MV-RGBT is the most diverse benchmark of its kind, featuring 36 different object categories captured across 19 distinct scenes. Furthermore, considering severe imaging conditions in MMW scenarios, a new problem is posed in RGBT tracking, named `when to fuse', to stimulate the development of fusion strategies for such scenarios. To facilitate its discussion, we propose a new solution with a mixture of experts, named MoETrack, where each expert generates independent tracking results along with a confidence score. Extensive results demonstrate the significant potential of MV-RGBT in advancing RGBT tracking and elicit the conclusion that fusion is not always beneficial, especially in MMW scenarios. Besides, MoETrack achieves state-of-the-art results on several benchmarks, including MV-RGBT, GTOT, and LasHeR. Github: https://github.com/Zhangyong-Tang/MVRGBT.
Authors: Junkai Fan, Jiangwei Weng, Kun Wang, Yijun Yang, Jianjun Qian, Jun Li, Jian Yang
Abstract: Real driving-video dehazing poses a significant challenge due to the inherent difficulty in acquiring precisely aligned hazy/clear video pairs for effective model training, especially in dynamic driving scenarios with unpredictable weather conditions. In this paper, we propose a pioneering approach that addresses this challenge through a nonaligned regularization strategy. Our core concept involves identifying clear frames that closely match hazy frames, serving as references to supervise a video dehazing network. Our approach comprises two key components: reference matching and video dehazing. Firstly, we introduce a non-aligned reference frame matching module, leveraging an adaptive sliding window to match high-quality reference frames from clear videos. Video dehazing incorporates flow-guided cosine attention sampler and deformable cosine attention fusion modules to enhance spatial multiframe alignment and fuse their improved information. To validate our approach, we collect a GoProHazy dataset captured effortlessly with GoPro cameras in diverse rural and urban road environments. Extensive experiments demonstrate the superiority of the proposed method over current state-of-the-art methods in the challenging task of real driving-video dehazing. Project page.
Authors: Run Luo, Yunshui Li, Longze Chen, Wanwei He, Ting-En Lin, Ziqiang Liu, Lei Zhang, Zikai Song, Xiaobo Xia, Tongliang Liu, Min Yang, Binyuan Hui
Abstract: The development of large language models (LLMs) has significantly advanced the emergence of large multimodal models (LMMs). While LMMs have achieved tremendous success by promoting the synergy between multimodal comprehension and creation, they often face challenges when confronted with out-of-distribution data, such as which can hardly distinguish orientation, quantity, color, structure, etc. This is primarily due to their reliance on image encoders trained to encode images into task-relevant features, which may lead them to disregard irrelevant details. Delving into the modeling capabilities of diffusion models for images naturally prompts the question: Can diffusion models serve as the eyes of large language models for image perception? In this paper, we propose DEEM, a simple but effective approach that utilizes the generative feedback of diffusion models to align the semantic distributions of the image encoder. This addresses the drawbacks of previous methods that solely relied on image encoders like CLIP-ViT, thereby enhancing the model's resilience against out-of-distribution samples and reducing visual hallucinations. Importantly, this is achieved without requiring additional training modules and with fewer training parameters. We extensively evaluated DEEM on both our newly constructed RobustVQA benchmark and other well-known benchmarks, POPE and MMVP, for visual hallucination and perception. In particular, DEEM improves LMM's visual perception performance to a large extent (e.g., 4% higher on RobustVQA, 6.5% higher on MMVP and 12.8 % higher on POPE ). Compared to the state-of-the-art interleaved content generation models, DEEM exhibits enhanced robustness and a superior capacity to alleviate model hallucinations while utilizing fewer trainable parameters, less pre-training data (10%), and a smaller base model size.
Authors: Yibo Zhang, Lihong Wang, Changqing Zou, Tieru Wu, Rui Ma
Abstract: 3D sketches are widely used for visually representing the 3D shape and structure of objects or scenes. However, the creation of 3D sketch often requires users to possess professional artistic skills. Existing research efforts primarily focus on enhancing the ability of interactive sketch generation in 3D virtual systems. In this work, we propose Diff3DS, a novel differentiable rendering framework for generating view-consistent 3D sketch by optimizing 3D parametric curves under various supervisions. Specifically, we perform perspective projection to render the 3D rational B\'ezier curves into 2D curves, which are subsequently converted to a 2D raster image via our customized differentiable rasterizer. Our framework bridges the domains of 3D sketch and raster image, achieving end-toend optimization of 3D sketch through gradients computed in the 2D image domain. Our Diff3DS can enable a series of novel 3D sketch generation tasks, including textto-3D sketch and image-to-3D sketch, supported by the popular distillation-based supervision, such as Score Distillation Sampling (SDS). Extensive experiments have yielded promising results and demonstrated the potential of our framework. Project page is at https://yiboz2001.github.io/Diff3DS/.
Authors: Abdelrahman Abdelhamed, Mahmoud Afifi, Alec Go
Abstract: Large language models (LLMs) have been effectively used for many computer vision tasks, including image classification. In this paper, we present a simple yet effective approach for zero-shot image classification using multimodal LLMs. Using multimodal LLMs, we generate comprehensive textual representations from input images. These textual representations are then utilized to generate fixed-dimensional features in a cross-modal embedding space. Subsequently, these features are fused together to perform zero-shot classification using a linear classifier. Our method does not require prompt engineering for each dataset; instead, we use a single, straightforward set of prompts across all datasets. We evaluated our method on several datasets and our results demonstrate its remarkable effectiveness, surpassing benchmark accuracy on multiple datasets. On average, for ten benchmarks, our method achieved an accuracy gain of 6.2 percentage points, with an increase of 6.8 percentage points on the ImageNet dataset, compared to prior methods re-evaluated with the same setup. Our findings highlight the potential of multimodal LLMs to enhance computer vision tasks such as zero-shot image classification, offering a significant improvement over traditional methods.
Authors: Zejun Li, Ruipu Luo, Jiwen Zhang, Minghui Qiu, Xuanjing Huang, Zhongyu Wei
Abstract: While large multi-modal models (LMMs) have exhibited impressive capabilities across diverse tasks, their effectiveness in handling complex tasks has been limited by the prevailing single-step reasoning paradigm. To this end, this paper proposes VoCoT, a multi-step Visually grounded object-centric Chain-of-Thought reasoning framework tailored for inference with LMMs. VoCoT is characterized by two key features: (1) object-centric reasoning paths that revolve around cross-modal shared object-level information, and (2) visually grounded representation of object concepts in a multi-modal interleaved and aligned manner, which effectively bridges the modality gap within LMMs during long-term generation. To adapt LMMs in reasoning with VoCoT, we further construct an instruction-tuning dataset. By combining VoCoT with the prevalent open-source LMM architectures, we develop a VoCoT-based model, VolCano. With only 7B parameters and limited input image resolution, VolCano demonstrates excellent performance across various scenarios. In benchmarks like CLEVR and EmbSpatial, which highly require complex reasoning capabilities, VolCano outperforms SOTA models, including powerful GPT-4V. Related code, data and models are released in https://github.com/RupertLuo/VoCoT.
Authors: Seokju Yun, Dongheon Lee, Youngmin Ro
Abstract: Transformer, composed of self-attention and Feed-Forward Network, has revolutionized the landscape of network design across various vision tasks. While self-attention is extensively explored as a key factor in performance, FFN has received little attention. FFN is a versatile operator seamlessly integrated into nearly all AI models to effectively harness rich representations. Recent works also show that FFN functions like key-value memories. Thus, akin to the query-key-value mechanism within self-attention, FFN can be viewed as a memory network, where the input serves as query and the two projection weights operate as keys and values, respectively. Based on these observations, we hypothesize that the importance lies in query-key-value framework itself for competitive performance. To verify this, we propose converting self-attention into a more FFN-like efficient token mixer with only convolutions while retaining query-key-value framework, namely FFNification. Specifically, FFNification replaces query-key-value interactions with large kernel convolutions and adopts GELU activation function instead of softmax. The derived token mixer, FFNified attention, serves as key-value memories for detecting locally distributed spatial patterns, and operates in the opposite dimension to the ConvNeXt block within each corresponding sub-operation of the query-key-value framework. Building upon the above two modules, we present a family of Fast-Forward Networks (FFNet). Despite being composed of only simple operators, FFNet outperforms sophisticated and highly specialized methods in each domain, with notable efficiency gains. These results validate our hypothesis, leading us to propose MetaMixer, a general mixer architecture that does not specify sub-operations within the query-key-value framework.
Authors: Man Liu, Huihui Bai, Feng Li, Chunjie Zhang, Yunchao Wei, Tat-Seng Chua, Yao Zhao
Abstract: Zero-shot learning (ZSL) endeavors to transfer knowledge from seen categories to recognize unseen categories, which mostly relies on the semantic-visual interactions between image and attribute tokens. Recently, prompt learning has emerged in ZSL and demonstrated significant potential as it allows the zero-shot transfer of diverse visual concepts to downstream tasks. However, current methods explore the fixed adaption of learnable prompt on seen domains, which makes them over-emphasize the primary visual features observed during training, limiting their generalization capabilities to unseen domains. In this work, we propose AENet, which endows semantic information into the visual prompt to distill semantic-enhanced prompt for visual representation enrichment, enabling effective knowledge transfer for ZSL. AENet comprises two key steps: 1) exploring the concept-harmonized tokens for the visual and attribute modalities, grounded on the modal-sharing token that represents consistent visual-semantic concepts; and 2) yielding semantic-enhanced prompt via the visual residual refinement unit with attribute consistency supervision. These are further integrated with primary visual features to attend to semantic-related information for visual enhancement, thus strengthening transferable ability. Experimental results on three benchmarks show that our AENet outperforms existing state-of-the-art ZSL methods. The code is provided in the zip file of supplementary materials.
Authors: Zhenyu Wang, Jianxi Huang, Zhida Sun, Yuanhao Gong, Daniel Cohen-Or, Min Lu
Abstract: This work presents a progressive image vectorization technique that reconstructs the raster image as layer-wise vectors from semantic-aligned macro structures to finer details. Our approach introduces a new image simplification method leveraging the feature-average effect in the Score Distillation Sampling mechanism, achieving effective visual abstraction from the detailed to coarse. Guided by the sequence of progressive simplified images, we propose a two-stage vectorization process of structural buildup and visual refinement, constructing the vectors in an organized and manageable manner. The resulting vectors are layered and well-aligned with the target image's explicit and implicit semantic structures. Our method demonstrates high performance across a wide range of images. Comparative analysis with existing vectorization methods highlights our technique's superiority in creating vectors with high visual fidelity, and more importantly, achieving higher semantic alignment and more compact layered representation. The project homepage is https://szuviz.github.io/layered_vectorization/.
Authors: Hao Fang, Jiawei Kong, Wenbo Yu, Bin Chen, Jiawei Li, Hao Wu, Shutao Xia, Ke Xu
Abstract: Vision-Language Pre-training (VLP) models have exhibited unprecedented capability in many applications by taking full advantage of the multimodal alignment. However, previous studies have shown they are vulnerable to maliciously crafted adversarial samples. Despite recent success, these methods are generally instance-specific and require generating perturbations for each input sample. In this paper, we reveal that VLP models are also vulnerable to the instance-agnostic universal adversarial perturbation (UAP). Specifically, we design a novel Contrastive-training Perturbation Generator with Cross-modal conditions (C-PGC) to achieve the attack. In light that the pivotal multimodal alignment is achieved through the advanced contrastive learning technique, we devise to turn this powerful weapon against themselves, i.e., employ a malicious version of contrastive learning to train the C-PGC based on our carefully crafted positive and negative image-text pairs for essentially destroying the alignment relationship learned by VLP models. Besides, C-PGC fully utilizes the characteristics of Vision-and-Language (V+L) scenarios by incorporating both unimodal and cross-modal information as effective guidance. Extensive experiments show that C-PGC successfully forces adversarial samples to move away from their original area in the VLP model's feature space, thus essentially enhancing attacks across various victim models and V+L tasks. The GitHub repository is available at https://github.com/ffhibnese/CPGC_VLP_Universal_Attacks.
URLs: https://github.com/ffhibnese/CPGC_VLP_Universal_Attacks.
Authors: Pranav Jeevan, Amit Sethi
Abstract: In contemporary computer vision applications, particularly image classification, architectural backbones pre-trained on large datasets like ImageNet are commonly employed as feature extractors. Despite the widespread use of these pre-trained convolutional neural networks (CNNs), there remains a gap in understanding the performance of various resource-efficient backbones across diverse domains and dataset sizes. Our study systematically evaluates multiple lightweight, pre-trained CNN backbones under consistent training settings across a variety of datasets, including natural images, medical images, galaxy images, and remote sensing images. This comprehensive analysis aims to aid machine learning practitioners in selecting the most suitable backbone for their specific problem, especially in scenarios involving small datasets where fine-tuning a pre-trained network is crucial. Even though attention-based architectures are gaining popularity, we observed that they tend to perform poorly under low data finetuning tasks compared to CNNs. We also observed that some CNN architectures such as ConvNeXt, RegNet and EfficientNet performs well compared to others on a diverse set of domains consistently. Our findings provide actionable insights into the performance trade-offs and effectiveness of different backbones, facilitating informed decision-making in model selection for a broad spectrum of computer vision domains. Our code is available here: https://github.com/pranavphoenix/Backbones
Authors: Yaoteng Tan, Zikui Cai, M. Salman Asif
Abstract: Deep networks are highly vulnerable to adversarial attacks, yet conventional attack methods utilize static adversarial perturbations that induce fixed mispredictions. In this work, we exploit an overlooked property of adversarial perturbations--their dependence on image transforms--and introduce transform-dependent adversarial attacks. Unlike traditional attacks, our perturbations exhibit metamorphic properties, enabling diverse adversarial effects as a function of transformation parameters. We demonstrate that this transform-dependent vulnerability exists across different architectures (e.g., CNN and transformer), vision tasks (e.g., image classification and object detection), and a wide range of image transforms. Additionally, we show that transform-dependent perturbations can serve as a defense mechanism, preventing sensitive information disclosure when image enhancement transforms pose a risk of revealing private content. Through analysis in blackbox and defended model settings, we show that transform-dependent perturbations achieve high targeted attack success rates, outperforming state-of-the-art transfer attacks by 17-31% in blackbox scenarios. Our work introduces novel, controllable paradigm for adversarial attack deployment, revealing a previously overlooked vulnerability in deep networks.
Authors: Yuang Peng, Yuxin Cui, Haomiao Tang, Zekun Qi, Runpei Dong, Jing Bai, Chunrui Han, Zheng Ge, Xiangyu Zhang, Shu-Tao Xia
Abstract: Personalized image generation holds great promise in assisting humans in everyday work and life due to its impressive ability to creatively generate personalized content across various contexts. However, current evaluations either are automated but misalign with humans or require human evaluations that are time-consuming and expensive. In this work, we present DreamBench++, a human-aligned benchmark that advanced multimodal GPT models automate. Specifically, we systematically design the prompts to let GPT be both human-aligned and self-aligned, empowered with task reinforcement. Further, we construct a comprehensive dataset comprising diverse images and prompts. By benchmarking 7 modern generative models, we demonstrate that DreamBench++ results in significantly more human-aligned evaluation, helping boost the community with innovative findings.
Authors: Yuxuan Zhang, Tianheng Cheng, Lianghui Zhu, Rui Hu, Lei Liu, Heng Liu, Longjin Ran, Xiaoxin Chen, Wenyu Liu, Xinggang Wang
Abstract: Segment Anything Model (SAM) has attracted widespread attention for its superior interactive segmentation capabilities with visual prompts while lacking further exploration of text prompts. In this paper, we empirically investigate what text prompt encoders (e.g., CLIP or LLM) are good for adapting SAM for referring expression segmentation and introduce the Early Vision-language Fusion-based SAM (EVF-SAM). EVF-SAM is a simple yet effective referring segmentation method which exploits multimodal prompts (i.e., image and text) and comprises a pre-trained vision-language model to generate referring prompts and a SAM model for segmentation. Surprisingly, we observe that: (1) multimodal prompts and (2) vision-language models with early fusion (e.g., BEIT-3) are beneficial for prompting SAM for accurate referring segmentation. Our experiments show that the proposed EVF-SAM based on BEIT-3 can obtain state-of-the-art performance on RefCOCO/+/g for referring expression segmentation and demonstrate the superiority of prompting SAM with early vision-language fusion. In addition, the proposed EVF-SAM with 1.32B parameters achieves remarkably higher performance while reducing nearly 82% of parameters compared to previous SAM methods based on large multimodal models.
Authors: Huiguo He, Huan Yang, Zixi Tuo, Yuan Zhou, Qiuyue Wang, Yuhang Zhang, Zeyu Liu, Wenhao Huang, Hongyang Chao, Jian Yin
Abstract: Story visualization aims to create visually compelling images or videos corresponding to textual narratives. Despite recent advances in diffusion models yielding promising results, existing methods still struggle to create a coherent sequence of subject-consistent frames based solely on a story. To this end, we propose DreamStory, an automatic open-domain story visualization framework by leveraging the LLMs and a novel multi-subject consistent diffusion model. DreamStory consists of (1) an LLM acting as a story director and (2) an innovative Multi-Subject consistent Diffusion model (MSD) for generating consistent multi-subject across the images. First, DreamStory employs the LLM to generate descriptive prompts for subjects and scenes aligned with the story, annotating each scene's subjects for subsequent subject-consistent generation. Second, DreamStory utilizes these detailed subject descriptions to create portraits of the subjects, with these portraits and their corresponding textual information serving as multimodal anchors (guidance). Finally, the MSD uses these multimodal anchors to generate story scenes with consistent multi-subject. Specifically, the MSD includes Masked Mutual Self-Attention (MMSA) and Masked Mutual Cross-Attention (MMCA) modules. MMSA and MMCA modules ensure appearance and semantic consistency with reference images and text, respectively. Both modules employ masking mechanisms to prevent subject blending. To validate our approach and promote progress in story visualization, we established a benchmark, DS-500, which can assess the overall performance of the story visualization framework, subject-identification accuracy, and the consistency of the generation model. Extensive experiments validate the effectiveness of DreamStory in both subjective and objective evaluations. Please visit our project homepage at https://dream-xyz.github.io/dreamstory.
Authors: Yushi Huang, Ruihao Gong, Xianglong Liu, Jing Liu, Yuhang Li, Jiwen Lu, Dacheng Tao
Abstract: The Diffusion models, widely used for image generation, face significant challenges related to their broad applicability due to prolonged inference times and high memory demands. Efficient Post-Training Quantization (PTQ) is crucial to address these issues. However, unlike traditional models, diffusion models critically rely on the time-step for the multi-round denoising. Typically, each time-step is encoded into a hypersensitive temporal feature by several modules. Despite this, existing PTQ methods do not optimize these modules individually. Instead, they employ unsuitable reconstruction objectives and complex calibration methods, leading to significant disturbances in the temporal feature and denoising trajectory, as well as reduced compression efficiency. To address these challenges, we introduce a novel quantization framework that includes three strategies: 1) TIB-based Maintenance: Based on our innovative Temporal Information Block (TIB) definition, Temporal Information-aware Reconstruction (TIAR) and Finite Set Calibration (FSC) are developed to efficiently align original temporal features. 2) Cache-based Maintenance: Instead of indirect and complex optimization for the related modules, pre-computing and caching quantized counterparts of temporal features are developed to minimize errors. 3) Disturbance-aware Selection: Employ temporal feature errors to guide a fine-grained selection between the two maintenance strategies for further disturbance reduction. This framework preserves most of the temporal information and ensures high-quality end-to-end generation. Extensive testing on various datasets, diffusion models and hardware confirms our superior performance and acceleration.
Authors: Yunfeng Diao, Naixin Zhai, Changtao Miao, Zitong Yu, Xingxing Wei, Xun Yang, Meng Wang
Abstract: Recent advancements in image synthesis, particularly with the advent of GAN and Diffusion models, have amplified public concerns regarding the dissemination of disinformation. To address such concerns, numerous AI-generated Image (AIGI) Detectors have been proposed and achieved promising performance in identifying fake images. However, there still lacks a systematic understanding of the adversarial robustness of AIGI detectors. In this paper, we examine the vulnerability of state-of-the-art AIGI detectors against adversarial attack under white-box and black-box settings, which has been rarely investigated so far. To this end, we propose a new method to attack AIGI detectors. First, inspired by the obvious difference between real images and fake images in the frequency domain, we add perturbations under the frequency domain to push the image away from its original frequency distribution. Second, we explore the full posterior distribution of the surrogate model to further narrow this gap between heterogeneous AIGI detectors, e.g. transferring adversarial examples across CNNs and ViTs. This is achieved by introducing a novel post-train Bayesian strategy that turns a single surrogate into a Bayesian one, capable of simulating diverse victim models using one pre-trained surrogate, without the need for re-training. We name our method as Frequency-based Post-train Bayesian Attack, or FPBA. Through FPBA, we show that adversarial attack is truly a real threat to AIGI detectors, because FPBA can deliver successful black-box attacks across models, generators, defense methods, and even evade cross-generator detection, which is a crucial real-world detection scenario. The code will be shared upon acceptance.
Authors: Zhuofan Zhang, Ziyu Zhu, Junhao Li, Pengxiang Li, Tianxu Wang, Tengyu Liu, Xiaojian Ma, Yixin Chen, Baoxiong Jia, Siyuan Huang, Qing Li
Abstract: Grounding natural language in 3D environments is a critical step toward achieving robust 3D vision-language alignment. Current datasets and models for 3D visual grounding predominantly focus on identifying and localizing objects from static, object-centric descriptions. These approaches do not adequately address the dynamic and sequential nature of task-oriented scenarios. In this work, we introduce a novel task: Task-oriented Sequential Grounding and Navigation in 3D Scenes, where models must interpret step-by-step instructions for daily activities by either localizing a sequence of target objects in indoor scenes or navigating toward them within a 3D simulator. To facilitate this task, we present SG3D, a large-scale dataset comprising 22,346 tasks with 112,236 steps across 4,895 real-world 3D scenes. The dataset is constructed by combining RGB-D scans from various 3D scene datasets with an automated task generation pipeline, followed by human verification for quality assurance. We benchmark contemporary methods on SG3D, revealing the significant challenges in understanding task-oriented context across multiple steps. Furthermore, we propose SG-LLM, a state-of-the-art approach leveraging a stepwise grounding paradigm to tackle the sequential grounding task. Our findings underscore the need for further research to advance the development of more capable and context-aware embodied agents.
Authors: Bohao Peng, Jian Wang, Yuechen Zhang, Wenbo Li, Ming-Chang Yang, Jiaya Jia
Abstract: Diffusion models have demonstrated remarkable and robust abilities in both image and video generation. To achieve greater control over generated results, researchers introduce additional architectures, such as ControlNet, Adapters and ReferenceNet, to integrate conditioning controls. However, current controllable generation methods often require substantial additional computational resources, especially for video generation, and face challenges in training or exhibit weak control. In this paper, we propose ControlNeXt: a powerful and efficient method for controllable image and video generation. We first design a more straightforward and efficient architecture, replacing heavy additional branches with minimal additional cost compared to the base model. Such a concise structure also allows our method to seamlessly integrate with other LoRA weights, enabling style alteration without the need for additional training. As for training, we reduce up to 90% of learnable parameters compared to the alternatives. Furthermore, we propose another method called Cross Normalization (CN) as a replacement for Zero-Convolution' to achieve fast and stable training convergence. We have conducted various experiments with different base models across images and videos, demonstrating the robustness of our method.
Authors: Zechen Bai, Tianjun Xiao, Tong He, Pichao Wang, Zheng Zhang, Thomas Brox, Mike Zheng Shou
Abstract: As online video content rapidly grows, the task of text-video retrieval (TVR) becomes increasingly important. A key challenge in TVR is the information asymmetry between video and text: videos are inherently richer in information, while their textual descriptions often capture only fragments of this complexity. This paper introduces a novel, data-centric framework to bridge this gap by enriching textual representations to better match the richness of video content. During training, videos are segmented into event-level clips and captioned to ensure comprehensive coverage. During retrieval, a large language model (LLM) generates semantically diverse queries to capture a broader range of possible matches. To enhance retrieval efficiency, we propose a query selection mechanism that identifies the most relevant and diverse queries, reducing computational cost while improving accuracy. Our method achieves state-of-the-art results across multiple benchmarks, demonstrating the power of data-centric approaches in addressing information asymmetry in TVR. This work paves the way for new research focused on leveraging data to improve cross-modal retrieval.
Authors: Xin Di, Long Peng, Peizhe Xia, Wenbo Li, Renjing Pei, Yang Cao, Yang Wang, Zheng-Jun Zha
Abstract: Burst super-resolution aims to reconstruct high-resolution images with higher quality and richer details by fusing the sub-pixel information from multiple burst low-resolution frames. In BusrtSR, the key challenge lies in extracting the base frame's content complementary sub-pixel details while simultaneously suppressing high-frequency noise disturbance. Existing methods attempt to extract sub-pixels by modeling inter-frame relationships frame by frame while overlooking the mutual correlations among multi-current frames and neglecting the intra-frame interactions, leading to inaccurate and noisy sub-pixels for base frame super-resolution. Further, existing methods mainly employ static upsampling with fixed parameters to improve spatial resolution for all scenes, failing to perceive the sub-pixel distribution difference across multiple frames and cannot balance the fusion weights of different frames, resulting in over-smoothed details and artifacts. To address these limitations, we introduce a novel Query Mamba Burst Super-Resolution (QMambaBSR) network, which incorporates a Query State Space Model (QSSM) and Adaptive Up-sampling module (AdaUp). Specifically, based on the observation that sub-pixels have consistent spatial distribution while random noise is inconsistently distributed, a novel QSSM is proposed to efficiently extract sub-pixels through inter-frame querying and intra-frame scanning while mitigating noise interference in a single step. Moreover, AdaUp is designed to dynamically adjust the upsampling kernel based on the spatial distribution of multi-frame sub-pixel information in the different burst scenes, thereby facilitating the reconstruction of the spatial arrangement of high-resolution details. Extensive experiments on four popular synthetic and real-world benchmarks demonstrate that our method achieves a new state-of-the-art performance.
Authors: Ce Wang, Zhenyu Hu, Wanjie Sun, Zhenzhong Chen
Abstract: Image rescaling aims to learn the optimal low-resolution (LR) image that can be accurately reconstructed to its original high-resolution (HR) counterpart, providing an efficient image processing and storage method for ultra-high definition media. However, extreme downscaling factors pose significant challenges to the upscaling process due to its highly ill-posed nature, causing existing image rescaling methods to struggle in generating semantically correct structures and perceptual friendly textures. In this work, we propose a novel framework called Timestep-Aware Diffusion Model (TADM) for extreme image rescaling, which performs rescaling operations in the latent space of a pre-trained autoencoder and effectively leverages powerful natural image priors learned by a pre-trained text-to-image diffusion model. Specifically, TADM adopts a pseudo-invertible module to establish the bidirectional mapping between the latent features of the HR image and the target-sized LR image. Then, the rescaled latent features are enhanced by a pre-trained diffusion model to generate more faithful details. Considering the spatially non-uniform degradation caused by the rescaling operation, we propose a novel time-step alignment strategy, which can adaptively allocate the generative capacity of the diffusion model based on the quality of the reconstructed latent features. Extensive experiments demonstrate the superiority of TADM over previous methods in both quantitative and qualitative evaluations.
Authors: Youjun Zhao, Jiaying Lin, Shuquan Ye, Qianshi Pang, Rynson W. H. Lau
Abstract: Open-vocabulary 3D scene understanding (OV-3D) aims to localize and classify novel objects beyond the closed set of object classes. However, existing approaches and benchmarks primarily focus on the open vocabulary problem within the context of object classes, which is insufficient in providing a holistic evaluation to what extent a model understands the 3D scene. In this paper, we introduce a more challenging task called Generalized Open-Vocabulary 3D Scene Understanding (GOV-3D) to explore the open vocabulary problem beyond object classes. It encompasses an open and diverse set of generalized knowledge, expressed as linguistic queries of fine-grained and object-specific attributes. To this end, we contribute a new benchmark named \textit{OpenScan}, which consists of 3D object attributes across eight representative linguistic aspects, including affordance, property, and material. We further evaluate state-of-the-art OV-3D methods on our OpenScan benchmark and discover that these methods struggle to comprehend the abstract vocabularies of the GOV-3D task, a challenge that cannot be addressed simply by scaling up object classes during training. We highlight the limitations of existing methodologies and explore promising directions to overcome the identified shortcomings.
Authors: Guoting Wei, Xia Yuan, Yu Liu, Zhenhao Shang, Xizhe Xue, Peng Wang, Kelu Yao, Chunxia Zhao, Haokui Zhang, Rong Xiao
Abstract: Aerial object detection plays a crucial role in numerous applications. However, most existing methods focus on detecting predefined object categories, limiting their applicability in real-world open scenarios. In this paper, we extend aerial object detection to open scenarios through image-text collaboration and propose OVA-Det, a highly efficient open-vocabulary detector for aerial scenes. Specifically, we first introduce an image-to-text alignment loss to replace the conventional category regression loss, thereby eliminating category limitations. Next, we propose a lightweight text-guided strategy that enhances the feature extraction process in the encoder and enables queries to focus on class-relevant image features within the decoder, further improving detection accuracy without introducing significant additional costs. Extensive comparison experiments demonstrate that the proposed OVA-Det outperforms state-of-the-art methods on all three widely used benchmark datasets by a large margin. For instance, for zero-shot detection on DIOR, OVA-Det achieves 37.2 mAP and 79.8 Recall, 12.4 and 42.0 higher than that of YOLO-World. In addition, the inference speed of OVA-Det reaches 36 FPS on RTX 4090, meeting the real-time detection requirements for various applications. The code is available at \href{https://github.com/GT-Wei/OVA-Det}{https://github.com/GT-Wei/OVA-Det}.
URLs: https://github.com/GT-Wei/OVA-Det, https://github.com/GT-Wei/OVA-Det
Authors: Xiaozhe Zhang, Fengying Xie, Haidong Ding, Linpeng Pan, Zhenwei Shi
Abstract: Haze removal aims to restore a clear image from a hazy input. Existing methods have shown significant efficacy by capturing either short-range dependencies for local detail preservation or long-range dependencies for global context modeling. Given the complementary strengths of both approaches, a intuitive advancement is to explicitly integrate them into a unified framework. However, this potential remains underexplored in current research. In this paper, we propose \textbf{DehazeMatic}, which leverages the proposed Transformer-Mamba Dual Aggregation block to simultaneously and explicitly captures both short- and long-range dependencies through dual-path design for improved restoration. To ensure that dependencies at varying ranges contribute optimally to performance, we conduct extensive experiments to identify key influencing factors and determine that an effective aggregation mechanism should be guided by the joint consideration of haze density and semantic information. Building on these insights, we introduce the CLIP-enhanced Dual-path Aggregator, which utilizes the rich semantic priors encapsulated in CLIP and the estimated haze density map, derived from its powerful generalization ability, to instruct the aggregation process. Extensive experiments demonstrate that DehazeMatic outperforms sort-of-the-art methods across various benchmarks.
Authors: George Tang, William Zhao, Logan Ford, David Benhaim, Paul Zhang
Abstract: We propose Segment Any Mesh, a novel zero-shot mesh part segmentation method that overcomes the limitations of shape analysis-based, learning-based, and contemporary approaches. Our approach operates in two phases: multimodal rendering and 2D-to-3D lifting. In the first phase, multiview renders of the mesh are individually processed through Segment Anything to generate 2D masks. These masks are then lifted into a mesh part segmentation by associating masks that refer to the same mesh part across the multiview renders. We find that applying Segment Anything to multimodal feature renders of normals and shape diameter scalars achieves better results than using only untextured renders of meshes. By building our method on top of Segment Anything, we seamlessly inherit any future improvements made to 2D segmentation. We compare our method with a robust, well-evaluated shape analysis method, Shape Diameter Function, and show that our method is comparable to or exceeds its performance. Since current benchmarks contain limited object diversity, we also curate and release a dataset of generated meshes and use it to demonstrate our method's improved generalization over Shape Diameter Function via human evaluation. We release the code and dataset at https://github.com/gtangg12/samesh
Authors: Baichuan Zhou, Haote Yang, Dairong Chen, Junyan Ye, Tianyi Bai, Jinhua Yu, Songyang Zhang, Dahua Lin, Conghui He, Weijia Li
Abstract: Recent evaluations of Large Multimodal Models (LMMs) have explored their capabilities in various domains, with only few benchmarks specifically focusing on urban environments. Moreover, existing urban benchmarks have been limited to evaluating LMMs with basic region-level urban tasks under singular views, leading to incomplete evaluations of LMMs' abilities in urban environments. To address these issues, we present UrBench, a comprehensive benchmark designed for evaluating LMMs in complex multi-view urban scenarios. UrBench contains 11.6K meticulously curated questions at both region-level and role-level that cover 4 task dimensions: Geo-Localization, Scene Reasoning, Scene Understanding, and Object Understanding, totaling 14 task types. In constructing UrBench, we utilize data from existing datasets and additionally collect data from 11 cities, creating new annotations using a cross-view detection-matching method. With these images and annotations, we then integrate LMM-based, rule-based, and human-based methods to construct large-scale high-quality questions. Our evaluations on 21 LMMs show that current LMMs struggle in the urban environments in several aspects. Even the best performing GPT-4o lags behind humans in most tasks, ranging from simple tasks such as counting to complex tasks such as orientation, localization and object attribute recognition, with an average performance gap of 17.4%. Our benchmark also reveals that LMMs exhibit inconsistent behaviors with different urban views, especially with respect to understanding cross-view relations.
Authors: Martin Obschonka, Christian Fisch, Tharindu Fernando, Clinton Fookes
Abstract: Occupational outcomes like entrepreneurship are generally considered personal information that individuals should have the autonomy to disclose. With the advancing capability of artificial intelligence (AI) to infer private details from widely available human-centric data (e.g., social media), it is crucial to investigate whether AI can accurately extract private occupational information from such data. In this study, we demonstrate that deep neural networks can classify individuals as entrepreneurs with high accuracy based on facial images sourced from Crunchbase, a premier source for entrepreneurship data. Utilizing a dataset comprising facial images of 40,728 individuals, including both entrepreneurs and non-entrepreneurs, we train a Convolutional Neural Network (CNN) using a contrastive learning approach based on pairs of facial images (one entrepreneur and one non-entrepreneur per pair). While human experts (n=650) and trained participants (n=133) were unable to classify entrepreneurs with accuracy above chance levels (>50%), our AI model achieved a classification accuracy of 79.51%. Several robustness tests indicate that this high level of accuracy is maintained under various conditions. These results indicate privacy risks for entrepreneurs.
Authors: Nan Chen, Mengqi Huang, Zhuowei Chen, Yang Zheng, Lei Zhang, Zhendong Mao
Abstract: Subject-driven text-to-image (T2I) customization has drawn significant interest in academia and industry. This task enables pre-trained models to generate novel images based on unique subjects. Existing studies adopt a self-reconstructive perspective, focusing on capturing all details of a single image, which will misconstrue the specific image's irrelevant attributes (e.g., view, pose, and background) as the subject intrinsic attributes. This misconstruction leads to both overfitting or underfitting of irrelevant and intrinsic attributes of the subject, i.e., these attributes are over-represented or under-represented simultaneously, causing a trade-off between similarity and controllability. In this study, we argue an ideal subject representation can be achieved by a cross-differential perspective, i.e., decoupling subject intrinsic attributes from irrelevant attributes via contrastive learning, which allows the model to focus more on intrinsic attributes through intra-consistency (features of the same subject are spatially closer) and inter-distinctiveness (features of different subjects have distinguished differences). Specifically, we propose CustomContrast, a novel framework, which includes a Multilevel Contrastive Learning (MCL) paradigm and a Multimodal Feature Injection (MFI) Encoder. The MCL paradigm is used to extract intrinsic features of subjects from high-level semantics to low-level appearance through crossmodal semantic contrastive learning and multiscale appearance contrastive learning. To facilitate contrastive learning, we introduce the MFI encoder to capture cross-modal representations. Extensive experiments show the effectiveness of CustomContrast in subject similarity and text controllability.
Authors: Guoyang Zhao, Yuxuan Liu, Weiqing Qi, Fulong Ma, Ming Liu, Jun Ma
Abstract: Accurate depth estimation is crucial for 3D scene comprehension in robotics and autonomous vehicles. Fisheye cameras, known for their wide field of view, have inherent geometric benefits. However, their use in depth estimation is restricted by a scarcity of ground truth data and image distortions. We present FisheyeDepth, a self-supervised depth estimation model tailored for fisheye cameras. We incorporate a fisheye camera model into the projection and reprojection stages during training to handle image distortions, thereby improving depth estimation accuracy and training stability. Furthermore, we incorporate real-scale pose information into the geometric projection between consecutive frames, replacing the poses estimated by the conventional pose network. Essentially, this method offers the necessary physical depth for robotic tasks, and also streamlines the training and inference procedures. Additionally, we devise a multi-channel output strategy to improve robustness by adaptively fusing features at various scales, which reduces the noise from real pose data. We demonstrate the superior performance and robustness of our model in fisheye image depth estimation through evaluations on public datasets and real-world scenarios. The project website is available at: https://github.com/guoyangzhao/FisheyeDepth.
Authors: Guoyang Zhao, Fulong Ma, Weiqing Qi, Chenguang Zhang, Yuxuan Liu, Ming Liu, Jun Ma
Abstract: Traffic sign is a critical map feature for navigation and traffic control. Nevertheless, current methods for traffic sign recognition rely on traditional deep learning models, which typically suffer from significant performance degradation considering the variations in data distribution across different regions. In this paper, we propose TSCLIP, a robust fine-tuning approach with the contrastive language-image pre-training (CLIP) model for worldwide cross-regional traffic sign recognition. We first curate a cross-regional traffic sign benchmark dataset by combining data from ten different sources. Then, we propose a prompt engineering scheme tailored to the characteristics of traffic signs, which involves specific scene descriptions and corresponding rules to generate targeted text descriptions. During the TSCLIP fine-tuning process, we implement adaptive dynamic weight ensembling (ADWE) to seamlessly incorporate outcomes from each training iteration with the zero-shot CLIP model. This approach ensures that the model retains its ability to generalize while acquiring new knowledge about traffic signs. To the best knowledge of authors, TSCLIP is the first contrastive language-image model used for the worldwide cross-regional traffic sign recognition task. The project website is available at: https://github.com/guoyangzhao/TSCLIP.
Authors: Dimitrije Anti\'c, Georgios Paschalidis, Shashank Tripathi, Theo Gevers, Sai Kumar Dwivedi, Dimitrios Tzionas
Abstract: Recovering 3D object pose and shape from a single image is a challenging and highly ill-posed problem. This is due to strong (self-)occlusions, depth ambiguities, the vast intra- and inter-class shape variance, and lack of 3D ground truth for natural images. While existing methods train deep networks on synthetic datasets to predict 3D shapes, they often struggle to generalize to real-world scenarios, lack an explicit feedback loop for refining noisy estimates, and primarily focus on geometry without explicitly considering pixel alignment. To this end, we make two key observations: (1) a robust solution requires a model that imposes a strong category-specific shape prior to constrain the search space, and (2) foundational models embed 2D images and 3D shapes in joint spaces; both help resolve ambiguities. Hence, we propose SDFit, a novel optimization framework that is built on three key innovations: First, we use a learned morphable signed-distance-function (mSDF) model that acts as a strong shape prior, thus constraining the shape space. Second, we use foundational models to establish rich 2D-to-3D correspondences between image features and the mSDF. Third, we develop a fitting pipeline that iteratively refines both shape and pose, aligning the mSDF to the image. We evaluate SDFit on the Pix3D, Pascal3D+, and COMIC image datasets. SDFit performs on par with SotA methods, while demonstrating exceptional robustness to occlusions and requiring no retraining for unseen images. Therefore, SDFit contributes new insights for generalizing in the wild, paving the way for future research. Code will be released.
Authors: Dapeng Feng, Zhiqiang Chen, Yizhen Yin, Shipeng Zhong, Yuhua Qi, Hongbo Chen
Abstract: Simultaneous Localization and Mapping (SLAM) is pivotal in robotics, with photorealistic scene reconstruction emerging as a key challenge. To address this, we introduce Computational Alignment for Real-Time Gaussian Splatting SLAM (CaRtGS), a novel method enhancing the efficiency and quality of photorealistic scene reconstruction in real-time environments. Leveraging 3D Gaussian Splatting (3DGS), CaRtGS achieves superior rendering quality and processing speed, which is crucial for scene photorealistic reconstruction. Our approach tackles computational misalignment in Gaussian Splatting SLAM (GS-SLAM) through an adaptive strategy that enhances optimization iterations, addresses long-tail optimization, and refines densification. Experiments on Replica, TUM-RGBD, and VECtor datasets demonstrate CaRtGS's effectiveness in achieving high-fidelity rendering with fewer Gaussian primitives. This work propels SLAM towards real-time, photorealistic dense rendering, significantly advancing photorealistic scene representation. For the benefit of the research community, we release the code and accompanying videos on our project website: https://dapengfeng.github.io/cartgs.
Authors: Zongming Li, Tianheng Cheng, Shoufa Chen, Peize Sun, Haocheng Shen, Longjin Ran, Xiaoxin Chen, Wenyu Liu, Xinggang Wang
Abstract: Autoregressive (AR) models have reformulated image generation as next-token prediction, demonstrating remarkable potential and emerging as strong competitors to diffusion models. However, control-to-image generation, akin to ControlNet, remains largely unexplored within AR models. Although a natural approach, inspired by advancements in Large Language Models, is to tokenize control images into tokens and prefill them into the autoregressive model before decoding image tokens, it still falls short in generation quality compared to ControlNet and suffers from inefficiency. To this end, we introduce ControlAR, an efficient and effective framework for integrating spatial controls into autoregressive image generation models. Firstly, we explore control encoding for AR models and propose a lightweight control encoder to transform spatial inputs (e.g., canny edges or depth maps) into control tokens. Then ControlAR exploits the conditional decoding method to generate the next image token conditioned on the per-token fusion between control and image tokens, similar to positional encodings. Compared to prefilling tokens, using conditional decoding significantly strengthens the control capability of AR models but also maintains the model's efficiency. Furthermore, the proposed ControlAR surprisingly empowers AR models with arbitrary-resolution image generation via conditional decoding and specific controls. Extensive experiments can demonstrate the controllability of the proposed ControlAR for the autoregressive control-to-image generation across diverse inputs, including edges, depths, and segmentation masks. Furthermore, both quantitative and qualitative results indicate that ControlAR surpasses previous state-of-the-art controllable diffusion models, e.g., ControlNet++. Code, models, and demo will soon be available at https://github.com/hustvl/ControlAR.
Authors: Ashish Kumar, Jaesik Park
Abstract: In the era of vision Transformers, the recent success of VanillaNet shows the huge potential of simple and concise convolutional neural networks (ConvNets). Where such models mainly focus on runtime, it is also crucial to simultaneously focus on other aspects, e.g., FLOPs, parameters, etc, to strengthen their utility further. To this end, we introduce a refreshing ConvNet macro design called Columnar Stage Network (CoSNet). CoSNet has a systematically developed simple and concise structure, smaller depth, low parameter count, low FLOPs, and attention-less operations, well suited for resource-constrained deployment. The key novelty of CoSNet is deploying parallel convolutions with fewer kernels fed by input replication, using columnar stacking of these convolutions, and minimizing the use of 1x1 convolution layers. Our comprehensive evaluations show that CoSNet rivals many renowned ConvNets and Transformer designs under resource-constrained scenarios. Code: https://github.com/ashishkumar822/CoSNet
Authors: Jianze Li, Jiezhang Cao, Zichen Zou, Xiongfei Su, Xin Yuan, Yulun Zhang, Yong Guo, Xiaokang Yang
Abstract: Diffusion models have demonstrated excellent performance for real-world image super-resolution (Real-ISR), albeit at high computational costs. Most existing methods are trying to derive one-step diffusion models from multi-step counterparts through knowledge distillation (KD) or variational score distillation (VSD). However, these methods are limited by the capabilities of the teacher model, especially if the teacher model itself is not sufficiently strong. To tackle these issues, we propose a new One-Step \textbf{D}iffusion model with a larger-scale \textbf{D}iffusion \textbf{D}iscriminator for SR, called D$^3$SR. Our discriminator is able to distill noisy features from any time step of diffusion models in the latent space. In this way, our diffusion discriminator breaks through the potential limitations imposed by the presence of a teacher model. Additionally, we improve the perceptual loss with edge-aware DISTS (EA-DISTS) to enhance the model's ability to generate fine details. Our experiments demonstrate that, compared with previous diffusion-based methods requiring dozens or even hundreds of steps, our D$^3$SR attains comparable or even superior results in both quantitative metrics and qualitative evaluations. Moreover, compared with other methods, D$^3$SR achieves at least $3\times$ faster inference speed and reduces parameters by at least 30\%. We will release code and models at https://github.com/JianzeLi-114/D3SR.
Authors: Yixiang Qiu, Hongyao Yu, Hao Fang, Tianqu Zhuang, Wenbo Yu, Bin Chen, Xuan Wang, Shu-Tao Xia, Ke Xu
Abstract: Model Inversion (MI) attacks aim at leveraging the output information of target models to reconstruct privacy-sensitive training data, raising critical concerns regarding the privacy vulnerabilities of Deep Neural Networks (DNNs). Unfortunately, in tandem with the rapid evolution of MI attacks, the absence of a comprehensive benchmark with standardized metrics and reproducible implementations has emerged as a formidable challenge. This deficiency has hindered objective comparison of methodological advancements and reliable assessment of defense efficacy. To address this critical gap, we build the first practical benchmark named MIBench for systematic evaluation of model inversion attacks and defenses. This benchmark bases on an extensible and reproducible modular-based toolbox which currently integrates a total of 19 state-of-the-art attack and defense methods and encompasses 9 standardized evaluation protocols. Capitalizing on this foundation, we conduct extensive evaluation from multiple perspectives to holistically compare and analyze various methods across different scenarios, such as the impact of target resolution, model predictive power, defense performance and adversarial robustness.
Authors: Mingxuan Liu, Zhun Zhong, Jun Li, Gianni Franchi, Subhankar Roy, Elisa Ricci
Abstract: Organizing unstructured visual data into semantic clusters is a key challenge in computer vision. Traditional deep clustering approaches focus on a single partition of data, while multiple clustering (MC) methods address this limitation by uncovering distinct clustering solutions. The rise of large language models (LLMs) and multimodal LLMs has enhanced MC by allowing users to define text clustering criteria. However, expecting users to manually define such criteria for large datasets before understanding the data is impractical. In this work, we introduce the task of Open-ended Semantic Multiple Clustering, that aims to automatically discover clustering criteria from large, unstructured image collections, uncovering interpretable substructures without requiring human input. Our framework, X-Cluster: eXploratory Clustering, uses text as a proxy to concurrently reason over large image collections, discover partitioning criteria, expressed in natural language, and reveal semantic substructures. To evaluate X-Cluster, we introduce the COCO-4c and Food-4c benchmarks, each containing four grouping criteria and ground-truth annotations. We apply X-Cluster to various real-world applications, such as discovering biases and analyzing social media image popularity, demonstrating its utility as a practical tool for organizing large unstructured image collections and revealing novel insights. We will open-source our code and benchmarks for reproducibility and future research.
Authors: Saemi Moon, Minjong Lee, Sangdon Park, Dongwoo Kim
Abstract: As text-to-image diffusion models gain widespread commercial applications, there are increasing concerns about unethical or harmful use, including the unauthorized generation of copyrighted or sensitive content. Concept unlearning has emerged as a promising solution to these challenges by removing undesired and harmful information from the pre-trained model. However, the previous evaluations primarily focus on whether target concepts are removed while preserving image quality, neglecting the broader impacts such as unintended side effects. In this work, we propose Holistic Unlearning Benchmark (HUB), a comprehensive framework for evaluating unlearning methods across six key dimensions: faithfulness, alignment, pinpoint-ness, multilingual robustness, attack robustness, and efficiency. Our benchmark covers 33 target concepts, including 16,000 prompts per concept, spanning four categories: Celebrity, Style, Intellectual Property, and NSFW. Our investigation reveals that no single method excels across all evaluation criteria. By releasing our evaluation code and dataset, we hope to inspire further research in this area, leading to more reliable and effective unlearning methods.
Authors: Zhe Li, Weihao Yuan, Yisheng He, Lingteng Qiu, Shenhao Zhu, Xiaodong Gu, Weichao Shen, Yuan Dong, Zilong Dong, Laurence T. Yang
Abstract: Language plays a vital role in the realm of human motion. Existing methods have largely depended on CLIP text embeddings for motion generation, yet they fall short in effectively aligning language and motion due to CLIP's pretraining on static image-text pairs. This work introduces LaMP, a novel Language-Motion Pretraining model, which transitions from a language-vision to a more suitable language-motion latent space. It addresses key limitations by generating motion-informative text embeddings, significantly enhancing the relevance and semantics of generated motion sequences. With LaMP, we advance three key tasks: text-to-motion generation, motion-text retrieval, and motion captioning through aligned language-motion representation learning. For generation, we utilize LaMP to provide the text condition instead of CLIP, and an autoregressive masked prediction is designed to achieve mask modeling without rank collapse in transformers. For retrieval, motion features from LaMP's motion transformer interact with query tokens to retrieve text features from the text transformer, and vice versa. For captioning, we finetune a large language model with the language-informative motion features to develop a strong motion captioning model. In addition, we introduce the LaMP-BertScore metric to assess the alignment of generated motions with textual descriptions. Extensive experimental results on multiple datasets demonstrate substantial improvements over previous methods across all three tasks. The code of our method will be made public.
Authors: Xiaoling Hu, Karthik Gopinath, Peirong Liu, Malte Hoffmann, Koen Van Leemput, Oula Puonti, Juan Eugenio Iglesias
Abstract: Over recent years, deep learning based image registration has achieved impressive accuracy in many domains, including medical imaging and, specifically, human neuroimaging with magnetic resonance imaging (MRI). However, the uncertainty estimation associated with these methods has been largely limited to the application of generic techniques (e.g., Monte Carlo dropout) that do not exploit the peculiarities of the problem domain, particularly spatial modeling. Here, we propose a principled way to propagate uncertainties (epistemic or aleatoric) estimated at the level of spatial location by these methods, to the level of global transformation models, and further to downstream tasks. Specifically, we justify the choice of a Gaussian distribution for the local uncertainty modeling, and then propose a framework where uncertainties spread across hierarchical levels, depending on the choice of transformation model. Experiments on publicly available data sets show that Monte Carlo dropout correlates very poorly with the reference registration error, whereas our uncertainty estimates correlate much better. Crucially, the results also show that uncertainty-aware fitting of transformations improves the registration accuracy of brain MRI scans. Finally, we illustrate how sampling from the posterior distribution of the transformations can be used to propagate uncertainties to downstream neuroimaging tasks. Code is available at: https://github.com/HuXiaoling/Regre4Regis.
Authors: Xirui Li, Charles Herrmann, Kelvin C. K. Chan, Yinxiao Li, Deqing Sun, Chao Ma, Ming-Hsuan Yang
Abstract: Recent progress in image generation has sparked research into controlling these models through condition signals, with various methods addressing specific challenges in conditional generation. Instead of proposing another specialized technique, we introduce a simple, unified framework to handle diverse conditional generation tasks involving a specific image-condition correlation. By learning a joint distribution over a correlated image pair (e.g. image and depth) with a diffusion model, our approach enables versatile capabilities via different inference-time sampling schemes, including controllable image generation (e.g. depth to image), estimation (e.g. image to depth), signal guidance, joint generation (image & depth), and coarse control. Previous attempts at unification often introduce significant complexity through multi-stage training, architectural modification, or increased parameter counts. In contrast, our simple formulation requires a single, computationally efficient training stage, maintains the standard model input, and adds minimal learned parameters (15% of the base model). Moreover, our model supports additional capabilities like non-spatially aligned and coarse conditioning. Extensive results show that our single model can produce comparable results with specialized methods and better results than prior unified methods. We also demonstrate that multiple models can be effectively combined for multi-signal conditional generation.
Authors: Hao He, Yixun Liang, Luozhou Wang, Yuanhao Cai, Xinli Xu, Hao-Xiang Guo, Xiang Wen, Yingcong Chen
Abstract: Recent large reconstruction models have made notable progress in generating high-quality 3D objects from single images. However, current reconstruction methods often rely on explicit camera pose estimation or fixed viewpoints, restricting their flexibility and practical applicability. We reformulate 3D reconstruction as image-to-image translation and introduce the Relative Coordinate Map (RCM), which aligns multiple unposed images to a main view without pose estimation. While RCM simplifies the process, its lack of global 3D supervision can yield noisy outputs. To address this, we propose Relative Coordinate Gaussians (RCG) as an extension to RCM, which treats each pixel's coordinates as a Gaussian center and employs differentiable rasterization for consistent geometry and pose recovery. Our LucidFusion framework handles an arbitrary number of unposed inputs, producing robust 3D reconstructions within seconds and paving the way for more flexible, pose-free 3D pipelines.
Authors: Kevis-Kokitsi Maninis, Kaifeng Chen, Soham Ghosh, Arjun Karpur, Koert Chen, Ye Xia, Bingyi Cao, Daniel Salz, Guangxing Han, Jan Dlabal, Dan Gnanapragasam, Mojtaba Seyedhosseini, Howard Zhou, Andre Araujo
Abstract: While image-text representation learning has become very popular in recent years, existing models tend to lack spatial awareness and have limited direct applicability for dense understanding tasks. For this reason, self-supervised image-only pretraining is still the go-to method for many dense vision applications (e.g. depth estimation, semantic segmentation), despite the lack of explicit supervisory signals. In this paper, we close this gap between image-text and self-supervised learning, by proposing a novel general-purpose image-text model, which can be effectively used off the shelf for dense and global vision tasks. Our method, which we refer to as Text-Image Pretraining with Spatial awareness (TIPS), leverages two simple and effective insights. First, on textual supervision: we reveal that replacing noisy web image captions by synthetically generated textual descriptions boosts dense understanding performance significantly, due to a much richer signal for learning spatially aware representations. We propose an adapted training method that combines noisy and synthetic captions, resulting in improvements across both dense and global understanding tasks. Second, on the learning technique: we propose to combine contrastive image-text learning with self-supervised masked image modeling, to encourage spatial coherence, unlocking substantial enhancements for downstream applications. Building on these two ideas, we scale our model using the transformer architecture, trained on a curated set of public images. Our experiments are conducted on 8 tasks involving 16 datasets in total, demonstrating strong off-the-shelf performance on both dense and global understanding, for several image-only and image-text tasks. Code and models are released at https://github.com/google-deepmind/tips.
Authors: Chuanxiang Yang, Yuanfeng Zhou, Guangshun Wei, Long Ma, Junhui Hou, Yuan Liu, Wenping Wang
Abstract: As commonly used implicit geometry representations, the signed distance function (SDF) is limited to modeling watertight shapes, while the unsigned distance function (UDF) is capable of representing various surfaces. However, its inherent theoretical shortcoming, i.e., the non-differentiability at the zero level set, would result in sub-optimal reconstruction quality. In this paper, we propose the scaled-squared distance function (S$^{2}$DF), a novel implicit surface representation for modeling arbitrary surface types. S$^{2}$DF does not distinguish between inside and outside regions while effectively addressing the non-differentiability issue of UDF at the zero level set. We demonstrate that S$^{2}$DF satisfies a second-order partial differential equation of Monge-Ampere-type, allowing us to develop a learning pipeline that leverages a novel Monge-Ampere regularization to directly learn S$^{2}$DF from raw unoriented point clouds without supervision from ground-truth S$^{2}$DF values. Extensive experiments across multiple datasets show that our method significantly outperforms state-of-the-art supervised approaches that require ground-truth surface information as supervision for training. The code will be publicly available at https://github.com/chuanxiang-yang/S2DF.
Authors: Xupeng Chen, Zhixin Lai, Kangrui Ruan, Shichu Chen, Jiaxiang Liu, Zuozhu Liu
Abstract: Artificial intelligence has made significant strides in medical visual question answering (Med-VQA), yet prevalent studies often interpret images holistically, overlooking the visual regions of interest that may contain crucial information, potentially aligning with a doctor's prior knowledge that can be incorporated with minimal annotations (e.g., bounding boxes). To address this gap, this paper introduces R-LLaVA, designed to enhance biomedical VQA understanding by integrating simple medical annotations as prior knowledge directly into the image space through CLIP. These annotated visual regions of interest are then fed into the LLaVA model during training, aiming to enrich the model's understanding of biomedical queries. Experimental evaluation on four standard Med-VQA datasets demonstrates R-LLaVA's superiority over existing state-of-the-art (SoTA) methods. Additionally, to verify the model's capability in visual comprehension, a novel multiple-choice medical visual understanding dataset is introduced, confirming the positive impact of focusing on visual regions of interest in advancing biomedical VQA understanding.
Authors: Panagiota Gatoula, Dimitrios E. Diamantis, Anastasios Koulaouzidis, Cristina Carretero, Stefania Chetcuti-Zammit, Pablo Cortegoso Valdivia, Bego\~na Gonz\'alez-Su\'arez, Alessandro Mussetto, John Plevris, Alexander Robertson, Bruno Rosa, Ervin Toth, Dimitris K. Iakovidis
Abstract: Synthetic Data Generation (SDG) based on Artificial Intelligence (AI) can transform the way clinical medicine is delivered by overcoming privacy barriers that currently render clinical data sharing difficult. This is the key to accelerating the development of digital tools contributing to enhanced patient safety. Such tools include robust data-driven clinical decision support systems, and example-based digital training tools that will enable healthcare professionals to improve their diagnostic performance for enhanced patient safety. This study focuses on the clinical evaluation of medical SDG, with a proof-of-concept investigation on diagnosing Inflammatory Bowel Disease (IBD) using Wireless Capsule Endoscopy (WCE) images. Its scientific contributions include a) a novel protocol for the systematic Clinical Evaluation of Medical Image Synthesis (CEMIS); b) a novel variational autoencoder-based model for the generation of high-resolution synthetic WCE images; and c) a comprehensive evaluation of the synthetic images using the CEMIS protocol by 10 international WCE specialists, in terms of image quality, diversity, and realism, as well as their utility for clinical decision-making. The results show that TIDE-II generates clinically plausible, very realistic WCE images, of improved quality compared to relevant state-of-the-art generative models. Concludingly, CEMIS can serve as a reference for future research on medical image-generation techniques, while the adaptation/extension of the architecture of TIDE-II to other imaging domains can be promising.
Authors: Ruofan Wang, Juncheng Li, Yixu Wang, Bo Wang, Xiaosen Wang, Yan Teng, Yingchun Wang, Xingjun Ma, Yu-Gang Jiang
Abstract: As large Vision-Language Models (VLMs) gain prominence, ensuring their safe deployment has become critical. Recent studies have explored VLM robustness against jailbreak attacks-techniques that exploit model vulnerabilities to elicit harmful outputs. However, the limited availability of diverse multimodal data has constrained current approaches to rely heavily on adversarial or manually crafted images derived from harmful text datasets, which often lack effectiveness and diversity across different contexts. In this paper, we propose IDEATOR, a novel jailbreak method that autonomously generates malicious image-text pairs for black-box jailbreak attacks. IDEATOR is grounded in the insight that VLMs themselves could serve as powerful red team models for generating multimodal jailbreak prompts. Specifically, IDEATOR leverages a VLM to create targeted jailbreak texts and pairs them with jailbreak images generated by a state-of-the-art diffusion model. Extensive experiments demonstrate IDEATOR's high effectiveness and transferability, achieving a 94% attack success rate (ASR) in jailbreaking MiniGPT-4 with an average of only 5.34 queries, and high ASRs of 82%, 88%, and 75% when transferred to LLaVA, InstructBLIP, and Chameleon, respectively. Building on IDEATOR's strong transferability and automated process, we introduce the VLBreakBench, a safety benchmark comprising 3,654 multimodal jailbreak samples. Our benchmark results on 11 recently released VLMs reveal significant gaps in safety alignment. For instance, our challenge set achieves ASRs of 46.31% on GPT-4o and 19.65% on Claude-3.5-Sonnet, underscoring the urgent need for stronger defenses.
Authors: Xiujin Zhu, Chee-Onn Chow, Joon Huang Chuah
Abstract: Image shadow removal is a common low-level vision problem. Shadows cause sudden brightness changes in some areas, which can affect the accuracy of downstream tasks. Currently, Transformer-based shadow removal methods improve computational efficiency by using a window mechanism. However, this approach reduces the effective receptive field and weakens the ability to model long-range dependencies in shadow images. Recently, Mamba has achieved significant success in computer vision by modeling long-sequence information globally with linear complexity. However, when applied to shadow removal, its original scanning mechanism overlooks the semantic continuity along shadow boundaries, and the coherence within each region. To solve this issue, we propose a new boundary-region selective scanning mechanism that scans shadow, boundary, and non-shadow regions separately, making pixels of the same type closer in the sequence. This increases semantic continuity and helps the model understand local details better. Incorporating this idea, we design the first Mamba-based lightweight shadow removal model, called ShadowMamba. It uses a hierarchical combination U-Net structure, which effectively reduces the number of parameters and computational complexity. Shallow layers rely on our boundary-region selective scanning to capture local details, while deeper layers use global cross-scanning to learn global brightness features. Extensive experiments show that ShadowMamba outperforms current state-of-the-art models on ISTD+, ISTD, and SRD datasets, and it also requires fewer parameters and less computational cost. (Code will be made available upon paper acceptance.)
Authors: Yutong Chen, Marko Mihajlovic, Xiyi Chen, Yiming Wang, Sergey Prokudin, Siyu Tang
Abstract: 3D Gaussian Splatting (3DGS) has recently transformed photorealistic reconstruction, achieving high visual fidelity and real-time performance. However, rendering quality significantly deteriorates when test views deviate from the camera angles used during training, posing a major challenge for applications in immersive free-viewpoint rendering and navigation. In this work, we conduct a comprehensive evaluation of 3DGS and related novel view synthesis methods under out-of-distribution (OOD) test camera scenarios. By creating diverse test cases with synthetic and real-world datasets, we demonstrate that most existing methods, including those incorporating various regularization techniques and data-driven priors, struggle to generalize effectively to OOD views. To address this limitation, we introduce SplatFormer, the first point transformer model specifically designed to operate on Gaussian splats. SplatFormer takes as input an initial 3DGS set optimized under limited training views and refines it in a single forward pass, effectively removing potential artifacts in OOD test views. To our knowledge, this is the first successful application of point transformers directly on 3DGS sets, surpassing the limitations of previous multi-scene training methods, which could handle only a restricted number of input views during inference. Our model significantly improves rendering quality under extreme novel views, achieving state-of-the-art performance in these challenging scenarios and outperforming various 3DGS regularization techniques, multi-scene models tailored for sparse view synthesis, and diffusion-based frameworks.
Authors: Xi Cheng, Ruiqi Lei, Di Huang, Zhichao Liao, Fengyuan Piao, Yan Chen, Pingfa Feng, Long Zeng
Abstract: Parametric point clouds are sampled from CAD shapes and are becoming increasingly common in industrial manufacturing. Most existing CAD-specific deep learning methods only focus on geometric features, while overlooking constraints which are inherent and important in CAD shapes. This limits their ability to discern CAD shapes with similar appearance but different constraints. To tackle this challenge, we first analyze the constraint importance via a simple validation experiment. Then, we introduce a deep learning-friendly constraints representation with three vectorized components, and design a constraint-aware feature learning network (CstNet), which includes two stages. Stage 1 extracts constraint feature from B-Rep data or point cloud based on shape local information. It enables better generalization ability to unseen dataset after model pre-training. Stage 2 employs attention layers to adaptively adjust the weights of three constraints' components. It facilitates the effective utilization of constraints. In addition, we built the first multi-modal parametric-purpose dataset, i.e. Param20K, comprising about 20K shape instances of 75 classes. On this dataset, we performed the classification and rotation robustness experiments, and CstNet achieved 3.52\% and 26.17\% absolute improvements in instance accuracy over the state-of-the-art methods, respectively. To the best of our knowledge, CstNet is the first constraint-aware deep learning method tailored for parametric point cloud analysis in CAD domain.
Authors: Wuzheng Dong
Abstract: This paper presents a method for object recognition and automatic labeling in large-area remote sensing images called LRSAA. The method integrates YOLOv11 and MobileNetV3-SSD object detection algorithms through ensemble learning to enhance model performance. Furthermore, it employs Poisson disk sampling segmentation techniques and the EIOU metric to optimize the training and inference processes of segmented images, followed by the integration of results. This approach not only reduces the demand for computational resources but also achieves a good balance between accuracy and speed. The source code for this project has been made publicly available on https://github.com/anaerovane/LRSAA.
Authors: David Svitov, Pietro Morerio, Lourdes Agapito, Alessio Del Bue
Abstract: We present billboard Splatting (BBSplat) - a novel approach for novel view synthesis based on textured geometric primitives. BBSplat represents the scene as a set of optimizable textured planar primitives with learnable RGB textures and alpha-maps to control their shape. BBSplat primitives can be used in any Gaussian Splatting pipeline as drop-in replacements for Gaussians. The proposed primitives close the rendering quality gap between 2D and 3D Gaussian Splatting (GS), enabling the accurate extraction of 3D mesh as in the 2DGS framework. Additionally, the explicit nature of planar primitives enables the use of the ray-tracing effects in rasterization. Our novel regularization term encourages textures to have a sparser structure, enabling an efficient compression that leads to a reduction in the storage space of the model up to x17 times compared to 3DGS. Our experiments show the efficiency of BBSplat on standard datasets of real indoor and outdoor scenes such as Tanks&Temples, DTU, and Mip-NeRF-360. Namely, we achieve a state-of-the-art PSNR of 29.72 for DTU at Full HD resolution.
Authors: Jun Xie, Wenxiao Li, Faqiang Wang, Liqiang Zhang, Zhengyang Hou, Jun Liu
Abstract: Morphological methods play a crucial role in remote sensing image processing, due to their ability to capture and preserve small structural details. However, most of the existing deep learning models for semantic segmentation are based on the encoder-decoder architecture including U-net and Segment Anything Model (SAM), where the downsampling process tends to discard fine details. In this paper, we propose a new approach that integrates learnable morphological skeleton prior into deep neural networks using the variational method. To address the difficulty in backpropagation in neural networks caused by the non-differentiability presented in classical morphological operations, we provide a smooth representation of the morphological skeleton and design a variational segmentation model integrating morphological skeleton prior by employing operator splitting and dual methods. Then, we integrate this model into the network architecture of SAM, which is achieved by adding a token to mask decoder and modifying the final sigmoid layer, ensuring the final segmentation results preserve the skeleton structure as much as possible. Experimental results on remote sensing datasets, including buildings, roads and water, demonstrate that our method outperforms the original SAM on slender object segmentation and exhibits better generalization capability.
Authors: Guanwen Feng, Zhihao Qian, Yunan Li, Siyu Jin, Qiguang Miao, Chi-Man Pun
Abstract: While existing one-shot talking head generation models have achieved progress in coarse-grained emotion editing, there is still a lack of fine-grained emotion editing models with high interpretability. We argue that for an approach to be considered fine-grained, it needs to provide clear definitions and sufficiently detailed differentiation. We present LES-Talker, a novel one-shot talking head generation model with high interpretability, to achieve fine-grained emotion editing across emotion types, emotion levels, and facial units. We propose a Linear Emotion Space (LES) definition based on Facial Action Units to characterize emotion transformations as vector transformations. We design the Cross-Dimension Attention Net (CDAN) to deeply mine the correlation between LES representation and 3D model representation. Through mining multiple relationships across different feature and structure dimensions, we enable LES representation to guide the controllable deformation of 3D model. In order to adapt the multimodal data with deviations to the LES and enhance visual quality, we utilize specialized network design and training strategies. Experiments show that our method provides high visual quality along with multilevel and interpretable fine-grained emotion editing, outperforming mainstream methods.
Authors: Huan Kang, Hui Li, Tianyang Xu, Rui Wang, Xiao-Jun Wu, Josef Kittler
Abstract: Euclidean representation learning methods have achieved commendable results in image fusion tasks, which can be attributed to their clear advantages in handling with linear space. However, data collected from a realistic scene usually have a non-Euclidean structure, where Euclidean metric might be limited in representing the true data relationships, degrading fusion performance. To address this issue, a novel SPD (symmetric positive definite) manifold learning framework is proposed for multi-modal image fusion, named SPDFusion, which extends the image fusion approach from the Euclidean space to the SPD manifolds. Specifically, we encode images according to the Riemannian geometry to exploit their intrinsic statistical correlations, thereby aligning with human visual perception. Actually, the SPD matrix underpins our network learning, with a cross-modal fusion strategy employed to harness modality-specific dependencies and augment complementary information. Subsequently, an attention module is designed to process the learned weight matrix, facilitating the weighting of spatial global correlation semantics via SPD matrix multiplication. Based on this, we design an end-to-end fusion network based on cross-modal manifold learning. Extensive experiments on public datasets demonstrate that our framework exhibits superior performance compared to the current state-of-the-art methods.
Authors: Qi Wang, Jinjia Zhou
Abstract: In previous studies on knowledge distillation, the significance of logit distillation has frequently been overlooked. To revitalize logit distillation, we present a novel perspective by reconsidering its computation based on the semantic properties of logits and exploring how to utilize it more efficiently. Logits often contain a substantial amount of high-level semantic information; however, the conventional approach of employing logits to compute Kullback-Leibler (KL) divergence does not account for their semantic properties. Furthermore, this direct KL divergence computation fails to fully exploit the potential of logits. To address these challenges, we introduce a novel and efficient logit distillation method, Multi-perspective Contrastive Logit Distillation (MCLD), which substantially improves the performance and efficacy of logit distillation. In comparison to existing logit distillation methods and complex feature distillation methods, MCLD attains state-of-the-art performance in image classification, and transfer learning tasks across multiple datasets, including CIFAR-100, ImageNet, Tiny-ImageNet, and STL-10. Additionally, MCLD exhibits superior training efficiency and outstanding performance with distilling on Vision Transformers, further emphasizing its notable advantages. This study unveils the vast potential of logits in knowledge distillation and seeks to offer valuable insights for future research.
Authors: Jeonghyeok Do, Munchurl Kim
Abstract: We firstly present a diffusion-based action recognition with zero-shot learning for skeleton inputs. In zero-shot skeleton-based action recognition, aligning skeleton features with the text features of action labels is essential for accurately predicting unseen actions. Previous methods focus on direct alignment between skeleton and text latent spaces, but the modality gaps between these spaces hinder robust generalization learning. Motivated from the remarkable performance of text-to-image diffusion models, we leverage their alignment capabilities between different modalities mostly by focusing on the training process during reverse diffusion rather than using their generative power. Based on this, our framework is designed as a Triplet Diffusion for Skeleton-Text Matching (TDSM) method which aligns skeleton features with text prompts through reverse diffusion, embedding the prompts into the unified skeleton-text latent space to achieve robust matching. To enhance discriminative power, we introduce a novel triplet diffusion (TD) loss that encourages our TDSM to correct skeleton-text matches while pushing apart incorrect ones. Our TDSM significantly outperforms the very recent state-of-the-art methods with large margins of 2.36%-point to 13.05%-point, demonstrating superior accuracy and scalability in zero-shot settings through effective skeleton-text matching.
Authors: Jeonghyeok Do, Jaehyup Lee, Munchurl Kim
Abstract: Synthetic Aperture Radar (SAR) imagery provides robust environmental and temporal coverage (e.g., during clouds, seasons, day-night cycles), yet its noise and unique structural patterns pose interpretation challenges, especially for non-experts. SAR-to-EO (Electro-Optical) image translation (SET) has emerged to make SAR images more perceptually interpretable. However, traditional approaches trained from scratch on limited SAR-EO datasets are prone to overfitting. To address these challenges, we introduce Confidence Diffusion for SAR-to-EO Translation, called C-DiffSET, a framework leveraging pretrained Latent Diffusion Model (LDM) extensively trained on natural images, thus enabling effective adaptation to the EO domain. Remarkably, we find that the pretrained VAE encoder aligns SAR and EO images in the same latent space, even with varying noise levels in SAR inputs. To further improve pixel-wise fidelity for SET, we propose a confidence-guided diffusion (C-Diff) loss that mitigates artifacts from temporal discrepancies, such as appearing or disappearing objects, thereby enhancing structural accuracy. C-DiffSET achieves state-of-the-art (SOTA) results on multiple datasets, significantly outperforming the very recent image-to-image translation methods and SET methods with large margins.
Authors: Xi Fang, Jiankun Wang, Xiaochen Cai, Shangqian Chen, Shuwen Yang, Haoyi Tao, Nan Wang, Lin Yao, Linfeng Zhang, Guolin Ke
Abstract: In recent decades, chemistry publications and patents have increased rapidly. A significant portion of key information is embedded in molecular structure figures, complicating large-scale literature searches and limiting the application of large language models in fields such as biology, chemistry, and pharmaceuticals. The automatic extraction of precise chemical structures is of critical importance. However, the presence of numerous Markush structures in real-world documents, along with variations in molecular image quality, drawing styles, and noise, significantly limits the performance of existing optical chemical structure recognition (OCSR) methods. We present MolParser, a novel end-to-end OCSR method that efficiently and accurately recognizes chemical structures from real-world documents, including difficult Markush structure. We use a extended SMILES encoding rule to annotate our training dataset. Under this rule, we build MolParser-7M, the largest annotated molecular image dataset to our knowledge. While utilizing a large amount of synthetic data, we employed active learning methods to incorporate substantial in-the-wild data, specifically samples cropped from real patents and scientific literature, into the training process. We trained an end-to-end molecular image captioning model, MolParser, using a curriculum learning approach. MolParser significantly outperforms classical and learning-based methods across most scenarios, with potential for broader downstream applications. The dataset is publicly available.
Authors: Jinxing Zhou, Dan Guo, Ruohao Guo, Yuxin Mao, Jingjing Hu, Yiran Zhong, Xiaojun Chang, Meng Wang
Abstract: The Audio-Visual Event Localization (AVEL) task aims to temporally locate and classify video events that are both audible and visible. Most research in this field assumes a closed-set setting, which restricts these models' ability to handle test data containing event categories absent (unseen) during training. Recently, a few studies have explored AVEL in an open-set setting, enabling the recognition of unseen events as ``unknown'', but without providing category-specific semantics. In this paper, we advance the field by introducing the Open-Vocabulary Audio-Visual Event Localization (OV-AVEL) problem, which requires localizing audio-visual events and predicting explicit categories for both seen and unseen data at inference. To address this new task, we propose the OV-AVEBench dataset, comprising 24,800 videos across 67 real-life audio-visual scenes (seen:unseen = 46:21), each with manual segment-level annotation. We also establish three evaluation metrics for this task. Moreover, we investigate two baseline approaches, one training-free and one using a further fine-tuning paradigm. Specifically, we utilize the unified multimodal space from the pretrained ImageBind model to extract audio, visual, and textual (event classes) features. The training-free baseline then determines predictions by comparing the consistency of audio-text and visual-text feature similarities. The fine-tuning baseline incorporates lightweight temporal layers to encode temporal relations within the audio and visual modalities, using OV-AVEBench training data for model fine-tuning. We evaluate these baselines on the proposed OV-AVEBench dataset and discuss potential directions for future work in this new field.
Authors: Markus Sch\"on, Michael Buchholz, Klaus Dietmayer
Abstract: Monocular geometric scene understanding combines panoptic segmentation and self-supervised depth estimation, focusing on real-time application in autonomous vehicles. We introduce MGNiceNet, a unified approach that uses a linked kernel formulation for panoptic segmentation and self-supervised depth estimation. MGNiceNet is based on the state-of-the-art real-time panoptic segmentation method RT-K-Net and extends the architecture to cover both panoptic segmentation and self-supervised monocular depth estimation. To this end, we introduce a tightly coupled self-supervised depth estimation predictor that explicitly uses information from the panoptic path for depth prediction. Furthermore, we introduce a panoptic-guided motion masking method to improve depth estimation without relying on video panoptic segmentation annotations. We evaluate our method on two popular autonomous driving datasets, Cityscapes and KITTI. Our model shows state-of-the-art results compared to other real-time methods and closes the gap to computationally more demanding methods. Source code and trained models are available at https://github.com/markusschoen/MGNiceNet.
Authors: Arundhati S. Shanbhag, Brian B. Moser, Tobias C. Nauen, Stanislav Frolov, Federico Raue, Andreas Dengel
Abstract: Diffusion models, celebrated for their generative capabilities, have recently demonstrated surprising effectiveness in image classification tasks by using Bayes' theorem. Yet, current diffusion classifiers must evaluate every label candidate for each input, creating high computational costs that impede their use in large-scale applications. To address this limitation, we propose a Hierarchical Diffusion Classifier (HDC) that exploits hierarchical label structures or well-defined parent-child relationships in the dataset. By pruning irrelevant high-level categories and refining predictions only within relevant subcategories (leaf nodes and sub-trees), HDC reduces the total number of class evaluations. As a result, HDC can speed up inference by as much as 60% while preserving and sometimes even improving classification accuracy. In summary, our work provides a tunable control mechanism between speed and precision, making diffusion-based classification more feasible for large-scale applications.
Authors: Ziyang Zong, Guanying Chen, Zhaohuan Zhan, Fengcheng Yu, Guang Tan
Abstract: This paper proposes a two-stage text-to-floorplan generation framework that combines the reasoning capability of Large Language Models (LLMs) with the generative power of diffusion models. In the first stage, we leverage a Chain-of-Thought (CoT) prompting strategy to guide an LLM in generating an initial layout (Layout-Init) from natural language descriptions, which ensures a user-friendly and intuitive design process. However, Layout-Init may lack precise geometric alignment and fine-grained structural details. To address this, the second stage employs a conditional diffusion model to refine Layout-Init into a final floorplan (Layout-Final) that better adheres to physical constraints and user requirements. Unlike prior methods, our approach effectively reduces the difficulty of floorplan generation learning without the need for extensive domain-specific training data. Experimental results demonstrate that our approach achieves state-of-the-art performance across all metrics, which validates its effectiveness in practical home design applications.
Authors: Andy Li, Aiden Durrant, Milan Markovic, Tianjin Huang, Souvik Kundu, Tianlong Chen, Lu Yin, Georgios Leontidis
Abstract: Pruning of deep neural networks has been an effective technique for reducing model size while preserving most of the performance of dense networks, crucial for deploying models on memory and power-constrained devices. While recent sparse learning methods have shown promising performance up to moderate sparsity levels such as 95% and 98%, accuracy quickly deteriorates when pushing sparsities to extreme levels due to unique challenges such as fragile gradient flow. In this work, we explore network performance beyond the commonly studied sparsities, and propose a collection of techniques that enable the continuous learning of networks without accuracy collapse even at extreme sparsities, including 99.90%, 99.95% and 99.99% on ResNet architectures. Our approach combines 1) Dynamic ReLU phasing, where DyReLU initially allows for richer parameter exploration before being gradually replaced by standard ReLU, 2) weight sharing which reuses parameters within a residual layer while maintaining the same number of learnable parameters, and 3) cyclic sparsity, where both sparsity levels and sparsity patterns evolve dynamically throughout training to better encourage parameter exploration. We evaluate our method, which we term Extreme Adaptive Sparse Training (EAST) at extreme sparsities using ResNet-34 and ResNet-50 on CIFAR-10, CIFAR-100, and ImageNet, achieving significant performance improvements over state-of-the-art methods we compared with.
Authors: Hao Ju, Shaofei Huang, Si Liu, Zhedong Zheng
Abstract: Existing approaches to drone visual geo-localization predominantly adopt the image-based setting, where a single drone-view snapshot is matched with images from other platforms. Such task formulation, however, underutilizes the inherent video output of the drone and is sensitive to occlusions and viewpoint disparity. To address these limitations, we formulate a new video-based drone geo-localization task and propose the Video2BEV paradigm. This paradigm transforms the video into a Bird's Eye View (BEV), simplifying the subsequent \textbf{inter-platform} matching process. In particular, we employ Gaussian Splatting to reconstruct a 3D scene and obtain the BEV projection. Different from the existing transform methods, \eg, polar transform, our BEVs preserve more fine-grained details without significant distortion. To facilitate the discriminative \textbf{intra-platform} representation learning, our Video2BEV paradigm also incorporates a diffusion-based module for generating hard negative samples. To validate our approach, we introduce UniV, a new video-based geo-localization dataset that extends the image-based University-1652 dataset. UniV features flight paths at $30^\circ$ and $45^\circ$ elevation angles with increased frame rates of up to 10 frames per second (FPS). Extensive experiments on the UniV dataset show that our Video2BEV paradigm achieves competitive recall rates and outperforms conventional video-based methods. Compared to other competitive methods, our proposed approach exhibits robustness at lower elevations with more occlusions.
Authors: Kaihong Wang, Lingzhi Zhang, Jianming Zhang
Abstract: Despite recent advancements, text-to-image generation models often produce images containing artifacts, especially in human figures. These artifacts appear as poorly generated human bodies, including distorted, missing, or extra body parts, leading to visual inconsistencies with typical human anatomy and greatly impairing overall fidelity. In this study, we address this challenge by curating Human Artifact Dataset (HAD), a diverse dataset specifically designed to localize human artifacts. HAD comprises over 37,000 images generated by several popular text-to-image models, annotated for human artifact localization. Using this dataset, we train the Human Artifact Detection Models (HADM), which can identify different artifacts across multiple generative domains and demonstrate strong generalization, even on images from unseen generators. Additionally, to further improve generators' perception of human structural coherence, we use the predictions from our HADM as feedback for diffusion model finetuning. Our experiments confirm a reduction in human artifacts in the resulting model. Furthermore, we showcase a novel application of our HADM in an iterative inpainting framework to correct human artifacts in arbitrary images directly, demonstrating its utility in improving image quality. Our dataset and detection models are available at: https://github.com/wangkaihong/HADM.
Authors: Hailang Huang, Yong Wang, Zixuan Huang, Huaqiu Li, Tongwen Huang, Xiangxiang Chu, Richong Zhang
Abstract: Large Multimodal Models (LMMs) demonstrate impressive capabilities. However, current benchmarks predominantly focus on image comprehension in specific domains, and these benchmarks are labor-intensive to construct. Moreover, their answers tend to be brief, making it difficult to assess the ability of LMMs to generate detailed descriptions of images. To address these limitations, we propose the MMGenBench-Pipeline, a straightforward and fully automated evaluation pipeline. This involves generating textual descriptions from input images, using these descriptions to create auxiliary images via text-to-image generative models, and then comparing the original and generated images. Furthermore, to ensure the effectiveness of MMGenBench-Pipeline, we design MMGenBench-Test, evaluating LMMs across 13 distinct image patterns, and MMGenBench-Domain, focusing on generative image performance. A thorough evaluation involving over 50 popular LMMs demonstrates the effectiveness and reliability of both the pipeline and benchmark. Our observations indicate that numerous LMMs excelling in existing benchmarks fail to adequately complete the basic tasks related to image understanding and description. This finding highlights the substantial potential for performance improvement in current LMMs and suggests avenues for future model optimization. Concurrently, MMGenBench-Pipeline can efficiently assess the performance of LMMs across diverse domains using only image inputs.
Authors: Yuanhao Cai, He Zhang, Kai Zhang, Yixun Liang, Mengwei Ren, Fujun Luan, Qing Liu, Soo Ye Kim, Jianming Zhang, Zhifei Zhang, Yuqian Zhou, Yulun Zhang, Xiaokang Yang, Zhe Lin, Alan Yuille
Abstract: Existing feedforward image-to-3D methods mainly rely on 2D multi-view diffusion models that cannot guarantee 3D consistency. These methods easily collapse when changing the prompt view direction and mainly handle object-centric cases. In this paper, we propose a novel single-stage 3D diffusion model, DiffusionGS, for object generation and scene reconstruction from a single view. DiffusionGS directly outputs 3D Gaussian point clouds at each timestep to enforce view consistency and allow the model to generate robustly given prompt views of any directions, beyond object-centric inputs. Plus, to improve the capability and generality of DiffusionGS, we scale up 3D training data by developing a scene-object mixed training strategy. Experiments show that DiffusionGS yields improvements of 2.20 dB/23.25 and 1.34 dB/19.16 in PSNR/FID for objects and scenes than the state-of-the-art methods, without depth estimator. Plus, our method enjoys over 5$\times$ faster speed ($\sim$6s on an A100 GPU). Our Project page at https://caiyuanhao1998.github.io/project/DiffusionGS/ shows the video and interactive results.
Authors: Tingyu Fan, Yueyu Hu, Yao Wang
Abstract: Point cloud video (PCV) is a versatile 3D representation of dynamic scenes with emerging applications. This paper introduces U-Motion, a learning-based compression scheme for both PCV geometry and attributes. We propose a U-Structured inter-frame prediction framework, U-Inter, which performs explicit motion estimation and compensation (ME/MC) at different scales with varying levels of detail. It integrates Top-Down (Fine-to-Coarse) Motion Propagation, Bottom-Up Motion Predictive Coding and Multi-scale Group Motion Compensation to enable accurate motion estimation and efficient motion compression at each scale. In addition, we design a multi-scale spatial-temporal predictive coding module to capture the cross-scale spatial redundancy remaining after U-Inter prediction. We conduct experiments following the MPEG Common Test Condition for dense dynamic point clouds and demonstrate that U-Motion can achieve significant gains over MPEG G-PCC-GesTM v3.0 and recently published learning-based methods for both geometry and attribute compression.
Authors: Yuzhou Tang, Dejun Xu, Yongjie Hou, Zhenzhong Wang, Min Jiang
Abstract: Photorealistic 3D reconstruction of unstructured real-world scenes remains challenging due to complex illumination variations and transient occlusions. Existing methods based on Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) struggle with inefficient light decoupling and structure-agnostic occlusion handling. To address these limitations, we propose NexusSplats, an approach tailored for efficient and high-fidelity 3D scene reconstruction under complex lighting and occlusion conditions. In particular, NexusSplats leverages a hierarchical light decoupling strategy that performs centralized appearance learning, efficiently and effectively decoupling varying lighting conditions. Furthermore, a structure-aware occlusion handling mechanism is developed, establishing a nexus between 3D and 2D structures for fine-grained occlusion handling. Experimental results demonstrate that NexusSplats achieves state-of-the-art rendering quality and reduces the number of total parameters by 65.4\%, leading to 2.7$\times$ faster reconstruction.
Authors: Youwei Zhou, Tianyang Xu, Cong Wu, Xiaojun Wu, Josef Kittler
Abstract: The shared topology of human skeletons motivated the recent investigation of graph convolutional network (GCN) solutions for action recognition. However, most of the existing GCNs rely on the binary connection of two neighboring vertices (joints) formed by an edge (bone), overlooking the potential of constructing multi-vertex convolution structures. Although some studies have attempted to utilize hyper-graphs to represent the topology, they rely on a fixed construction strategy, which limits their adaptivity in uncovering the intricate latent relationships within the action. In this paper, we address this oversight and explore the merits of an adaptive hyper-graph convolutional network (Hyper-GCN) to achieve the aggregation of rich semantic information conveyed by skeleton vertices. In particular, our Hyper-GCN adaptively optimises the hyper-graphs during training, revealing the action-driven multi-vertex relations. Besides, virtual connections are often designed to support efficient feature aggregation, implicitly extending the spectrum of dependencies within the skeleton. By injecting virtual connections into hyper-graphs, the semantic clues of diverse action categories can be highlighted. The results of experiments conducted on the NTU-60, NTU-120, and NW-UCLA datasets demonstrate the merits of our Hyper-GCN, compared to the state-of-the-art methods. Specifically, we outperform the existing solutions on NTU-120, achieving 90.5\% and 91.7\% in terms of the top-1 recognition accuracy on X-Sub and X-Set.
Authors: Tomas Berriel Martins, Martin R. Oswald, Javier Civera
Abstract: This paper presents an Open-Vocabulary Online 3D semantic mapping pipeline, that we denote by its acronym OVO. Given a sequence of posed RGB-D frames, we detect and track 3D segments, which we describe using CLIP vectors. These are computed from the viewpoints where they are observed by a novel CLIP merging method. Notably, our OVO has a significantly lower computational and memory footprint than offline baselines, while also showing better segmentation metrics than them. Along with superior segmentation performance, we also show experimental results of our mapping contributions integrated with two different SLAM backbones (Gaussian-SLAM and ORB-SLAM2), being the first ones demonstrating end-to-end open-vocabulary online 3D reconstructions without relying on ground-truth camera poses or scene geometry.
Authors: Evelyn J. Mannix, Liam Hodgkinson, Howard Bondell
Abstract: Knowledge distillation approaches compress models by training a student network using the classification outputs of a high quality teacher model, but can fail to effectively transfer the properties of computer vision foundation models from the teacher to the student. While it has been recently shown that feature distillation$\unicode{x2013}$where a teacher model's output features are replicated instead$\unicode{x2013}$can reproduce performance for foundation models across numerous downstream tasks, they fall short in matching critical properties such as robustness and out-of-distribution (OOD) detection performance. This paper overcomes this shortcoming by introducing Cosine-similarity Preserving Compression (CosPress), a feature distillation technique that learns a mapping to compress the latent space of the teacher model into the smaller latent space of the student, by preserving the cosine similarities between image embeddings. This enables direct optimisation of the student network and produces a more faithful reproduction of the teacher's properties. It is shown that distillation with CosPress on a variety of datasets, including ImageNet, produces more accurate models with greater performance on generalisability, robustness and OOD detection benchmarks, and that this technique provides a competitive pathway for training highly performant lightweight models on small datasets. Code is available at https://github.com/emannix/cospress.
Authors: Jing Wu, Mehrtash Harandi
Abstract: Machine Unlearning (MU) aims to selectively erase harmful behaviors from models while retaining the overall utility of the model. As a multi-task learning problem, MU involves balancing objectives related to forgetting specific concepts/data and preserving general performance. A naive integration of these forgetting and preserving objectives can lead to gradient conflicts and dominance, impeding MU algorithms from reaching optimal solutions. To address the gradient conflict and dominance issue, we reformulate MU as a two-player cooperative game, where the two players, namely, the forgetting player and the preservation player, contribute via their gradient proposals to maximize their overall gain and balance their contributions. To this end, inspired by the Nash bargaining theory, we derive a closed-form solution to guide the model toward the Pareto stationary point. Our formulation of MU guarantees an equilibrium solution, where any deviation from the final state would lead to a reduction in the overall objectives for both players, ensuring optimality in each objective. We evaluate our algorithm's effectiveness on a diverse set of tasks across image classification and image generation. Extensive experiments with ResNet, vision-language model CLIP, and text-to-image diffusion models demonstrate that our method outperforms state-of-the-art MU algorithms, achieving a better trade-off between forgetting and preserving. Our results also highlight improvements in forgetting precision, preservation of generalization, and robustness against adversarial attacks.
Authors: Ryugo Morita, Stanislav Frolov, Brian Bernhard Moser, Takahiro Shirakawa, Ko Watanabe, Andreas Dengel, Jinjia Zhou
Abstract: Diffusion models have enabled the generation of high-quality images with a strong focus on realism and textual fidelity. Yet, large-scale text-to-image models, such as Stable Diffusion, struggle to generate images where foreground objects are placed over a chroma key background, limiting their ability to separate foreground and background elements without fine-tuning. To address this limitation, we present a novel Training-Free Chroma Key Content Generation Diffusion Model (TKG-DM), which optimizes the initial random noise to produce images with foreground objects on a specifiable color background. Our proposed method is the first to explore the manipulation of the color aspects in initial noise for controlled background generation, enabling precise separation of foreground and background without fine-tuning. Extensive experiments demonstrate that our training-free method outperforms existing methods in both qualitative and quantitative evaluations, matching or surpassing fine-tuned models. Finally, we successfully extend it to other tasks (e.g., consistency models and text-to-video), highlighting its transformative potential across various generative applications where independent control of foreground and background is crucial.
Authors: Teng Zhou, Xiaoyu Zhang, Yongchuan Tang
Abstract: Panoramic Image Generation (PIG) aims to create coherent images of arbitrary lengths. Most existing methods fall in the joint diffusion paradigm, but their complex and heuristic crop connection designs often limit their ability to achieve multilevel coherence. By deconstructing this challenge into its core components, we find it naturally aligns with next-token prediction, leading us to adopt an autoregressive (AR) paradigm for PIG modeling. However, existing visual AR (VAR) models are limited to fixed-size generation, lacking the capability to produce panoramic images. In this paper, we propose PanoLlama, a novel framework that achieves endless and coherent panorama generation with the autoregressive paradigm. Our approach develops a training-free strategy that utilizes token redirection to overcome the size limitations of existing VAR models, enabling next-crop prediction in both horizontal and vertical directions. This refreshes the PIG pipeline while achieving SOTA performance in coherence (47.50\%), fidelity(28.16\%), and aesthetics (15\%). Additionally, PanoLlama supports applications other PIG methods cannot achieve, including mask-free layout control, multi-scale and multi-guidance synthesis. To facilitate standardized evaluation, we also establish a dataset with 1,000 prompts spanning 100+ themes, providing a new testing benchmark for PIG research.
Authors: Sule Bai, Yong Liu, Yifei Han, Haoji Zhang, Yansong Tang
Abstract: Recent advancements in pre-trained vision-language models like CLIP, have enabled the task of open-vocabulary segmentation. CLIP demonstrates impressive zero-shot capabilities in various downstream tasks that require holistic image understanding. However, due to its image-level pre-training, CLIP struggles to capture local details, resulting in poor performance in segmentation tasks. Our analysis reveals that anomaly tokens emerge during the forward pass, drawing excessive attention from normal patch tokens, thereby diminishing spatial awareness. To address this issue, we propose Self-Calibrated CLIP (SC-CLIP), a training-free method that calibrates CLIP to produce finer representations while preserving its original generalization ability, without introducing new parameters or relying on additional backbones. Specifically, we first identify and resolve the anomaly tokens to mitigate their negative impact. Next, we enhance feature discriminability and attention correlation by leveraging the semantic consistency found in CLIP's intermediate features. Furthermore, we explore how to effectively employ multi-level feature fusion under the training-free setting. Collectively, these strategies enhance CLIP's feature representation with greater granularity and coherence. Experimental results demonstrate the effectiveness of SC-CLIP, achieving state-of-the-art results across all datasets and surpassing previous methods by 9.5%. Notably, SC-CLIP boosts the performance of vanilla CLIP ViT-L/14 by 6.8 times. Our source code is available at https://github.com/SuleBai/SC-CLIP.
Authors: Ryo Fujii, Hideo Saito, Ryo Hachiuma
Abstract: This paper jointly addresses three key limitations in conventional pedestrian trajectory forecasting: pedestrian perception errors, real-world data collection costs, and person ID annotation costs. We propose a novel framework, RealTraj, that enhances the real-world applicability of trajectory forecasting. Our approach includes two training phases -- self-supervised pretraining on synthetic data and weakly-supervised fine-tuning with limited real-world data -- to minimize data collection efforts. To improve robustness to real-world errors, we focus on both model design and training objectives. Specifically, we present Det2TrajFormer, a trajectory forecasting model that remains invariant to tracking noise by using past detections as inputs. Additionally, we pretrain the model using multiple pretext tasks, which enhance robustness and improve forecasting performance based solely on detection data. Unlike previous trajectory forecasting methods, our approach fine-tunes the model using only ground-truth detections, reducing the need for costly person ID annotations. In the experiments, we comprehensively verify the effectiveness of the proposed method against the limitations, and the method outperforms state-of-the-art trajectory forecasting methods on multiple datasets. The code will be released at https://fujiry0.github.io/RealTraj-project-page.
Authors: Stuti Pathak, Prashant Kumar, Dheeraj Baiju, Nicholus Mboga, Gunther Steenackers, Rudi Penne
Abstract: Point clouds acquired in constrained, challenging, uncontrolled, and multi-sensor real-world settings are noisy, incomplete, and non-uniformly sparse. This presents acute challenges for the vital task of point cloud completion. Using tools from Algebraic Topology and Persistent Homology (PH), we demonstrate that current benchmark object point clouds lack rich topological features that are integral part of point clouds captured in realistic environments. To facilitate research in this direction, we contribute the first real-world industrial dataset for point cloud completion, RealPC - a diverse, rich and varied set of point clouds. It consists of ~ 40,000 pairs across 21 categories of industrial structures in railway establishments. Benchmark results on several strong baselines reveal that existing methods fail in real-world scenarios. We discover a striking observation - unlike current datasets, RealPC consists of multiple 0- and 1-dimensional PH-based topological features. We prove that integrating these topological priors into existing works helps improve completion. We present how 0-dimensional PH priors extract the global topology of a complete shape in the form of a 3D skeleton and assist a model in generating topologically consistent complete shapes. Since computing Homology is expensive, we present a simple, yet effective Homology Sampler guided network, BOSHNet that bypasses the Homology computation by sampling proxy backbones akin to 0-dim PH. These backbones provide similar benefits of 0-dim PH right from the start of the training, unlike similar methods where accurate backbones are obtained only during later phases of the training.
Authors: Xinyu Zhang, Lingling Zhang, Yanrui Wu, Muye Huang, Wenjun Wu, Bo Li, Shaowei Wang, Basura Fernando, Jun Liu
Abstract: Visual Question Generation (VQG) has gained significant attention due to its potential in educational applications. However, VQG research mainly focuses on natural images, largely neglecting diagrams in educational materials used to assess students' conceptual understanding. To address this gap, we construct DiagramQG, a dataset containing 8,372 diagrams and 19,475 questions across various subjects. DiagramQG introduces concept and target text constraints, guiding the model to generate concept-focused questions for educational purposes. Meanwhile, we present the Hierarchical Knowledge Integration framework for Diagram Question Generation (HKI-DQG) as a strong baseline. This framework obtains multi-scale patches of diagrams and acquires knowledge using a visual language model with frozen parameters. It then integrates knowledge, text constraints, and patches to generate concept-focused questions. We evaluate the performance of existing VQG models, open-source and closed-source vision-language models, and HKI-DQG on the DiagramQG dataset. Our novel HKI-DQG consistently outperforms existing methods, demonstrating that it serves as a strong baseline. Furthermore, we apply HKI-DQG to four other VQG datasets of natural images, namely VQG-COCO, K-VQG, OK-VQA, and A-OKVQA, achieving state-of-the-art performance.
Authors: Ronglai Zuo, Rolandos Alexandros Potamias, Evangelos Ververas, Jiankang Deng, Stefanos Zafeiriou
Abstract: Sign language is a visual language that encompasses all linguistic features of natural languages and serves as the primary communication method for the deaf and hard-of-hearing communities. Although many studies have successfully adapted pretrained language models (LMs) for sign language translation (sign-to-text), the reverse task-sign language generation (text-to-sign)-remains largely unexplored. In this work, we introduce a multilingual sign language model, Signs as Tokens (SOKE), which can generate 3D sign avatars autoregressively from text inputs using a pretrained LM. To align sign language with the LM, we leverage a decoupled tokenizer that discretizes continuous signs into token sequences representing various body parts. During decoding, unlike existing approaches that flatten all part-wise tokens into a single sequence and predict one token at a time, we propose a multi-head decoding method capable of predicting multiple tokens simultaneously. This approach improves inference efficiency while maintaining effective information fusion across different body parts. To further ease the generation process, we propose a retrieval-enhanced SLG approach, which incorporates external sign dictionaries to provide accurate word-level signs as auxiliary conditions, significantly improving the precision of generated signs. Extensive qualitative and quantitative evaluations demonstrate the effectiveness of SOKE. Code, models, and data will be made publicly available.
Authors: Huiguo He, Qiuyue Wang, Yuan Zhou, Yuxuan Cai, Hongyang Chao, Jian Yin, Huan Yang
Abstract: Training-free diffusion models have achieved remarkable progress in generating multi-subject consistent images within open-domain scenarios. The key idea of these methods is to incorporate reference subject information within the attention layer. However, existing methods still obtain suboptimal performance when handling numerous subjects. This paper reveals two primary issues contributing to this deficiency. Firstly, the undesired internal attraction between different subjects within the target image can lead to the convergence of multiple subjects into a single entity. Secondly, tokens tend to reference nearby tokens, which reduces the effectiveness of the attention mechanism when there is a significant positional difference between subjects in reference and target images. To address these issues, we propose a training-free diffusion model with Isolation and Reposition Attention, named IR-Diffusion. Specifically, Isolation Attention ensures that multiple subjects in the target image do not reference each other, effectively eliminating the subject convergence. On the other hand, Reposition Attention involves scaling and repositioning subjects in both reference and target images to the same position within the images. This ensures that subjects in the target image can better reference those in the reference image, thereby maintaining better consistency. Extensive experiments demonstrate that IR-Diffusion significantly enhances multi-subject consistency, outperforming all existing methods in open-domain scenarios.
Authors: Prajwal Singh, Ashish Tiwari, Gautam Vashishtha, Shanmuganathan Raman
Abstract: Neural radiance fields (NeRF) have revolutionized photorealistic rendering of novel views for 3D scenes. Despite their growing popularity and efficiency as 3D resources, NeRFs face scalability challenges due to the need for separate models per scene and the cumulative increase in training time for multiple scenes. The potential for incrementally encoding multiple 3D scenes into a single NeRF model remains largely unexplored. To address this, we introduce Continual-Neural Graphics Primitives (C-NGP), a novel continual learning framework that integrates multiple scenes incrementally into a single neural radiance field. Using a generative replay approach, C-NGP adapts to new scenes without requiring access to old data. We demonstrate that C-NGP can accommodate multiple scenes without increasing the parameter count, producing high-quality novel-view renderings on synthetic and real datasets. Notably, C-NGP models all 8 scenes from the Real-LLFF dataset together, with only a 2.2% drop in PSNR compared to vanilla NeRF, which models each scene independently. Further, C-NGP allows multiple style edits in the same network. The implementation details and dynamic visualizations are in the supplementary material.
Authors: Shukang Yin, Chaoyou Fu, Sirui Zhao, Yunhang Shen, Chunjiang Ge, Yan Yang, Zuwei Long, Yuhan Dai, Yongdong Luo, Haoyu Cao, Tong Xu, Xing Sun, Caifeng Shan, Ran He, Enhong Chen
Abstract: Recent years have witnessed the success of Multimodal Large Language Models (MLLMs) in the vision understanding domain. The success of these models can largely be attributed to the dominant scaling law, which states that larger parameter sizes and data volumes contribute to better performance. Notably, data scaling has mainly been powered by automatic data pipelines, which center around the self-instruction of LLMs. The paradigm has been taken for granted for quite some time, but the study of the effectiveness of scaling with these data has been neglected for a long time. In this context, this work revisits scaling with synthetic data and focuses on developing video-LLMs from a data-centric perspective. Our main study approach is fine-tuning pre-trained image-LLMs with video data and investigating learning efficiency through data scaling. Results from our preliminary experiments reveal a low learning efficiency phenomenon when simply scaling up video data samples, which, through our probing, can be ascribed to a lack of instruction diversity. Aiming at this issue, we propose a data augmentation method called Sparrow, which synthesizes video-like samples from pure text instruction data. Mixing these synthetic samples with the video data enables a more efficient training scheme. Through comprehensive experiments, we demonstrate that our proposed method achieves performance comparable to or even superior to baselines trained with many more samples. Meanwhile, we find that incorporating these synthetic samples can boost the performance of long video understanding without training with long video data. The code and data examples are available at https://github.com/VITA-MLLM/Sparrow.
Authors: Wenda Shi, Yiren Song, Dengming Zhang, Jiaming Liu, Xingxing Zou
Abstract: Visual text rendering are widespread in various real-world applications, requiring careful font selection and typographic choices. Recent progress in diffusion transformer (DiT)-based text-to-image (T2I) models show promise in automating these processes. However, these methods still encounter challenges like inconsistent fonts, style variation, and limited fine-grained control, particularly at the word-level. This paper proposes a two-stage DiT-based pipeline to address these problems by enhancing controllability over typography and style in text rendering. We introduce typography control fine-tuning (TC-FT), an parameter-efficient fine-tuning method (on $5\%$ key parameters) with enclosing typography control tokens (ETC-tokens), which enables precise word-level application of typographic features. To further address style inconsistency in text rendering, we propose a text-agnostic style control adapter (SCA) that prevents content leakage while enhancing style consistency. To implement TC-FT and SCA effectively, we incorporated HTML-render into the data synthesis pipeline and proposed the first word-level controllable dataset. Through comprehensive experiments, we demonstrate the effectiveness of our approach in achieving superior word-level typographic control, font consistency, and style consistency in text rendering tasks. The datasets and models will be available for academic use.
Authors: Alexander Markin, Vadim Pryadilshchikov, Artem Komarichev, Ruslan Rakhimov, Peter Wonka, Evgeny Burnaev
Abstract: Transient objects in video sequences can significantly degrade the quality of 3D scene reconstructions. To address this challenge, we propose T-3DGS, a novel framework that robustly filters out transient distractors during 3D reconstruction using Gaussian Splatting. Our framework consists of two steps. First, we employ an unsupervised classification network that distinguishes transient objects from static scene elements by leveraging their distinct training dynamics within the reconstruction process. Second, we refine these initial detections by integrating an off-the-shelf segmentation method with a bidirectional tracking module, which together enhance boundary accuracy and temporal coherence. Evaluations on both sparsely and densely captured video datasets demonstrate that T-3DGS significantly outperforms state-of-the-art approaches, enabling high-fidelity 3D reconstructions in challenging, real-world scenarios.
Authors: Tianshuo Xu, Zhifei Chen, Leyi Wu, Hao Lu, Yuying Chen, Lihui Jiang, Bingbing Liu, Yingcong Chen
Abstract: Recent advances in video generation have shown promise for generating future scenarios, critical for planning and control in autonomous driving and embodied intelligence. However, real-world applications demand more than visually plausible predictions; they require reasoning about object motions based on explicitly defined boundary conditions, such as initial scene image and partial object motion. We term this capability Boundary Conditional Motion Reasoning. Current approaches either neglect explicit user-defined motion constraints, producing physically inconsistent motions, or conversely demand complete motion inputs, which are rarely available in practice. Here we introduce Motion Dreamer, a two-stage framework that explicitly separates motion reasoning from visual synthesis, addressing these limitations. Our approach introduces instance flow, a sparse-to-dense motion representation enabling effective integration of partial user-defined motions, and the motion inpainting strategy to robustly enable reasoning motions of other objects. Extensive experiments demonstrate that Motion Dreamer significantly outperforms existing methods, achieving superior motion plausibility and visual realism, thus bridging the gap towards practical boundary conditional motion reasoning. Our webpage is available: https://envision-research.github.io/MotionDreamer/.
Authors: Haodong Lu, Chongyang Zhao, Jason Xue, Lina Yao, Kristen Moore, Dong Gong
Abstract: We investigate whether the pre-trained knowledge in vision-language models (VLMs), such as CLIP, can be retained -- or even enhanced -- in continual learning (CL) while incorporating new knowledge from the data stream. Existing CL methods primarily focus on continual downstream adaptation using components isolated from pre-trained model (PTM), increasing inference complexity and limiting improvements to the PTM itself; some also retain knowledge relying on additional reference data, leading to high training costs. To address these limitations, we propose a universal and efficient Continual Learning approach for VLM based on Dynamic Rank-Selective LoRA (CoDyRA), which directly improves the PTMs while preserving the existing knowledge from both pre-training and CL. Through analyses on how LoRA rank and placement impact/regularize the learning and forgetting in CL, we propose CoDyRA that adaptively performs rank-minimized parameter updates in different modules, based on their importance to the current data. This ensures a balance between knowledge acquisition (plasticity) and forgetting mitigation (stability). Our method operates without explicit domain or distribution prediction and does not rely on reference data, enabling seamless task integration while maintaining pre-trained capabilities. Moreover, CoDyRA preserves the original model architecture and deployment pipeline, introducing no additional inference overhead. Extensive experiments demonstrate that our approach enhances representations based on new downstream data while retaining pre-trained knowledge, achieving state-of-the-art results.
Authors: Ivan DeAndres-Tame, Ruben Tolosana, Pietro Melzi, Ruben Vera-Rodriguez, Minchul Kim, Christian Rathgeb, Xiaoming Liu, Luis F. Gomez, Aythami Morales, Julian Fierrez, Javier Ortega-Garcia, Zhizhou Zhong, Yuge Huang, Yuxi Mi, Shouhong Ding, Shuigeng Zhou, Shuai He, Lingzhi Fu, Heng Cong, Rongyu Zhang, Zhihong Xiao, Evgeny Smirnov, Anton Pimenov, Aleksei Grigorev, Denis Timoshenko, Kaleb Mesfin Asfaw, Cheng Yaw Low, Hao Liu, Chuyi Wang, Qing Zuo, Zhixiang He, Hatef Otroshi Shahreza, Anjith George, Alexander Unnervik, Parsa Rahimi, S\'ebastien Marcel, Pedro C. Neto, Marco Huber, Jan Niklas Kolf, Naser Damer, Fadi Boutros, Jaime S. Cardoso, Ana F. Sequeira, Andrea Atzori, Gianni Fenu, Mirko Marras, Vitomir \v{S}truc, Jiang Yu, Zhangjie Li, Jichun Li, Weisong Zhao, Zhen Lei, Xiangyu Zhu, Xiao-Yu Zhang, Bernardo Biesseck, Pedro Vidal, Luiz Coelho, Roger Granada, David Menotti
Abstract: Synthetic data is gaining increasing popularity for face recognition technologies, mainly due to the privacy concerns and challenges associated with obtaining real data, including diverse scenarios, quality, and demographic groups, among others. It also offers some advantages over real data, such as the large amount of data that can be generated or the ability to customize it to adapt to specific problem-solving needs. To effectively use such data, face recognition models should also be specifically designed to exploit synthetic data to its fullest potential. In order to promote the proposal of novel Generative AI methods and synthetic data, and investigate the application of synthetic data to better train face recognition systems, we introduce the 2nd FRCSyn-onGoing challenge, based on the 2nd Face Recognition Challenge in the Era of Synthetic Data (FRCSyn), originally launched at CVPR 2024. This is an ongoing challenge that provides researchers with an accessible platform to benchmark i) the proposal of novel Generative AI methods and synthetic data, and ii) novel face recognition systems that are specifically proposed to take advantage of synthetic data. We focus on exploring the use of synthetic data both individually and in combination with real data to solve current challenges in face recognition such as demographic bias, domain adaptation, and performance constraints in demanding situations, such as age disparities between training and testing, changes in the pose, or occlusions. Very interesting findings are obtained in this second edition, including a direct comparison with the first one, in which synthetic databases were restricted to DCFace and GANDiffFace.
Authors: Rongkun Xue, Jinouwen Zhang, Yazhe Niu, Dazhong Shen, Bingqi Ma, Yu Liu, Jing Yang
Abstract: Recent generative models based on score matching and flow matching have significantly advanced generation tasks, but their potential in discriminative tasks remains underexplored. Previous approaches, such as generative classifiers, have not fully leveraged the capabilities of these models for discriminative tasks due to their intricate designs. We propose Pretrained Reversible Generation (PRG), which extracts unsupervised representations by reversing the generative process of a pretrained continuous generation model. PRG effectively reuses unsupervised generative models, leveraging their high capacity to serve as robust and generalizable feature extractors for downstream tasks. This framework enables the flexible selection of feature hierarchies tailored to specific downstream tasks. Our method consistently outperforms prior approaches across multiple benchmarks, achieving state-of-the-art performance among generative model based methods, including 78% top-1 accuracy on ImageNet at a resolution of 64. Extensive ablation studies, including out-of-distribution evaluations, further validate the effectiveness of our approach.
Authors: Jiahuan Cheng, Jan-Nico Zaech, Luc Van Gool, Danda Pani Paudel
Abstract: TL;DR: Gaussian Splatting is a widely adopted approach for 3D scene representation, offering efficient, high-quality reconstruction and rendering. A key reason for its success is the simplicity of representing scenes with sets of Gaussians, making it interpretable and adaptable. To enhance understanding beyond visual representation, recent approaches extend Gaussian Splatting with semantic vision-language features, enabling open-set tasks. Typically, these language features are aggregated from multiple 2D views, however, existing methods rely on cumbersome techniques, resulting in high computational costs and longer training times. In this work, we show that the complicated pipelines for language 3D Gaussian Splatting are simply unnecessary. Instead, we follow a probabilistic formulation of Language Gaussian Splatting and apply Occam's razor to the task at hand, leading to a highly efficient weighted multi-view feature aggregation technique. Doing so offers us state-of-the-art results with a speed-up of two orders of magnitude without any compression, allowing for easy scene manipulation. Project Page: https://insait-institute.github.io/OccamLGS/
Authors: Bo Wen, Haochen Zhang, Dirk-Uwe G. Bartsch, William R. Freeman, Truong Q. Nguyen, Cheolhong An
Abstract: Topological correctness is critical for segmentation of tubular structures. Existing topological segmentation loss functions are primarily based on the persistent homology of the image. They match the persistent features from the segmentation with the persistent features from the ground truth and minimize the difference between them. However, these methods suffer from an ambiguous matching problem since the matching only relies on the information in the topological space. In this work, we propose an effective and efficient Spatial-Aware Topological Loss Function that further leverages the information in the original spatial domain of the image to assist the matching of persistent features. Extensive experiments on images of various types of tubular structures show that the proposed method has superior performance in improving the topological accuracy of the segmentation compared with state-of-the-art methods.
Authors: Fan-Yun Sun, Weiyu Liu, Siyi Gu, Dylan Lim, Goutam Bhat, Federico Tombari, Manling Li, Nick Haber, Jiajun Wu
Abstract: Spatial reasoning is a fundamental aspect of human cognition, enabling intuitive understanding and manipulation of objects in three-dimensional space. However, Large Language Models (LLMs) struggle with simple tasks such as arranging 3D assets in space according to open-ended language instructions, particularly in dense and physically constrained environments. We introduce LayoutVLM, a framework and scene layout representation that exploits the semantic knowledge of Vision-Language Models (VLMs) and supports differentiable optimization to ensure physical plausibility. LayoutVLM employs VLMs to generate two mutually reinforcing representations from visually marked images, and a self-consistent decoding process to improve VLMs spatial planning. Our experiments show that LayoutVLM addresses the limitations of existing LLM and constraint-based approaches, producing physically plausible 3D layouts better aligned with the semantic intent of input language instructions. We also demonstrate that fine-tuning VLMs with the proposed scene layout representation extracted from existing scene datasets can improve their reasoning performance.
Authors: Kazuto Nakashima, Xiaowen Liu, Tomoya Miyawaki, Yumi Iwashita, Ryo Kurazume
Abstract: Building LiDAR generative models holds promise as powerful data priors for restoration, scene manipulation, and scalable simulation in autonomous mobile robots. In recent years, approaches using diffusion models have emerged, significantly improving training stability and generation quality. Despite their success, diffusion models require numerous iterations of running neural networks to generate high-quality samples, making the increasing computational cost a potential barrier for robotics applications. To address this challenge, this paper presents R2Flow, a fast and high-fidelity generative model for LiDAR data. Our method is based on rectified flows that learn straight trajectories, simulating data generation with significantly fewer sampling steps compared to diffusion models. We also propose an efficient Transformer-based model architecture for processing the image representation of LiDAR range and reflectance measurements. Our experiments on unconditional LiDAR data generation using the KITTI-360 dataset demonstrate the effectiveness of our approach in terms of both efficiency and quality.
Authors: Conghao Wong, Ziqian Zou, Beihao Xia, Xinge You
Abstract: Learning to forecast trajectories of intelligent agents has caught much more attention recently. However, it remains a challenge to accurately account for agents' intentions and social behaviors when forecasting, and in particular, to simulate the unique randomness within each of those components in an explainable and decoupled way. Inspired by vibration systems and their resonance properties, we propose the Resonance (short for Re) model to encode and forecast pedestrian trajectories in the form of ``co-vibrations''. It decomposes trajectory modifications and randomnesses into multiple vibration portions to simulate agents' reactions to each single cause, and forecasts trajectories as the superposition of these independent vibrations separately. Also, benefiting from such vibrations and their spectral properties, representations of social interactions can be learned by emulating the resonance phenomena, further enhancing its explainability. Experiments on multiple datasets have verified its usefulness both quantitatively and qualitatively.
Authors: Fengyuan Shi, Zhuoyan Luo, Yixiao Ge, Yujiu Yang, Ying Shan, Limin Wang
Abstract: Existing vector quantization (VQ) methods struggle with scalability, largely attributed to the instability of the codebook that undergoes partial updates during training. The codebook is prone to collapse as utilization decreases, due to the progressively widening distribution gap between non-activated codes and visual features. To solve the problem, we propose Index Backpropagation Quantization (IBQ), a new VQ method for the joint optimization of all codebook embeddings and the visual encoder. Applying a straight-through estimator on the one-hot categorical distribution between the encoded feature and codebook, all codes are differentiable and maintain a consistent latent space with the visual encoder. IBQ enables scalable training of visual tokenizers and, for the first time, achieves a large-scale codebook ($2^{18}$) with high dimension ($256$) and high utilization. Experiments on the standard ImageNet benchmark demonstrate the scalability and superiority of IBQ, achieving competitive results on reconstruction and the application of autoregressive visual generation. The code and models are available at https://github.com/TencentARC/SEED-Voken.
Authors: Sarthak Kumar Maharana, Baoming Zhang, Leonid Karlinsky, Rogerio Feris, Yunhui Guo
Abstract: Although open-vocabulary classification models like Contrastive Language Image Pretraining (CLIP) have demonstrated strong zero-shot learning capabilities, their robustness to common image corruptions remains poorly understood. Through extensive experiments, we show that zero-shot CLIP lacks robustness to common image corruptions during test-time, necessitating the adaptation of CLIP to unlabeled corrupted images using test-time adaptation (TTA). However, we found that existing TTA methods have severe limitations in adapting CLIP due to their unimodal nature. To address these limitations, we propose $\texttt{BATCLIP}$, a bimodal $\textbf{online}$ TTA method designed to improve CLIP's robustness to common image corruptions. The key insight of our approach is not only to adapt the visual encoders for improving image features but also to strengthen the alignment between image and text features by promoting a stronger association between the image class prototype, computed using pseudo-labels, and the corresponding text feature. We evaluate our approach on benchmark image corruption datasets and achieve state-of-the-art results in online TTA for CLIP. Furthermore, we evaluate our proposed TTA approach on various domain generalization datasets to demonstrate its generalization capabilities.
Authors: Fawad Javed Fateh, Umer Ahmed, Hamza Khan, M. Zeeshan Zia, Quoc-Huy Tran
Abstract: This paper introduces TemporalVLM, a video large language model (video LLM) capable of effective temporal reasoning and fine-grained understanding in long videos. At the core, our approach includes a visual encoder for mapping a long-term input video into features which are time-aware and contain both local and global cues. In particular, it first divides the input video into short-term clips, which are jointly encoded with their timestamps into time-sensitive local features. Next, the local features are passed through a bidirectional long short-term memory (BiLSTM) module for global feature aggregation. The extracted time-aware and multi-level features are important for accurate temporal reasoning and fine-grained understanding in long videos. Moreover, to facilitate the evaluation of TemporalVLM, we present a large-scale long video dataset of industry assembly processes, namely IndustryASM, which consists of videos recorded on factory floors with actions and timestamps annotated by industrial engineers for time and motion studies and temporal action segmentation evaluation. Finally, extensive experiments on datasets of long videos, including TimeIT and IndustryASM, show that TemporalVLM achieves superior performance than previous methods across temporal reasoning and fine-grained understanding tasks, namely dense video captioning, temporal video grounding, video highlight detection, and temporal action segmentation. To the best of our knowledge, our work is the first to incorporate LSTMs into video LLMs.
Authors: Shouwei Ruan, Hanqing Liu, Yao Huang, Xiaoqi Wang, Caixin Kang, Hang Su, Yinpeng Dong, Xingxing Wei
Abstract: Vision Language Models (VLMs) have exhibited remarkable generalization capabilities, yet their robustness in dynamic real-world scenarios remains largely unexplored. To systematically evaluate VLMs' robustness to real-world 3D variations, we propose AdvDreamer, the first framework capable of generating physically reproducible Adversarial 3D Transformation (Adv-3DT) samples from single-view observations. In AdvDreamer, we integrate three key innovations: Firstly, to characterize real-world 3D variations with limited prior knowledge precisely, we design a zero-shot Monocular Pose Manipulation pipeline built upon generative 3D priors. Secondly, to ensure the visual quality of worst-case Adv-3DT samples, we propose a Naturalness Reward Model that provides continuous naturalness regularization during adversarial optimization, effectively preventing convergence to hallucinated or unnatural elements. Thirdly, to enable systematic evaluation across diverse VLM architectures and visual-language tasks, we introduce the Inverse Semantic Probability loss as the adversarial optimization objective, which solely operates in the fundamental visual-textual alignment space. Based on the captured Adv-3DT samples with high aggressiveness and transferability, we establish MM3DTBench, the first VQA benchmark dataset tailored to evaluate VLM robustness under challenging 3D variations. Extensive evaluations of representative VLMs with varying architectures reveal that real-world 3D variations can pose severe threats to model performance across various tasks.
Authors: Runjian Chen, Hang Zhang, Avinash Ravichandran, Hyoungseob Park, Wenqi Shao, Alex Wong, Ping Luo
Abstract: Unsupervised 3D representation learning reduces the burden of labeling multimodal 3D data for fusion perception tasks. Among different pre-training paradigms, differentiable-rendering-based methods have shown most promise. However, existing works separately conduct pre-training for each modalities due to computational costs of processing large point clouds with images. As such, mutual benefit of high-level semantics (from image) and 3D structure (from point cloud) has not been exploited. To address this gap, we propose a joint unsupervised differentiable-rendering-based pre-training method for images and point clouds, termed CLAP, short for Curvature sampLing and leArnable Prototype. Specifically, our method overcomes the computational hurdle by Curvature Sampling to select the more informative points/pixels for pre-training. To uncover the performance benefits brought by their complementarity, we propose to use learnable prototypes to represent parts of the 3D scenes in a common feature space and an Expectation-Maximization training scheme to associate embeddings of each modality to prototypes. We further propose a swapping prediction loss that explores their interplay through prototypes along with a Gram Matrix Regularization term to maintain training stability. Experiments on NuScenes and Waymo datasets show that CLAP achieves up to 100% more performance gain as compared to previous SOTA pre-training methods. Codes and models will be released.
Authors: Zhaopeng Gu, Bingke Zhu, Guibo Zhu, Yingying Chen, Ming Tang, Jinqiao Wang
Abstract: Visual Anomaly Detection (VAD) aims to identify abnormal samples in images that deviate from normal patterns, covering multiple domains, including industrial, logical, and medical fields. Due to the domain gaps between these fields, existing VAD methods are typically tailored to each domain, with specialized detection techniques and model architectures that are difficult to generalize across different domains. Moreover, even within the same domain, current VAD approaches often follow a "one-category-one-model" paradigm, requiring large amounts of normal samples to train class-specific models, resulting in poor generalizability and hindering unified evaluation across domains. To address this issue, we propose a generalized few-shot VAD method, UniVAD, capable of detecting anomalies across various domains, such as industrial, logical, and medical anomalies, with a training-free unified model. UniVAD only needs few normal samples as references during testing to detect anomalies in previously unseen objects, without training on the specific domain. Specifically, UniVAD employs a Contextual Component Clustering ($C^3$) module based on clustering and vision foundation models to segment components within the image accurately, and leverages Component-Aware Patch Matching (CAPM) and Graph-Enhanced Component Modeling (GECM) modules to detect anomalies at different semantic levels, which are aggregated to produce the final detection result. We conduct experiments on nine datasets spanning industrial, logical, and medical fields, and the results demonstrate that UniVAD achieves state-of-the-art performance in few-shot anomaly detection tasks across multiple domains, outperforming domain-specific anomaly detection models. Code is available at https://github.com/FantasticGNU/UniVAD.
Authors: Kangan Qian, Jinyu Miao, Xinyu Jiao, Ziang Luo, Zheng Fu, Yining Shi, Yunlong Wang, Kun Jiang, Diange Yang
Abstract: Reliable spatial and motion perception is essential for safe autonomous navigation. Recently, class-agnostic motion prediction on bird's-eye view (BEV) cell grids derived from LiDAR point clouds has gained significant attention. However, existing frameworks typically perform cell classification and motion prediction on a per-pixel basis, neglecting important motion field priors such as rigidity constraints, temporal consistency, and future interactions between agents. These limitations lead to degraded performance, particularly in sparse and distant regions. To address these challenges, we introduce \textbf{PriorMotion}, an innovative generative framework designed for class-agnostic motion prediction that integrates essential motion priors by modeling them as distributions within a structured latent space. Specifically, our method captures structured motion priors using raster-vector representations and employs a variational autoencoder with distinct dynamic and static components to learn future motion distributions in the latent space. Experiments on the nuScenes dataset demonstrate that \textbf{PriorMotion} outperforms state-of-the-art methods across both traditional metrics and our newly proposed evaluation criteria. Notably, we achieve improvements of approximately 15.24\% in accuracy for fast-moving objects, an 3.59\% increase in generalization, a reduction of 0.0163 in motion stability, and a 31.52\% reduction in prediction errors in distant regions. Further validation on FMCW LiDAR sensors confirms the robustness of our approach.
Authors: Bingchen Li, Xin Li, Yiting Lu, Zhibo Chen
Abstract: We present the first loss agent, dubbed LossAgent, for low-level image processing tasks, e.g., image super-resolution and restoration, intending to achieve any customized optimization objectives of low-level image processing in different practical applications. Notably, not all optimization objectives, such as complex hand-crafted perceptual metrics, text description, and intricate human feedback, can be instantiated with existing low-level losses, e.g., MSE loss, which presents a crucial challenge in optimizing image processing networks in an end-to-end manner. To eliminate this, our LossAgent introduces the powerful large language model (LLM) as the loss agent, where the rich textual understanding of prior knowledge empowers the loss agent with the potential to understand complex optimization objectives, trajectory, and state feedback from external environments in the optimization process of the low-level image processing networks. In particular, we establish the loss repository by incorporating existing loss functions that support the end-to-end optimization for low-level image processing. Then, we design the optimization-oriented prompt engineering for the loss agent to actively and intelligently decide the compositional weights for each loss in the repository at each optimization interaction, thereby achieving the required optimization trajectory for any customized optimization objectives. Extensive experiments on three typical low-level image processing tasks and multiple optimization objectives have shown the effectiveness and applicability of our proposed LossAgent.
Authors: Yixin Zhang, Nicholas Konz, Kevin Kramer, Maciej A. Mazurowski
Abstract: Image segmentation foundation models (SFMs) like Segment Anything Model (SAM) have achieved impressive zero-shot and interactive segmentation across diverse domains. However, they struggle to segment objects with certain structures, particularly those with dense, tree-like morphology and low textural contrast from their surroundings. These failure modes are crucial for understanding the limitations of SFMs in real-world applications. To systematically study this issue, we introduce interpretable metrics quantifying object tree-likeness and textural separability. On carefully controlled synthetic experiments and real-world datasets, we show that SFM performance (e.g., SAM, SAM 2, HQ-SAM) noticeably correlates with these factors. We link these failures to "textural confusion", where models misinterpret local structure as global texture, causing over-segmentation or difficulty distinguishing objects from similar backgrounds. Notably, targeted fine-tuning fails to resolve this issue, indicating a fundamental limitation. Our study provides the first quantitative framework for modeling the behavior of SFMs on challenging structures, offering interpretable insights into their segmentation capabilities.
Authors: Zhenglin Huang, Jinwei Hu, Xiangtai Li, Yiwei He, Xingyu Zhao, Bei Peng, Baoyuan Wu, Xiaowei Huang, Guangliang Cheng
Abstract: The rapid advancement of generative models in creating highly realistic images poses substantial risks for misinformation dissemination. For instance, a synthetic image, when shared on social media, can mislead extensive audiences and erode trust in digital content, resulting in severe repercussions. Despite some progress, academia has not yet created a large and diversified deepfake detection dataset for social media, nor has it devised an effective solution to address this issue. In this paper, we introduce the Social media Image Detection dataSet (SID-Set), which offers three key advantages: (1) extensive volume, featuring 300K AI-generated/tampered and authentic images with comprehensive annotations, (2) broad diversity, encompassing fully synthetic and tampered images across various classes, and (3) elevated realism, with images that are predominantly indistinguishable from genuine ones through mere visual inspection. Furthermore, leveraging the exceptional capabilities of large multimodal models, we propose a new image deepfake detection, localization, and explanation framework, named SIDA (Social media Image Detection, localization, and explanation Assistant). SIDA not only discerns the authenticity of images, but also delineates tampered regions through mask prediction and provides textual explanations of the model's judgment criteria. Compared with state-of-the-art deepfake detection models on SID-Set and other benchmarks, extensive experiments demonstrate that SIDA achieves superior performance among diversified settings. The code, model, and dataset will be released.
Authors: Yongkang Li, Tianheng Cheng, Bin Feng, Wenyu Liu, Xinggang Wang
Abstract: Recent open-vocabulary segmentation methods adopt mask generators to predict segmentation masks and leverage pre-trained vision-language models, e.g., CLIP, to classify these masks via mask pooling. Although these approaches show promising results, it is counterintuitive that accurate masks often fail to yield accurate classification results through pooling CLIP image embeddings within the mask regions. In this paper, we reveal the performance limitations of mask pooling and introduce Mask-Adapter, a simple yet effective method to address these challenges in open-vocabulary segmentation. Compared to directly using proposal masks, our proposed Mask-Adapter extracts semantic activation maps from proposal masks, providing richer contextual information and ensuring alignment between masks and CLIP. Additionally, we propose a mask consistency loss that encourages proposal masks with similar IoUs to obtain similar CLIP embeddings to enhance models' robustness to varying predicted masks. Mask-Adapter integrates seamlessly into open-vocabulary segmentation methods based on mask pooling in a plug-and-play manner, delivering more accurate classification results. Extensive experiments across several zero-shot benchmarks demonstrate significant performance gains for the proposed Mask-Adapter on several well-established methods. Notably, Mask-Adapter also extends effectively to SAM and achieves impressive results on several open-vocabulary segmentation datasets. Code and models are available at https://github.com/hustvl/MaskAdapter.
Authors: Ziyi Wu, Aliaksandr Siarohin, Willi Menapace, Ivan Skorokhodov, Yuwei Fang, Varnith Chordia, Igor Gilitschenski, Sergey Tulyakov
Abstract: Real-world videos consist of sequences of events. Generating such sequences with precise temporal control is infeasible with existing video generators that rely on a single paragraph of text as input. When tasked with generating multiple events described using a single prompt, such methods often ignore some of the events or fail to arrange them in the correct order. To address this limitation, we present MinT, a multi-event video generator with temporal control. Our key insight is to bind each event to a specific period in the generated video, which allows the model to focus on one event at a time. To enable time-aware interactions between event captions and video tokens, we design a time-based positional encoding method, dubbed ReRoPE. This encoding helps to guide the cross-attention operation. By fine-tuning a pre-trained video diffusion transformer on temporally grounded data, our approach produces coherent videos with smoothly connected events. For the first time in the literature, our model offers control over the timing of events in generated videos. Extensive experiments demonstrate that MinT outperforms existing commercial and open-source models by a large margin.
Authors: Tiancheng Li, Weijian Luo, Zhiyang Chen, Liyuan Ma, Guo-Jun Qi
Abstract: Proper guidance strategies are essential to achieve high-quality generation results without retraining diffusion and flow-based text-to-image models. Existing guidance either requires specific training or strong inductive biases of diffusion model networks, potentially limiting their applications. Motivated by the observation that artifact outliers can be detected by a significant decline in the density from a noisier to a cleaner noise level, we propose Self-Guidance (SG), which improves the image quality by suppressing the generation of low-quality samples. SG only relies on the sampling probabilities of its own diffusion model at different noise levels with no need of any guidance-specific training. This makes it flexible to be used in a plug-and-play manner with other sampling algorithms, maximizing its potential to achieve competitive performances in many generative tasks. We conduct experiments on text-to-image and text-to-video generation with different architectures, including UNet and transformer models. With open-sourced diffusion models such as Stable Diffusion 3.5 and FLUX, Self-Guidance surpasses existing algorithms on multiple metrics, including both FID and Human Preference Score. Moreover, we find that SG has a surprisingly positive effect on the generation of physiologically correct human body structures such as hands, faces, and arms, showing its ability of eliminating human body artifacts with minimal efforts. We will release our code along with this paper.
Authors: Yunheng Li, Yuxuan Li, Quansheng Zeng, Wenhai Wang, Qibin Hou, Ming-Ming Cheng
Abstract: Pre-trained vision-language models (VLMs), such as CLIP, have demonstrated impressive zero-shot recognition capability, but still underperform in dense prediction tasks. Self-distillation recently is emerging as a promising approach for fine-tuning VLMs to better adapt to local regions without requiring extensive annotations. However, previous state-of-the-art approaches often suffer from significant `foreground bias', where models tend to wrongly identify background regions as foreground objects. To alleviate this issue, we propose DenseVLM, a framework designed to learn unbiased region-language alignment from powerful pre-trained VLM representations. To alleviate this issue, we propose DenseVLM, a framework designed to learn unbiased region-language alignment from powerful pre-trained VLM representations. DenseVLM leverages the pre-trained VLM to retrieve categories for unlabeled regions and then decouples the interference between foreground and background features. We show that DenseVLM can directly replace the original VLM in open-vocabulary object detection and image segmentation methods, leading to notable performance improvements. Furthermore, it exhibits promising zero-shot scalability when training on more extensive and diverse datasets. Our code is available at https://github.com/HVision-NKU/DenseVLM.
Authors: Ilya A. Petrov, Riccardo Marin, Julian Chibane, Gerard Pons-Moll
Abstract: Modeling 3D human-object interaction (HOI) is a problem of great interest for computer vision and a key enabler for virtual and mixed-reality applications. Existing methods work in a one-way direction: some recover plausible human interactions conditioned on a 3D object; others recover the object pose conditioned on a human pose. Instead, we provide the first unified model - TriDi which works in any direction. Concretely, we generate Human, Object, and Interaction modalities simultaneously with a new three-way diffusion process, allowing to model seven distributions with one network. We implement TriDi as a transformer attending to the various modalities' tokens, thereby discovering conditional relations between them. The user can control the interaction either as a text description of HOI or a contact map. We embed these two representations into a shared latent space, combining the practicality of text descriptions with the expressiveness of contact maps. Using a single network, TriDi unifies all the special cases of prior work and extends to new ones, modeling a family of seven distributions. Remarkably, despite using a single model, TriDi generated samples surpass one-way specialized baselines on GRAB and BEHAVE in terms of both qualitative and quantitative metrics, and demonstrating better diversity. We show the applicability of TriDi to scene population, generating objects for human-contact datasets, and generalization to unseen object geometry. The project page is available at: https://virtualhumans.mpi-inf.mpg.de/tridi.
Authors: Ellington Kirby, Mickael Chen, Renaud Marlet, Nermin Samet
Abstract: The generation of LiDAR scans is a growing area of research with diverse applications to autonomous driving. However, scan generation remains challenging, especially when compared to the rapid advancement of 2D and 3D object generation. We introduce a novel task: LiDAR object generation, requiring models to produce 3D objects as viewed by a LiDAR scan. This task focuses LiDAR scan generation on the most interesting aspect of scenes, the objects, while also benefiting from advancements in 3D object generative methods. We introduce a novel diffusion-based model to produce LiDAR point clouds of dataset objects, including intensity, and with an extensive control of the generation via conditioning information. Our experiments on nuScenes show the quality of our generations measured with new 3D metrics developed to suit LiDAR objects.
Authors: Mu Zhang, Yunfan Liu, Yue Liu, Yuzhong Zhao, Qixiang Ye
Abstract: Existing image synthesis methods for natural scenes focus primarily on foreground control, often reducing the background to simplistic textures. Consequently, these approaches tend to overlook the intrinsic correlation between foreground and background, which may lead to incoherent and unrealistic synthesis results in remote sensing (RS) scenarios. In this paper, we introduce CC-Diff, a $\underline{\textbf{Diff}}$usion Model-based approach for RS image generation with enhanced $\underline{\textbf{C}}$ontext $\underline{\textbf{C}}$oherence. Specifically, we propose a novel Dual Re-sampler for feature extraction, with a built-in `Context Bridge' to explicitly capture the intricate interdependency between foreground and background. Moreover, we reinforce their connection by employing a foreground-aware attention mechanism during the generation of background features, thereby enhancing the plausibility of the synthesized context. Extensive experiments show that CC-Diff outperforms state-of-the-art methods across critical quality metrics, excelling in the RS domain and effectively generalizing to natural images. Remarkably, CC-Diff also shows high trainability, boosting detection accuracy by 1.83 mAP on DOTA and 2.25 mAP on the COCO benchmark.
Authors: Vahid Balazadeh, Mohammadmehdi Ataei, Hyunmin Cheong, Amir Hosein Khasahmadi, Rahul G. Krishnan
Abstract: Physical reasoning, which involves interpreting object behaviors within dynamic environments, remains a significant challenge for Vision-Language Models (VLMs). The limitations in physical reasoning arise from an inability to translate learned knowledge into predictions about physical behavior. We perform a careful study to show how continual fine-tuning can mitigate this issue. However, fine-tuning is expensive for large models and impractical to repeatedly perform for every task. This necessitates the creation of modular and scalable ways to teach VLMs about physical reasoning. To that end, we introduce Physics Context Builders (PCBs), a novel modular framework where specialized VLMs are fine-tuned to generate detailed physical scene descriptions. These can be used as physical contexts for larger VLMs to enhance their reasoning capabilities. PCBs enable the separation of visual perception from reasoning, allowing us to analyze their relative contributions to physical understanding. We perform careful experiments on CLEVRER and on Falling Tower, a stability detection dataset with both simulated and real-world scenes, to demonstrate that PCBs provide substantial performance improvements, increasing average accuracy by up to 13.8% on complex physical reasoning tasks. Notably, PCBs show strong Sim2Real transfer, successfully generalizing from simulated training data to real-world scenes. Our work demonstrates that enhancing visual perception through modular, simulation-trained components offers a practical approach to improving physical reasoning in VLMs, while providing insights into the factors affecting physical understanding in these models.
Authors: Huawei Lin, Yingjie Lao, Weijie Zhao
Abstract: Identifying the training data samples that most influence a generated image is a critical task in understanding diffusion models (DMs), yet existing influence estimation methods are constrained to small-scale or LoRA-tuned models due to computational limitations. To address this challenge, we propose DMin (Diffusion Model influence), a scalable framework for estimating the influence of each training data sample on a given generated image. To the best of our knowledge, it is the first method capable of influence estimation for DMs with billions of parameters. Leveraging efficient gradient compression, DMin reduces storage requirements from hundreds of TBs to mere MBs or even KBs, and retrieves the top-k most influential training samples in under 1 second, all while maintaining performance. Our empirical results demonstrate DMin is both effective in identifying influential training samples and efficient in terms of computational and storage requirements.
Authors: Jaehwan Jeong, Sumin In, Sieun Kim, Hannie Shin, Jongheon Jeong, Sang Ho Yoon, Jaewook Chung, Sangpil Kim
Abstract: The rising use of deepfakes in criminal activities presents a significant issue, inciting widespread controversy. While numerous studies have tackled this problem, most primarily focus on deepfake detection. These reactive solutions are insufficient as a fundamental approach for crimes where authenticity is disregarded. Existing proactive defenses also have limitations, as they are effective only for deepfake models based on specific Generative Adversarial Networks (GANs), making them less applicable in light of recent advancements in diffusion-based models. In this paper, we propose a proactive defense method named FaceShield, which introduces novel defense strategies targeting deepfakes generated by Diffusion Models (DMs) and facilitates defenses on various existing GAN-based deepfake models through facial feature extractor manipulations. Our approach consists of three main components: (i) manipulating the attention mechanism of DMs to exclude protected facial features during the denoising process, (ii) targeting prominent facial feature extraction models to enhance the robustness of our adversarial perturbation, and (iii) employing Gaussian blur and low-pass filtering techniques to improve imperceptibility while enhancing robustness against JPEG compression. Experimental results on the CelebA-HQ and VGGFace2-HQ datasets demonstrate that our method achieves state-of-the-art performance against the latest deepfake models based on DMs, while also exhibiting transferability to GANs and showcasing greater imperceptibility of noise along with enhanced robustness.
Authors: Xinhao Zhong, Shuoyang Sun, Xulin Gu, Zhaoyang Xu, Yaowei Wang, Min Zhang, Bin Chen
Abstract: Dataset distillation offers an efficient way to reduce memory and computational costs by optimizing a smaller dataset with performance comparable to the full-scale original. However, for large datasets and complex deep networks (e.g., ImageNet-1K with ResNet-101), the extensive optimization space limits performance, reducing its practicality. Recent approaches employ pre-trained diffusion models to generate informative images directly, avoiding pixel-level optimization and achieving notable results. However, these methods often face challenges due to distribution shifts between pre-trained models and target datasets, along with the need for multiple distillation steps across varying settings. To address these issues, we propose a novel framework orthogonal to existing diffusion-based distillation methods, leveraging diffusion models for selection rather than generation. Our method starts by predicting noise generated by the diffusion model based on input images and text prompts (with or without label text), then calculates the corresponding loss for each pair. With the loss differences, we identify distinctive regions of the original images. Additionally, we perform intra-class clustering and ranking on selected patches to maintain diversity constraints. This streamlined framework enables a single-step distillation process, and extensive experiments demonstrate that our approach outperforms state-of-the-art methods across various metrics.
Authors: Zeyuan Yang, Delin Chen, Xueyang Yu, Maohao Shen, Chuang Gan
Abstract: Long video understanding poses unique challenges due to their temporal complexity and low information density. Recent works address this task by sampling numerous frames or incorporating auxiliary tools using LLMs, both of which result in high computational costs. In this work, we introduce a curiosity-driven video agent with self-exploration capability, dubbed as VCA. Built upon VLMs, VCA autonomously navigates video segments and efficiently builds a comprehensive understanding of complex video sequences. Instead of directly sampling frames, VCA employs a tree-search structure to explore video segments and collect frames. Rather than relying on external feedback or reward, VCA leverages VLM's self-generated intrinsic reward to guide its exploration, enabling it to capture the most crucial information for reasoning. Experimental results on multiple long video benchmarks demonstrate our approach's superior effectiveness and efficiency.
Authors: Wenhao Sun, Rong-Cheng Tu, Jingyi Liao, Zhao Jin, Dacheng Tao
Abstract: Diffusion Transformers (DiTs) have proven effective in generating high-quality videos but are hindered by high computational costs. Existing video DiT sampling acceleration methods often rely on costly fine-tuning or exhibit limited generalization capabilities. We propose Asymmetric Reduction and Restoration (AsymRnR), a training-free and model-agnostic method to accelerate video DiTs. It builds on the observation that redundancies of feature tokens in DiTs vary significantly across different model blocks, denoising steps, and feature types. Our AsymRnR asymmetrically reduces redundant tokens in the attention operation, achieving acceleration with negligible degradation in output quality and, in some cases, even improving it. We also tailored a reduction schedule to distribute the reduction across components adaptively. To further accelerate this process, we introduce a matching cache for more efficient reduction. Backed by theoretical foundations and extensive experimental validation, AsymRnR integrates into state-of-the-art video DiTs and offers substantial speedup.
Authors: Chengzhou Yu (South China University of Technology), Huihui Fang (Pazhou Laboratory), Hongqiu Wang (The Hong Kong University of Science,Technology), Ting Deng (South China University of Technology), Qing Du (South China University of Technology), Yanwu Xu (South China University of Technology), Weihua Yang (Shenzhen Eye Hospital)
Abstract: Fundus imaging is a critical tool in ophthalmology, with different imaging modalities offering unique advantages. For instance, fundus fluorescein angiography (FFA) can accurately identify eye diseases. However, traditional invasive FFA involves the injection of sodium fluorescein, which can cause discomfort and risks. Generating corresponding FFA images from non-invasive fundus images holds significant practical value but also presents challenges. First, limited datasets constrain the performance and effectiveness of models. Second, previous studies have primarily focused on generating FFA for single diseases or single modalities, often resulting in poor performance for patients with various ophthalmic conditions. To address these issues, we propose a novel latent diffusion model-based framework, Diffusion, which introduces a fine-tuning protocol to overcome the challenge of limited medical data and unleash the generative capabilities of diffusion models. Furthermore, we designed a new approach to tackle the challenges of generating across different modalities and disease types. On limited datasets, our framework achieves state-of-the-art results compared to existing methods, offering significant potential to enhance ophthalmic diagnostics and patient care. Our code will be released soon to support further research in this field.
Authors: Xing Liufu, Chaolei Tan, Xiaotong Lin, Yonggang Qi, Jinxuan Li, Jian-Fang Hu
Abstract: Edge labels are typically at various granularity levels owing to the varying preferences of annotators, thus handling the subjectivity of per-pixel labels has been a focal point for edge detection. Previous methods often employ a simple voting strategy to diminish such label uncertainty or impose a strong assumption of labels with a pre-defined distribution, e.g., Gaussian. In this work, we unveil that the segment anything model (SAM) provides strong prior knowledge to model the uncertainty in edge labels. Our key insight is that the intermediate SAM features inherently correspond to object edges at various granularities, which reflects different edge options due to uncertainty. Therefore, we attempt to align uncertainty with granularity by regressing intermediate SAM features from different layers to object edges at multi-granularity levels. In doing so, the model can fully and explicitly explore diverse ``uncertainties'' in a data-driven fashion. Specifically, we inject a lightweight module (~ 1.5% additional parameters) into the frozen SAM to progressively fuse and adapt its intermediate features to estimate edges from coarse to fine. It is crucial to normalize the granularity level of human edge labels to match their innate uncertainty. For this, we simply perform linear blending to the real edge labels at hand to create pseudo labels with varying granularities. Consequently, our uncertainty-aligned edge detector can flexibly produce edges at any desired granularity (including an optimal one). Thanks to SAM, our model uniquely demonstrates strong generalizability for cross-dataset edge detection. Extensive experimental results on BSDS500, Muticue and NYUDv2 validate our model's superiority.
Authors: Nikitha SR, Tarun Ram Menta, Mausoom Sarkar
Abstract: The advent of multimodal learning has brought a significant improvement in document AI. Documents are now treated as multimodal entities, incorporating both textual and visual information for downstream analysis. However, works in this space are often focused on the textual aspect, using the visual space as auxiliary information. While some works have explored pure vision based techniques for document image understanding, they require OCR identified text as input during inference, or do not align with text in their learning procedure. Therefore, we present a novel image-text alignment technique specially designed for leveraging the textual information in document images to improve performance on visual tasks. Our document encoder model DoPTA - trained with this technique demonstrates strong performance on a wide range of document image understanding tasks, without requiring OCR during inference. Combined with an auxiliary reconstruction objective, DoPTA consistently outperforms larger models, while using significantly lesser pre-training compute. DoPTA also sets new state-of-the art results on D4LA, and FUNSD, two challenging document visual analysis benchmarks.
Authors: Zihui Cheng, Qiguang Chen, Jin Zhang, Hao Fei, Xiaocheng Feng, Wanxiang Che, Min Li, Libo Qin
Abstract: Large Vision-Language Models (LVLMs) have recently demonstrated amazing success in multi-modal tasks, including advancements in Multi-modal Chain-of-Thought (MCoT) reasoning. Despite these successes, current benchmarks still follow a traditional paradigm with multi-modal input and text-modal output, which leads to significant drawbacks such as missing visual operations and vague expressions. Motivated by this, we introduce a novel Chain of Multi-modal Thought (CoMT) benchmark to address these limitations. Different from the traditional MCoT benchmark, CoMT requires both multi-modal input and multi-modal reasoning output, aiming to mimic human-like reasoning that inherently integrates visual operation. Specifically, CoMT consists of four categories: (1) Visual Creation, (2) Visual Deletion, (3) Visual Update, and (4) Visual Selection to comprehensively explore complex visual operations and concise expression in real scenarios. We evaluate various LVLMs and strategies on CoMT, revealing some key insights into the capabilities and limitations of the current approaches. We hope that CoMT can inspire more research on introducing multi-modal generation into the reasoning process.
Authors: Vasiliki Sideri-Lampretsa, Nil Stolt-Ans\'o, Huaqi Qiu, Julian McGinnis, Wenke Karbole, Martin Menten, Daniel Rueckert
Abstract: Deformable image registration poses a challenging problem where, unlike most deep learning tasks, a complex relationship between multiple coordinate systems has to be considered. Although data-driven methods have shown promising capabilities to model complex non-linear transformations, existing works employ standard deep learning architectures assuming they are general black-box solvers. We argue that understanding how learned operations perform pattern-matching between the features in the source and target domains is the key to building robust, data-efficient, and interpretable architectures. We present a theoretical foundation for designing an interpretable registration framework: separated feature extraction and deformation modeling, dynamic receptive fields, and a data-driven deformation functions awareness of the relationship between both spatial domains. Based on this foundation, we formulate an end-to-end process that refines transformations in a coarse-to-fine fashion. Our architecture employs spatially continuous deformation modeling functions that use geometric deep-learning principles, therefore avoiding the problematic approach of resampling to a regular grid between successive refinements of the transformation. We perform a qualitative investigation to highlight interesting interpretability properties of our architecture. We conclude by showing significant improvement in performance metrics over state-of-the-art approaches for both mono- and multi-modal inter-subject brain registration, as well as the challenging task of longitudinal retinal intra-subject registration. We make our code publicly available
Authors: Hanzhe Liang, Guoyang Xie, Chengbin Hou, Bingshu Wang, Can Gao, Jinbao Wang
Abstract: 3D anomaly detection has recently become a significant focus in computer vision. Several advanced methods have achieved satisfying anomaly detection performance. However, they typically concentrate on the external structure of 3D samples and struggle to leverage the internal information embedded within samples. Inspired by the basic intuition of why not look inside for more, we introduce a straightforward method named Internal Spatial Modality Perception~(ISMP) to explore the feature representation from internal views fully. Specifically, our proposed ISMP consists of a critical perception module, Spatial Insight Engine~(SIE), which abstracts complex internal information of point clouds into essential global features. Besides, to better align structural information with point data, we propose an enhanced key point feature extraction module for amplifying spatial structure feature representation. Simultaneously, a novel feature filtering module is incorporated to reduce noise and redundant features for further aligning precise spatial structure. Extensive experiments validate the effectiveness of our proposed method, achieving object-level and pixel-level AUROC improvements of 3.2\% and 13.1\%, respectively, on the Real3D-AD benchmarks. Note that the strong generalization ability of SIE has been theoretically proven and is verified in both classification and segmentation tasks.
Authors: Yuning Peng, Haiping Wang, Yuan Liu, Chenglu Wen, Zhen Dong, Bisheng Yang
Abstract: 3D open-vocabulary scene understanding, which accurately perceives complex semantic properties of objects in space, has gained significant attention in recent years. In this paper, we propose GAGS, a framework that distills 2D CLIP features into 3D Gaussian splatting, enabling open-vocabulary queries for renderings on arbitrary viewpoints. The main challenge of distilling 2D features for 3D fields lies in the multiview inconsistency of extracted 2D features, which provides unstable supervision for the 3D feature field. GAGS addresses this challenge with two novel strategies. First, GAGS associates the prompt point density of SAM with the camera distances, which significantly improves the multiview consistency of segmentation results. Second, GAGS further decodes a granularity factor to guide the distillation process and this granularity factor can be learned in a unsupervised manner to only select the multiview consistent 2D features in the distillation process. Experimental results on two datasets demonstrate significant performance and stability improvements of GAGS in visual grounding and semantic segmentation, with an inference speed 2$\times$ faster than baseline methods. The code and additional results are available at https://pz0826.github.io/GAGS-Webpage/ .
Authors: Dimitrios Mallis, Ahmet Serdar Karadeniz, Sebastian Cavada, Danila Rukhovich, Niki Foteinopoulou, Kseniya Cherenkova, Anis Kacem, Djamila Aouada
Abstract: We propose CAD-Assistant, a general-purpose CAD agent for AI-assisted design. Our approach is based on a powerful Vision and Large Language Model (VLLM) as a planner and a tool-augmentation paradigm using CAD-specific tools. CAD-Assistant addresses multimodal user queries by generating actions that are iteratively executed on a Python interpreter equipped with the FreeCAD software, accessed via its Python API. Our framework is able to assess the impact of generated CAD commands on geometry and adapts subsequent actions based on the evolving state of the CAD design. We consider a wide range of CAD-specific tools including a sketch image parameterizer, rendering modules, a 2D cross-section generator, and other specialized routines. CAD-Assistant is evaluated on multiple CAD benchmarks, where it outperforms VLLM baselines and supervised task-specific methods. Beyond existing benchmarks, we qualitatively demonstrate the potential of tool-augmented VLLMs as general-purpose CAD solvers across diverse workflows.
Authors: Xijun Wang, Prateek Chennuri, Yu Yuan, Bole Ma, Xingguang Zhang, Stanley Chan
Abstract: While smartphone cameras today can produce astonishingly good photos, their performance in low light is still not completely satisfactory because of the fundamental limits in photon shot noise and sensor read noise. Generative image restoration methods have demonstrated promising results compared to traditional methods, but they suffer from hallucinatory content generation when the signal-to-noise ratio (SNR) is low. Recognizing the availability of personalized photo galleries on users' smartphones, we propose Personalized Generative Denoising (PGD) by building a diffusion model customized for different users. Our core innovation is an identity-consistent physical buffer that extracts the physical attributes of the person from the gallery. This ID-consistent physical buffer provides a strong prior that can be integrated with the diffusion model to restore the degraded images, without the need of fine-tuning. Over a wide range of low-light testing scenarios, we show that PGD achieves superior image denoising and enhancement performance compared to existing diffusion-based denoising approaches.
Authors: Hao Huang, Yujie Lin, Siyu Chen, Haiyang Liu
Abstract: Skeleton-based action recognition using GCNs has achieved remarkable performance, but recognizing ambiguous actions, such as "waving" and "saluting", remains a significant challenge. Existing methods typically rely on a serial combination of GCNs and TCNs, where spatial and temporal features are extracted independently, leading to an unbalanced spatial-temporal information, which hinders accurate action recognition. Moreover, existing methods for ambiguous actions often overemphasize local details, resulting in the loss of crucial global context, which further complicates the task of differentiating ambiguous actions. To address these challenges, we propose a lightweight plug-and-play module called SF-Head, inserted between GCN and TCN layers. SF-Head first conducts SSTE with a Feature Redundancy Loss (F-RL), ensuring a balanced interaction. It then performs AC-FA, with a Feature Consistency Loss (F-CL), which aligns the aggregated feature with their original spatial-temporal feature. Experimental results on NTU RGB+D 60, NTU RGB+D 120, NW-UCLA and PKU-MMD I datasets demonstrate significant improvements in distinguishing ambiguous actions.
Authors: Or Madar, Ohad Fried
Abstract: Image tiling -- the seamless connection of disparate images to create a coherent visual field -- is crucial for applications such as texture creation, video game asset development, and digital art. Traditionally, tiles have been constructed manually, a method that poses significant limitations in scalability and flexibility. Recent research has attempted to automate this process using generative models. However, current approaches primarily focus on tiling textures and manipulating models for single-image generation, without inherently supporting the creation of multiple interconnected tiles across diverse domains. This paper presents Tiled Diffusion, a novel approach that extends the capabilities of diffusion models to accommodate the generation of cohesive tiling patterns across various domains of image synthesis that require tiling. Our method supports a wide range of tiling scenarios, from self-tiling to complex many-to-many connections, enabling seamless integration of multiple images. Tiled Diffusion automates the tiling process, eliminating the need for manual intervention and enhancing creative possibilities in various applications, such as seamlessly tiling of existing images, tiled texture creation, and 360{\deg} synthesis.
Authors: Mushfiqur Rahman Abir, Md. Tanzib Hosain, Md. Abdullah-Al-Jubair, M. F. Mridha
Abstract: Human behavior and interactions are profoundly influenced by visual stimuli present in their surroundings. This influence extends to various aspects of life, notably food consumption and selection. In our study, we employed various models to extract different attributes from the environmental images. Specifically, we identify five key attributes and employ an ensemble model IMVB7 based on five distinct models for some of their detection resulted 0.85 mark. In addition, we conducted surveys to discern patterns in food preferences in response to visual stimuli. Leveraging the insights gleaned from these surveys, we formulate recommendations using decision tree for dishes based on the amalgamation of identified attributes resulted IMVB7t 0.96 mark. This study serves as a foundational step, paving the way for further exploration of this interdisciplinary domain.
Authors: Zhaoxing Zhang, Junda Cheng, Gangwei Xu, Xiaoxiang Wang, Can Zhang, Xin Yang
Abstract: Recent approaches to VO have significantly improved performance by using deep networks to predict optical flow between video frames. However, existing methods still suffer from noisy and inconsistent flow matching, making it difficult to handle challenging scenarios and long-sequence estimation. To overcome these challenges, we introduce Spatio-Temporal Visual Odometry (STVO), a novel deep network architecture that effectively leverages inherent spatio-temporal cues to enhance the accuracy and consistency of multi-frame flow matching. With more accurate and consistent flow matching, STVO can achieve better pose estimation through the bundle adjustment (BA). Specifically, STVO introduces two innovative components: 1) the Temporal Propagation Module that utilizes multi-frame information to extract and propagate temporal cues across adjacent frames, maintaining temporal consistency; 2) the Spatial Activation Module that utilizes geometric priors from the depth maps to enhance spatial consistency while filtering out excessive noise and incorrect matches. Our STVO achieves state-of-the-art performance on TUM-RGBD, EuRoc MAV, ETH3D and KITTI Odometry benchmarks. Notably, it improves accuracy by 77.8% on ETH3D benchmark and 38.9% on KITTI Odometry benchmark over the previous best methods.
Authors: Hongsong Wang, Andi Xu, Pinle Ding, Jie Gui
Abstract: Video Anomaly Detection (VAD) is essential for computer vision research. Existing VAD methods utilize either reconstruction-based or prediction-based frameworks. The former excels at detecting irregular patterns or structures, whereas the latter is capable of spotting abnormal deviations or trends. We address pose-based video anomaly detection and introduce a novel framework called Dual Conditioned Motion Diffusion (DCMD), which enjoys the advantages of both approaches. The DCMD integrates conditioned motion and conditioned embedding to comprehensively utilize the pose characteristics and latent semantics of observed movements, respectively. In the reverse diffusion process, a motion transformer is proposed to capture potential correlations from multi-layered characteristics within the spectrum space of human motion. To enhance the discriminability between normal and abnormal instances, we design a novel United Association Discrepancy (UAD) regularization that primarily relies on a Gaussian kernel-based time association and a self-attention-based global association. Finally, a mask completion strategy is introduced during the inference stage of the reverse diffusion process to enhance the utilization of conditioned motion for the prediction branch of anomaly detection. Extensive experiments on four datasets demonstrate that our method dramatically outperforms state-of-the-art methods and exhibits superior generalization performance.
Authors: Yidi Shao, Mu Huang, Chen Change Loy, Bo Dai
Abstract: We introduce GausSim, a novel neural network-based simulator designed to capture the dynamic behaviors of real-world elastic objects represented through Gaussian kernels. We leverage continuum mechanics and treat each kernel as a Center of Mass System (CMS) that represents continuous piece of matter, accounting for realistic deformations without idealized assumptions. To improve computational efficiency and fidelity, we employ a hierarchical structure that further organizes kernels into CMSs with explicit formulations, enabling a coarse-to-fine simulation approach. This structure significantly reduces computational overhead while preserving detailed dynamics. In addition, GausSim incorporates explicit physics constraints, such as mass and momentum conservation, ensuring interpretable results and robust, physically plausible simulations. To validate our approach, we present a new dataset, READY, containing multi-view videos of real-world elastic deformations. Experimental results demonstrate that GausSim achieves superior performance compared to existing physics-driven baselines, offering a practical and accurate solution for simulating complex dynamic behaviors. Code and model will be released. Project page: https://www.mmlab-ntu.com/project/gausim/index.html .
Authors: Changwoon Choi (Seoul National University), Jeongjun Kim (Seoul National University), Geonho Cha (NAVER Cloud), Minkwan Kim (Seoul National University), Dongyoon Wee (NAVER Cloud), Young Min Kim (Seoul National University)
Abstract: Recent works on dynamic 3D neural field reconstruction assume the input from synchronized multi-view videos whose poses are known. The input constraints are often not satisfied in real-world setups, making the approach impractical. We show that unsynchronized videos from unknown poses can generate dynamic neural fields as long as the videos capture human motion. Humans are one of the most common dynamic subjects captured in videos, and their shapes and poses can be estimated using state-of-the-art libraries. While noisy, the estimated human shape and pose parameters provide a decent initialization point to start the highly non-convex and under-constrained problem of training a consistent dynamic neural representation. Given the shape and pose parameters of humans in individual frames, we formulate methods to calculate the time offsets between videos, followed by camera pose estimations that analyze the 3D joint positions. Then, we train the dynamic neural fields employing multiresolution grids while we concurrently refine both time offsets and camera poses. The setup still involves optimizing many parameters; therefore, we introduce a robust progressive learning strategy to stabilize the process. Experiments show that our approach achieves accurate spatio-temporal calibration and high-quality scene reconstruction in challenging conditions.
Authors: Jingbo Lin, Zhilu Zhang, Wenbo Li, Renjing Pei, Hang Xu, Hongzhi Zhang, Wangmeng Zuo
Abstract: Recently, considerable progress has been made in all-in-one image restoration. Generally, existing methods can be degradation-agnostic or degradation-aware. However, the former are limited in leveraging degradation-specific restoration, and the latter suffer from the inevitable error in degradation estimation. Consequently, the performance of existing methods has a large gap compared to specific single-task models. In this work, we make a step forward in this topic, and present our UniRestorer with improved restoration performance. Specifically, we perform hierarchical clustering on degradation space, and train a multi-granularity mixture-of-experts (MoE) restoration model. Then, UniRestorer adopts both degradation and granularity estimation to adaptively select an appropriate expert for image restoration. In contrast to existing degradation-agnostic and -aware methods, UniRestorer can leverage degradation estimation to benefit degradation specific restoration, and use granularity estimation to make the model robust to degradation estimation error. Experimental results show that our UniRestorer outperforms state-of-the-art all-in-one methods by a large margin, and is promising in closing the performance gap to specific single task models. The code and pre-trained models will be publicly available at https://github.com/mrluin/UniRestorer.
Authors: Zhiyu Tan, WenXu Qian, Hesen Chen, Mengping Yang, Lei Chen, Hao Li
Abstract: Diffusion models have established themselves as the de facto primary paradigm in visual generative modeling, revolutionizing the field through remarkable success across various diverse applications ranging from high-quality image synthesis to temporal aware video generation. Despite these advancements, three fundamental limitations persist, including 1) discrepancy between training and inference processes, 2) progressive information leakage throughout the noise corruption procedures, and 3) inherent constraints preventing effective integration of modern optimization criteria like perceptual and adversarial loss. To mitigate these critical challenges, we in this paper present a novel end-to-end learning paradigm that establishes direct optimization from the final generated samples to initial noises. Our proposed End-to-End Differentiable Diffusion, dubbed E2ED^2, introduces several key improvements: it eliminates the sequential training-sampling mismatch and intermediate information leakage via conceptualizing training as a direct transformation from isotropic Gaussian noise to the target data distribution. Additionally, such training framework enables seamless incorporation of adversarial and perceptual losses into the core optimization objective. Comprehensive evaluation across standard benchmarks including COCO30K and HW30K reveals that our method achieves substantial performance gains in terms of Fr\'echet Inception Distance (FID) and CLIP score, even with fewer sampling steps (less than 4). Our findings highlight that the end-to-end mechanism might pave the way for more robust and efficient solutions, \emph{i.e.,} combining diffusion stability with GAN-like discriminative optimization in an end-to-end manner.
Authors: Xinhao Li, Yi Wang, Jiashuo Yu, Xiangyu Zeng, Yuhan Zhu, Haian Huang, Jianfei Gao, Kunchang Li, Yinan He, Chenting Wang, Yu Qiao, Yali Wang, Limin Wang
Abstract: Long-context video modeling is critical for multimodal large language models (MLLMs), enabling them to process movies, online video streams, and so on. Despite its advances, handling long videos remains challenging due to the difficulty in efficiently understanding the extremely long video context. This paper aims to address this issue from aspects of model architecture, training data, training strategy and evaluation benchmark. First, we propose a novel Hierarchical video token Compression (HiCo) method, which leverages visual redundancy in long videos to compress long video context from Clip-level to Video-level, reducing the computation significantly while preserving essential details, achieving an extreme compression ratio of approximately 1/50 with almost no performance loss. Second, we introduce a multi-stage short-to-long learning scheme, a large-scale dataset of real-world long videos named LongVid, and a challenging ``Multi-Hop Needle-In-A-Video-Haystack'' benchmark. Finally, we build a powerful video MLLM named VideoChat-Flash, which shows a leading performance on both mainstream long and short video benchmarks at the 2B and 7B model scale. It first gets 99.1% accuracy over 10,000 frames in NIAH among open-source models.
Authors: Jingfeng Yao, Bin Yang, Xinggang Wang
Abstract: Latent diffusion models with Transformer architectures excel at generating high-fidelity images. However, recent studies reveal an optimization dilemma in this two-stage design: while increasing the per-token feature dimension in visual tokenizers improves reconstruction quality, it requires substantially larger diffusion models and more training iterations to achieve comparable generation performance. Consequently, existing systems often settle for sub-optimal solutions, either producing visual artifacts due to information loss within tokenizers or failing to converge fully due to expensive computation costs. We argue that this dilemma stems from the inherent difficulty in learning unconstrained high-dimensional latent spaces. To address this, we propose aligning the latent space with pre-trained vision foundation models when training the visual tokenizers. Our proposed VA-VAE (Vision foundation model Aligned Variational AutoEncoder) significantly expands the reconstruction-generation frontier of latent diffusion models, enabling faster convergence of Diffusion Transformers (DiT) in high-dimensional latent spaces. To exploit the full potential of VA-VAE, we build an enhanced DiT baseline with improved training strategies and architecture designs, termed LightningDiT. The integrated system achieves state-of-the-art (SOTA) performance on ImageNet 256x256 generation with an FID score of 1.35 while demonstrating remarkable training efficiency by reaching an FID score of 2.11 in just 64 epochs--representing an over 21 times convergence speedup compared to the original DiT. Models and codes are available at: https://github.com/hustvl/LightningDiT.
Authors: Wenyan Cong, Hanqing Zhu, Kevin Wang, Jiahui Lei, Colton Stearns, Yuanhao Cai, Dilin Wang, Rakesh Ranjan, Matt Feiszli, Leonidas Guibas, Zhangyang Wang, Weiyao Wang, Zhiwen Fan
Abstract: Efficiently reconstructing 3D scenes from monocular video remains a core challenge in computer vision, vital for applications in virtual reality, robotics, and scene understanding. Recently, frame-by-frame progressive reconstruction without camera poses is commonly adopted, incurring high computational overhead and compounding errors when scaling to longer videos. To overcome these issues, we introduce VideoLifter, a novel video-to-3D pipeline that leverages a local-to-global strategy on a fragment basis, achieving both extreme efficiency and SOTA quality. Locally, VideoLifter leverages learnable 3D priors to register fragments, extracting essential information for subsequent 3D Gaussian initialization with enforced inter-fragment consistency and optimized efficiency. Globally, it employs a tree-based hierarchical merging method with key frame guidance for inter-fragment alignment, pairwise merging with Gaussian point pruning, and subsequent joint optimization to ensure global consistency while efficiently mitigating cumulative errors. This approach significantly accelerates the reconstruction process, reducing training time by over 82% while holding better visual quality than current SOTA methods.
Authors: Tingyang Zhang, Chen Wang, Zhiyang Dou, Qingzhe Gao, Jiahui Lei, Baoquan Chen, Lingjie Liu
Abstract: We propose ProTracker, a novel framework for accurate and robust long-term dense tracking of arbitrary points in videos. Previous methods relying on global cost volumes effectively handle large occlusions and scene changes but lack precision and temporal awareness. In contrast, local iteration-based methods accurately track smoothly transforming scenes but face challenges with occlusions and drift. To address these issues, we propose a probabilistic framework that marries the strengths of both paradigms by leveraging local optical flow for predictions and refined global heatmaps for observations. This design effectively combines global semantic information with temporally aware low-level features, enabling precise and robust long-term tracking of arbitrary points in videos. Extensive experiments demonstrate that ProTracker attains state-of-the-art performance among optimization-based approaches and surpasses supervised feed-forward methods on multiple benchmarks. The code and model will be released after publication.
Authors: Guozhen Zhang, Yuhan Zhu, Yutao Cui, Xiaotong Zhao, Kai Ma, Limin Wang
Abstract: Flow-based frame interpolation methods ensure motion stability through estimated intermediate flow but often introduce severe artifacts in complex motion regions. Recent generative approaches, boosted by large-scale pre-trained video generation models, show promise in handling intricate scenes. However, they frequently produce unstable motion and content inconsistencies due to the absence of explicit motion trajectory constraints. To address these challenges, we propose Motion-aware Generative frame interpolation (MoG) that synergizes intermediate flow guidance with generative capacities to enhance interpolation fidelity. Our key insight is to simultaneously enforce motion smoothness through flow constraints while adaptively correcting flow estimation errors through generative refinement. Specifically, we first introduce a dual guidance injection that propagates condition information using intermediate flow at both latent and feature levels, aligning the generated motion with flow-derived motion trajectories. Meanwhile, we implemented two critical designs, encoder-only guidance injection and selective parameter fine-tuning, which enable dynamic artifact correction in the complex motion regions. Extensive experiments on both real-world and animation benchmarks demonstrate that MoG outperforms state-of-the-art methods in terms of video quality and visual fidelity. Our work bridges the gap between flow-based stability and generative flexibility, offering a versatile solution for frame interpolation across diverse scenarios.
Authors: Ulindu De Silva, Didula Samaraweera, Sasini Wanigathunga, Kavindu Kariyawasam, Kanchana Ranasinghe, Muzammal Naseer, Ranga Rodrigo
Abstract: We present Seg-TTO, a novel framework for zero-shot, open-vocabulary semantic segmentation (OVSS), designed to excel in specialized domain tasks. While current open-vocabulary approaches show impressive performance on standard segmentation benchmarks under zero-shot settings, they fall short of supervised counterparts on highly domain-specific datasets. We focus on segmentation-specific test-time optimization to address this gap. Segmentation requires an understanding of multiple concepts within a single image while retaining the locality and spatial structure of representations. We propose a novel self-supervised objective adhering to these requirements and use it to align the model parameters with input images at test time. In the textual modality, we learn multiple embeddings for each category to capture diverse concepts within an image, while in the visual modality, we calculate pixel-level losses followed by embedding aggregation operations specific to preserving spatial structure. Our resulting framework termed Seg-TTO is a plug-and-play module. We integrate Seg-TTO with three state-of-the-art OVSS approaches and evaluate across 22 challenging OVSS tasks covering a range of specialized domains. Our Seg-TTO demonstrates clear performance improvements (up to 27% mIoU increase on some datasets) establishing new state-of-the-art. Our code and models will be released publicly.
Authors: Xiaojie Li, Yibo Yang, Jianlong Wu, Jie Liu, Yue Yu, Liqiang Nie, Min Zhang
Abstract: Few-shot class-incremental learning (FSCIL) involves learning new classes from limited data while retaining prior knowledge, and often results in catastrophic forgetting. Existing methods either freeze backbone networks to preserve knowledge, which limits adaptability, or rely on additional modules or prompts, introducing inference overhead. To this end, we propose Continuous Knowledge-Preserving Decomposition for FSCIL (CKPD-FSCIL), a framework that decomposes a model's weights into two parts: one that compacts existing knowledge (knowledge-sensitive components) and another that carries redundant capacity to accommodate new abilities (redundant-capacity components). The decomposition is guided by a covariance matrix from replay samples, ensuring principal components align with classification abilities. During adaptation, we freeze the knowledge-sensitive components and only adapt the redundant-capacity components, fostering plasticity while minimizing interference without changing the architecture or increasing overhead. Additionally, CKPD introduces an adaptive layer selection strategy to identify layers with redundant capacity, dynamically allocating adapters. Experiments on multiple benchmarks show that CKPD-FSCIL outperforms state-of-the-art methods.
Authors: Yingjie Chen, Yifang Men, Yuan Yao, Miaomiao Cui, Liefeng Bo
Abstract: Motion-controllable image animation is a fundamental task with a wide range of potential applications. Recent works have made progress in controlling camera or object motion via various motion representations, while they still struggle to support collaborative camera and object motion control with adaptive control granularity. To this end, we introduce 3D-aware motion representation and propose an image animation framework, called Perception-as-Control, to achieve fine-grained collaborative motion control. Specifically, we construct 3D-aware motion representation from a reference image, manipulate it based on interpreted user instructions, and perceive it from different viewpoints. In this way, camera and object motions are transformed into intuitive and consistent visual changes. Then, our framework leverages the perception results as motion control signals, enabling it to support various motion-related video synthesis tasks in a unified and flexible way. Experiments demonstrate the superiority of the proposed approach. For more details and qualitative results, please refer to our anonymous project webpage: https://chen-yingjie.github.io/projects/Perception-as-Control.
URLs: https://chen-yingjie.github.io/projects/Perception-as-Control.
Authors: Xianqi Wang, Hao Yang, Gangwei Xu, Junda Cheng, Min Lin, Yong Deng, Jinliang Zang, Yurui Chen, Xin Yang
Abstract: State-of-the-art supervised stereo matching methods have achieved remarkable performance on various benchmarks. However, their generalization to real-world scenarios remains challenging due to the scarcity of annotated real-world stereo data. In this paper, we propose ZeroStereo, a novel stereo image generation pipeline for zero-shot stereo matching. Our approach synthesizes high-quality right images from arbitrary single images by leveraging pseudo disparities generated by a monocular depth estimation model. Unlike previous methods that address occluded regions by filling missing areas with neighboring pixels or random backgrounds, we fine-tune a diffusion inpainting model to recover missing details while preserving semantic structure. Additionally, we propose Training-Free Confidence Generation, which mitigates the impact of unreliable pseudo labels without additional training, and Adaptive Disparity Selection, which ensures a diverse and realistic disparity distribution while preventing excessive occlusion and foreground distortion. Experiments demonstrate that models trained with our pipeline achieve state-of-the-art zero-shot generalization across multiple datasets with only a dataset volume comparable to Scene Flow. Code: https://github.com/Windsrain/ZeroStereo.
Authors: Shiu-hong Kao, Xiao Li, Jinglu Wang, Yang Li, Chi-Keung Tang, Yu-Wing Tai, Yan Lu
Abstract: Large Reconstruction Models (LRMs) have recently become a popular method for creating 3D foundational models. Training 3D reconstruction models with 2D visual data traditionally requires prior knowledge of camera poses for the training samples, a process that is both time-consuming and prone to errors. Consequently, 3D reconstruction training has been confined to either synthetic 3D datasets or small-scale datasets with annotated poses. In this study, we investigate the feasibility of 3D reconstruction using unposed video data of various objects. We introduce UVRM, a novel 3D reconstruction model capable of being trained and evaluated on monocular videos without requiring any information about the pose. UVRM uses a transformer network to implicitly aggregate video frames into a pose-invariant latent feature space, which is then decoded into a tri-plane 3D representation. To obviate the need for ground-truth pose annotations during training, UVRM employs a combination of the score distillation sampling (SDS) method and an analysis-by-synthesis approach, progressively synthesizing pseudo novel-views using a pre-trained diffusion model. We qualitatively and quantitatively evaluate UVRM's performance on the G-Objaverse and CO3D datasets without relying on pose information. Extensive experiments show that UVRM is capable of effectively and efficiently reconstructing a wide range of 3D objects from unposed videos.
Authors: Jan Skvrna, Lukas Neumann
Abstract: Inferring object 3D position and orientation from a single RGB camera is a foundational task in computer vision with many important applications. Traditionally, 3D object detection methods are trained in a fully-supervised setup, requiring LiDAR and vast amounts of human annotations, which are laborious, costly, and do not scale well with the ever-increasing amounts of data being captured. We present a novel method to train a 3D object detector from a single RGB camera without domain-specific human annotations, making orders of magnitude more data available for training. The method uses newly proposed Local Object Motion Model to disentangle object movement source between subsequent frames, is approximately 700 times faster than previous work and compensates camera focal length differences to aggregate multiple datasets. The method is evaluated on three public datasets, where despite using no human labels, it outperforms prior work by a significant margin. It also shows its versatility as a pre-training tool for fully-supervised training and shows that combining pseudo-labels from multiple datasets can achieve comparable accuracy to using human labels from a single dataset. The source code and model will be published soon.
Authors: Xiaohui Li, Yihao Liu, Shuo Cao, Ziyan Chen, Shaobin Zhuang, Xiangyu Chen, Yinan He, Yi Wang, Yu Qiao
Abstract: Diffusion models have demonstrated exceptional capabilities in image restoration, yet their application to video super-resolution (VSR) faces significant challenges in balancing fidelity with temporal consistency. Our evaluation reveals a critical gap: existing approaches consistently fail on severely degraded videos--precisely where diffusion models' generative capabilities are most needed. We identify that existing diffusion-based VSR methods struggle primarily because they face an overwhelming learning burden: simultaneously modeling complex degradation distributions, content representations, and temporal relationships with limited high-quality training data. To address this fundamental challenge, we present DiffVSR, featuring a Progressive Learning Strategy (PLS) that systematically decomposes this learning burden through staged training, enabling superior performance on complex degradations. Our framework additionally incorporates an Interweaved Latent Transition (ILT) technique that maintains competitive temporal consistency without additional training overhead. Experiments demonstrate that our approach excels in scenarios where competing methods struggle, particularly on severely degraded videos. Our work reveals that addressing the learning strategy, rather than focusing solely on architectural complexity, is the critical path toward robust real-world video super-resolution with diffusion models.
Authors: Jinrong Cui, Xiaohuang Wu, Haitao Zhang, Chongjie Dong, Jie Wen
Abstract: Deep multi-view clustering seeks to utilize the abundant information from multiple views to improve clustering performance. However, most of the existing clustering methods often neglect to fully mine multi-view structural information and fail to explore the distribution of multi-view data, limiting clustering performance. To address these limitations, we propose a structure-guided deep multi-view clustering model. Specifically, we introduce a positive sample selection strategy based on neighborhood relationships, coupled with a corresponding loss function. This strategy constructs multi-view nearest neighbor graphs to dynamically redefine positive sample pairs, enabling the mining of local structural information within multi-view data and enhancing the reliability of positive sample selection. Additionally, we introduce a Gaussian distribution model to uncover latent structural information and introduce a loss function to reduce discrepancies between view embeddings. These two strategies explore multi-view structural information and data distribution from different perspectives, enhancing consistency across views and increasing intra-cluster compactness. Experimental evaluations demonstrate the efficacy of our method, showing significant improvements in clustering performance on multiple benchmark datasets compared to state-of-the-art multi-view clustering approaches.
Authors: Junyu Xia, Jiesong Bai, Yihang Dong
Abstract: Low-light image enhancement (LLE) aims to improve the visual quality of images captured in poorly lit conditions, which often suffer from low brightness, low contrast, noise, and color distortions. These issues hinder the performance of computer vision tasks such as object detection, facial recognition, and autonomous driving.Traditional enhancement techniques, such as multi-scale fusion and histogram equalization, fail to preserve fine details and often struggle with maintaining the natural appearance of enhanced images under complex lighting conditions. Although the Retinex theory provides a foundation for image decomposition, it often amplifies noise, leading to suboptimal image quality. In this paper, we propose the Dual Light Enhance Network (DLEN), a novel architecture that incorporates two distinct attention mechanisms, considering both spatial and frequency domains. Our model introduces a learnable wavelet transform module in the illumination estimation phase, preserving high- and low-frequency components to enhance edge and texture details. Additionally, we design a dual-branch structure that leverages the power of the Transformer architecture to enhance both the illumination and structural components of the image.Through extensive experiments, our model outperforms state-of-the-art methods on standard benchmarks.Code is available here: https://github.com/LaLaLoXX/DLEN
Authors: Hao Fang, Xiaohang Sui, Hongyao Yu, Kuofeng Gao, Jiawei Kong, Sijin Yu, Bin Chen, Hao Wu, Shu-Tao Xia
Abstract: Diffusion models (DMs) have recently demonstrated remarkable generation capability. However, their training generally requires huge computational resources and large-scale datasets. To solve these, recent studies empower DMs with the advanced Retrieval-Augmented Generation (RAG) technique and propose retrieval-augmented diffusion models (RDMs). By incorporating rich knowledge from an auxiliary database, RAG enhances diffusion models' generation and generalization ability while significantly reducing model parameters. Despite the great success, RAG may introduce novel security issues that warrant further investigation. In this paper, we reveal that the RDM is susceptible to backdoor attacks by proposing a multimodal contrastive attack approach named BadRDM. Our framework fully considers RAG's characteristics and is devised to manipulate the retrieved items for given text triggers, thereby further controlling the generated contents. Specifically, we first insert a tiny portion of images into the retrieval database as target toxicity surrogates. Subsequently, a malicious variant of contrastive learning is adopted to inject backdoors into the retriever, which builds shortcuts from triggers to the toxicity surrogates. Furthermore, we enhance the attacks through novel entropy-based selection and generative augmentation strategies that can derive better toxicity surrogates. Extensive experiments on two mainstream tasks demonstrate the proposed BadRDM achieves outstanding attack effects while preserving the model's benign utility.
Authors: Aman Urumbekov, Zheng Chen
Abstract: Transformers have become increasingly popular for image super-resolution (SR) tasks due to their strong global context modeling capabilities. However, their quadratic computational complexity necessitates the use of window-based attention mechanisms, which restricts the receptive field and limits effective context expansion. Recently, the Mamba architecture has emerged as a promising alternative with linear computational complexity, allowing it to avoid window mechanisms and maintain a large receptive field. Nevertheless, Mamba faces challenges in handling long-context dependencies when high pixel-level precision is required, as in SR tasks. This is due to its hidden state mechanism, which can compress and store a substantial amount of context but only in an approximate manner, leading to inaccuracies that transformers do not suffer from. In this paper, we propose \textbf{Contrast}, a hybrid SR model that combines \textbf{Con}volutional, \textbf{Tra}nsformer, and \textbf{St}ate Space components, effectively blending the strengths of transformers and Mamba to address their individual limitations. By integrating transformer and state space mechanisms, \textbf{Contrast} compensates for the shortcomings of each approach, enhancing both global context modeling and pixel-level accuracy. We demonstrate that combining these two architectures allows us to mitigate the problems inherent in each, resulting in improved performance on image super-resolution tasks.
Authors: Fu Rong, Meng Lan, Qian Zhang, Lefei Zhang
Abstract: Referring video object segmentation (RVOS) aims to segment objects in a video according to textual descriptions, which requires the integration of multimodal information and temporal dynamics perception. The Segment Anything Model 2 (SAM 2) has shown great effectiveness across various video segmentation tasks. However, its application to offline RVOS is challenged by the translation of the text into effective prompts and a lack of global context awareness. In this paper, we propose a novel RVOS framework, termed MPG-SAM 2, to address these challenges. Specifically, MPG-SAM 2 employs a unified multimodal encoder to jointly encode video and textual features, generating semantically aligned video and text embeddings, along with multimodal class tokens. A mask prior generator utilizes the video embeddings and class tokens to create pseudo masks of target objects and global context. These masks are fed into the prompt encoder as dense prompts along with multimodal class tokens as sparse prompts to generate accurate prompts for SAM 2. To provide the online SAM 2 with a global view, we introduce a hierarchical global-historical aggregator, which allows SAM 2 to aggregate global and historical information of target objects at both pixel and object levels, enhancing the target representation and temporal consistency. Extensive experiments on several RVOS benchmarks demonstrate the superiority of MPG-SAM 2 and the effectiveness of our proposed modules.
Authors: Siqi Wang, Yuanze Hu, Xinwang Liu, Siwei Wang, Guangpu Wang, Chuanfu Xu, Jie Liu, Ping Chen
Abstract: Industrial image anomaly detection (IAD) is a pivotal topic with huge value. Due to anomaly's nature, real anomalies in a specific modern industrial domain (i.e. domain-specific anomalies) are usually too rare to collect, which severely hinders IAD. Thus, zero-shot anomaly synthesis (ZSAS), which synthesizes pseudo anomaly images without any domain-specific anomaly, emerges as a vital technique for IAD. However, existing solutions are either unable to synthesize authentic pseudo anomalies, or require cumbersome training. Thus, we focus on ZSAS and propose a brand-new paradigm that can realize both authentic and training-free ZSAS. It is based on a chronically-ignored fact: Although domain-specific anomalies are rare, real anomalies from other domains (i.e. cross-domain anomalies) are actually abundant and directly applicable to ZSAS. Specifically, our new ZSAS paradigm makes three-fold contributions: First, we propose a novel method named Cross-domain Anomaly Injection (CAI), which directly exploits cross-domain anomalies to enable highly authentic ZSAS in a training-free manner. Second, to supply CAI with sufficient cross-domain anomalies, we build the first Domain-agnostic Anomaly Dataset within our best knowledge, which provides ZSAS with abundant real anomaly patterns. Third, we propose a CAI-guided Diffusion Mechanism, which further breaks the quantity limit of real anomalies and enable unlimited anomaly synthesis. Our head-to-head comparison with existing ZSAS solutions justifies our paradigm's superior performance for IAD and demonstrates it as an effective and pragmatic ZSAS solution.
Authors: Zhibo Ren, Pritthijit Nath, Pancham Shukla
Abstract: Tropical cyclone (TC) forecasting is crucial for disaster preparedness and mitigation. While recent deep learning approaches have shown promise, existing methods often treat TC evolution as a series of independent frame-to-frame predictions, limiting their ability to capture long-term dynamics. We present a novel application of video diffusion models for TC forecasting that explicitly models temporal dependencies through additional temporal layers. Our approach enables the model to generate multiple frames simultaneously, better capturing cyclone evolution patterns. We introduce a two-stage training strategy that significantly improves individual-frame quality and performance in low-data regimes. Experimental results show our method outperforms the previous approach of Nath et al. by 19.3% in MAE, 16.2% in PSNR, and 36.1% in SSIM. Most notably, we extend the reliable forecasting horizon from 36 to 50 hours. Through comprehensive evaluation using both traditional metrics and Fr\'echet Video Distance (FVD), we demonstrate that our approach produces more temporally coherent forecasts while maintaining competitive single-frame quality. Code accessible at https://github.com/Ren-creater/forecast-video-diffmodels.
URLs: https://github.com/Ren-creater/forecast-video-diffmodels.
Authors: Md Kaykobad Reza, Ameya Patil, Mashhour Solh, M. Salman Asif
Abstract: Multimodal models often experience a significant performance drop when one or more modalities are missing during inference. To address this challenge, we propose a simple yet effective approach that enhances robustness to missing modalities while maintaining strong performance when all modalities are available. Our method introduces cross-modal proxy tokens (CMPTs), which approximate the class token of a missing modality by attending only to the tokens of the available modality. To efficiently learn the approximation for the missing modality via CMPTs with minimal computational overhead, we employ low-rank adapters in frozen unimodal encoders and jointly optimize an alignment loss with a task-specific loss. Extensive experiments on five multimodal datasets show that our method outperforms state-of-the-art baselines across various missing rates while achieving competitive results in complete-modality settings. Overall, our method offers a flexible and efficient solution for robust multimodal learning. The code and pretrained models will be released on GitHub.
Authors: Yicheng Wu, Tao Song, Zhonghua Wu, Jin Ye, Zongyuan Ge, Zhaolin Chen, Jianfei Cai
Abstract: Unified MRI imputation, which can adapt to diverse imputation scenarios, is highly desirable as it reduces scanning costs and provides comprehensive MRI information for improved clinical diagnosis. Existing unified MRI imputation methods either rely on specific prompts to guide their transformation network or require multiple modality-specific modules. However, these approaches struggle to capture large modality and instance variations or become too complex to generalize effectively. To address these limitations, we propose CodeBrain, a fundamentally different pipeline for unified brain MRI imputation. Our key idea is to reframe various inter-modality transformations as a full-modality code prediction task via a two-stage framework. In the first stage, CodeBrain reconstructs a target modality from any other modalities by learning a compact scalar-quantized code for each instance and modality. Any target modality can then be reconstructed with high fidelity by combining the corresponding code with shared features extracted from any available modality. In the second stage, a projection encoder is trained to predict full-modality compact codes from any incomplete MRI samples, effectively simulating various imputation scenarios. We evaluate our CodeBrain on two public brain MRI datasets (i.e., IXI and BraTS 2023). Extensive experiments demonstrate that CodeBrain outperforms state-of-the-art methods, setting a new benchmark for unified brain MRI imputation. Our code will be released.
Authors: Pinxin Liu, Luchuan Song, Junhua Huang, Haiyang Liu, Chenliang Xu
Abstract: Generating full-body human gestures based on speech signals remains challenges on quality and speed. Existing approaches model different body regions such as body, legs and hands separately, which fail to capture the spatial interactions between them and result in unnatural and disjointed movements. Additionally, their autoregressive/diffusion-based pipelines show slow generation speed due to dozens of inference steps. To address these two challenges, we propose GestureLSM, a flow-matching-based approach for Co-Speech Gesture Generation with spatial-temporal modeling. Our method i) explicitly model the interaction of tokenized body regions through spatial and temporal attention, for generating coherent full-body gestures. ii) introduce the flow matching to enable more efficient sampling by explicitly modeling the latent velocity space. To overcome the suboptimal performance of flow matching baseline, we propose latent shortcut learning and beta distribution time stamp sampling during training to enhance gesture synthesis quality and accelerate inference. Combining the spatial-temporal modeling and improved flow matching-based framework, GestureLSM achieves state-of-the-art performance on BEAT2 while significantly reducing inference time compared to existing methods, highlighting its potential for enhancing digital humans and embodied agents in real-world applications. Project Page: https://andypinxinliu.github.io/GestureLSM
Authors: Lei Jiang, Ye Wei, Hao Ni
Abstract: Diffusion models have become a popular choice for human motion synthesis due to their powerful generative capabilities. However, their high computational complexity and large sampling steps pose challenges for real-time applications. Fortunately, the Consistency Model (CM) provides a solution to greatly reduce the number of sampling steps from hundreds to a few, typically fewer than four, significantly accelerating the synthesis of diffusion models. However, applying CM to text-conditioned human motion synthesis in latent space yields unsatisfactory generation results. In this paper, we introduce \textbf{MotionPCM}, a phased consistency model-based approach designed to improve the quality and efficiency for real-time motion synthesis in latent space. Experimental results on the HumanML3D dataset show that our model achieves real-time inference at over 30 frames per second in a single sampling step while outperforming the previous state-of-the-art with a 38.9\% improvement in FID. The code will be available for reproduction.
Authors: Mian Muhammad Naeem Abid, Nancy Mehta, Zongwei Wu, Radu Timofte
Abstract: Semantic segmentation assigns labels to pixels in images, a critical yet challenging task in computer vision. Convolutional methods, although capturing local dependencies well, struggle with long-range relationships. Vision Transformers (ViTs) excel in global context capture but are hindered by high computational demands, especially for high-resolution inputs. Most research optimizes the encoder architecture, leaving the bottleneck underexplored - a key area for enhancing performance and efficiency. We propose ContextFormer, a hybrid framework leveraging the strengths of CNNs and ViTs in the bottleneck to balance efficiency, accuracy, and robustness for real-time semantic segmentation. The framework's efficiency is driven by three synergistic modules: the Token Pyramid Extraction Module (TPEM) for hierarchical multi-scale representation, the Transformer and Branched DepthwiseConv (Trans-BDC) block for dynamic scale-aware feature modeling, and the Feature Merging Module (FMM) for robust integration with enhanced spatial and contextual consistency. Extensive experiments on ADE20K, Pascal Context, CityScapes, and COCO-Stuff datasets show ContextFormer significantly outperforms existing models, achieving state-of-the-art mIoU scores, setting a new benchmark for efficiency and performance. The codes will be made publicly available upon acceptance.
Authors: Zhixi Cai, Fucai Ke, Simindokht Jahangard, Maria Garcia de la Banda, Reza Haffari, Peter J. Stuckey, Hamid Rezatofighi
Abstract: Visual Grounding (VG) tasks, such as referring expression detection and segmentation tasks are important for linking visual entities to context, especially in complex reasoning tasks that require detailed query interpretation. This paper explores VG beyond basic perception, highlighting challenges for methods that require reasoning like human cognition. Recent advances in large language methods (LLMs) and Vision-Language methods (VLMs) have improved abilities for visual comprehension, contextual understanding, and reasoning. These methods are mainly split into end-to-end and compositional methods, with the latter offering more flexibility. Compositional approaches that integrate LLMs and foundation models show promising performance but still struggle with complex reasoning with language-based logical representations. To address these limitations, we propose NAVER, a compositional visual grounding method that integrates explicit probabilistic logic reasoning within a finite-state automaton, equipped with a self-correcting mechanism. This design improves robustness and interpretability in inference through explicit logic reasoning. Our results show that NAVER achieves SoTA performance comparing to recent end-to-end and compositional baselines. The code is available at https://github.com/ControlNet/NAVER .
Authors: Zhiteng Li, Mingyuan Xia, Jingyuan Zhang, Zheng Hui, Linghe Kong, Yulun Zhang, Xiaokang Yang
Abstract: Large language models (LLMs) have achieved remarkable success in natural language processing (NLP) tasks, yet their substantial memory requirements present significant challenges for deployment on resource-constrained devices. Singular Value Decomposition (SVD) has emerged as a promising compression technique for LLMs, offering considerable reductions in memory overhead. However, existing SVD-based methods often struggle to effectively mitigate the errors introduced by SVD truncation, leading to a noticeable performance gap when compared to the original models. Furthermore, applying a uniform compression ratio across all transformer layers fails to account for the varying importance of different layers. To address these challenges, we propose AdaSVD, an adaptive SVD-based LLM compression approach. Specifically, AdaSVD introduces adaComp, which adaptively compensates for SVD truncation errors by alternately updating the singular matrices $\mathcal{U}$ and $\mathcal{V}^\top$. Additionally, AdaSVD introduces adaCR, which adaptively assigns layer-specific compression ratios based on the relative importance of each layer. Extensive experiments across multiple LLM/VLM families and evaluation metrics demonstrate that AdaSVD consistently outperforms state-of-the-art (SOTA) SVD-based methods, achieving superior performance with significantly reduced memory requirements. Code and models of AdaSVD will be available at https://github.com/ZHITENGLI/AdaSVD.
Authors: Xiaomeng Yang, Zhiyu Tan, Hao Li
Abstract: Video foundation models have achieved significant advancement with the help of network upgrade as well as model scale-up. However, they are still hard to meet requirements of applications due to unsatisfied generation quality. To solve this problem, we propose to align video foundation models with human preferences from the perspective of post-training in this paper. Consequently, we introduce an Iterative Preference Optimization strategy to enhance generated video quality by incorporating human feedback. Specifically, IPO exploits a critic model to justify video generations for pairwise ranking as in Direct Preference Optimization or point-wise scoring as in Kahneman-Tversky Optimization. Given this, IPO optimizes video foundation models with guidance of signals from preference feedback, which helps improve generated video quality in subject consistency, motion smoothness and aesthetic quality, etc. In addition, IPO incorporates the critic model with the multi-modality large language model, which enables it to automatically assign preference labels without need of retraining or relabeling. In this way, IPO can efficiently perform multi-round preference optimization in an iterative manner, without the need of tediously manual labeling. Comprehensive experiments demonstrate that the proposed IPO can effectively improve the video generation quality of a pretrained model and help a model with only 2B parameters surpass the one with 5B parameters. Besides, IPO achieves new state-of-the-art performance on VBench benchmark.
Authors: Hongyu Ge, Longkun Hao, Zihui Xu, Zhenxin Lin, Bin Li, Shoujun Zhou, Hongjin Zhao, Yihang Liu
Abstract: Medical Visual Question Answering (Med-VQA) represents a critical and challenging subtask within the general VQA domain. Despite significant advancements in general Visual Question Answering (VQA), multimodal large language models (MLLMs) still exhibit substantial limitations when handling multi-task VQA scenarios. These limitations manifest through erroneous spatial localization and misinterpretation of medical images, which primarily arise from two fundamental issues: inadequate image-text alignment and insufficient medical knowledge in general-purpose MLLMs for specialized medical applications. To address these issues, we introduce the Cross-Modal Clinical Knowledge Distiller (ClinKD), an innovative framework designed to enhance image-text alignment and establish more effective medical knowledge adaptation mechanisms, which enables MLLMs to adapt to medical knowledge. Our extensive experimental evaluations demonstrate that the ClinKD achieves state-of-the-art performance on the Med-GRIT-270k dataset, a challenging medical benchmark containing fine-grained multi-task QA pairs. The results indicate that our approach not only significantly improves image-text alignment but also effectively enables MLLMs to adapt to the medical knowledge. The source code for ClinKD is available at: https://github.com/overloadedHenry/ClinKD.
Authors: Xinyu Liu, Ailing Zeng, Wei Xue, Harry Yang, Wenhan Luo, Qifeng Liu, Yike Guo
Abstract: Crafting magic and illusions is one of the most thrilling aspects of filmmaking, with visual effects (VFX) serving as the powerhouse behind unforgettable cinematic experiences. While recent advances in generative artificial intelligence have driven progress in generic image and video synthesis, the domain of controllable VFX generation remains relatively underexplored. In this work, we propose a novel paradigm for animated VFX generation as image animation, where dynamic effects are generated from user-friendly textual descriptions and static reference images. Our work makes two primary contributions: (i) Open-VFX, the first high-quality VFX video dataset spanning 15 diverse effect categories, annotated with textual descriptions, instance segmentation masks for spatial conditioning, and start-end timestamps for temporal control. (ii) VFX Creator, a simple yet effective controllable VFX generation framework based on a Video Diffusion Transformer. The model incorporates a spatial and temporal controllable LoRA adapter, requiring minimal training videos. Specifically, a plug-and-play mask control module enables instance-level spatial manipulation, while tokenized start-end motion timestamps embedded in the diffusion process, alongside the text encoder, allow precise temporal control over effect timing and pace. Extensive experiments on the Open-VFX test set demonstrate the superiority of the proposed system in generating realistic and dynamic effects, achieving state-of-the-art performance and generalization ability in both spatial and temporal controllability. Furthermore, we introduce a specialized metric to evaluate the precision of temporal control. By bridging traditional VFX techniques with generative approaches, VFX Creator unlocks new possibilities for efficient and high-quality video effect generation, making advanced VFX accessible to a broader audience.
Authors: Laura Dodds, Tara Boroushaki, Cusuh Ham, Fadel Adib
Abstract: The ability to observe the world is fundamental to reasoning and making informed decisions on how to interact with the environment. However, optical perception can often be disrupted due to common occurrences, such as occlusions, which can pose challenges to existing vision systems. We present MITO, the first millimeter-wave (mmWave) dataset of diverse, everyday objects, collected using a UR5 robotic arm with two mmWave radars operating at different frequencies and an RGB-D camera. Unlike visible light, mmWave signals can penetrate common occlusions (e.g., cardboard boxes, fabric, plastic) but each mmWave frame has much lower resolution than typical cameras. To capture higher-resolution mmWave images, we leverage the robot's mobility and fuse frames over the synthesized aperture. MITO captures over 24 million mmWave frames and uses them to generate 550 high-resolution mmWave (synthetic aperture) images in line-of-sight and non-light-of-sight (NLOS), as well as RGB-D images, segmentation masks, and raw mmWave signals, taken from 76 different objects. We develop an open-source simulation tool that can be used to generate synthetic mmWave images for any 3D triangle mesh. Finally, we demonstrate the utility of our dataset and simulator for enabling broader NLOS perception by developing benchmarks for NLOS segmentation and classification.
Authors: Huaying Yuan, Jian Ni, Yueze Wang, Junjie Zhou, Zhengyang Liang, Zheng Liu, Zhao Cao, Zhicheng Dou, Ji-Rong Wen
Abstract: Retrieval augmented generation (RAG) holds great promise in addressing challenges associated with long video understanding. These methods retrieve useful moments from long videos for their presented tasks, thereby enabling multimodal large language models (MLLMs) to generate high-quality answers in a cost-effective way. In this work, we present MomentSeeker, a comprehensive benchmark to evaluate retrieval models' performance in handling general long-video moment retrieval (LVMR) tasks. MomentSeeker offers three key advantages. First, it incorporates long videos of over 500 seconds on average, making it the first benchmark specialized for long-video moment retrieval. Second, it covers a wide range of task categories (including Moment Search, Caption Alignment, Image-conditioned Moment Search, and Video-conditioned Moment Search) and diverse application scenarios (e.g., sports, movies, cartoons, and ego), making it a comprehensive tool for assessing retrieval models' general LVMR performance. Additionally, the evaluation tasks are carefully curated through human annotation, ensuring the reliability of assessment. We further fine-tune an MLLM-based LVMR retriever on synthetic data, which demonstrates strong performance on our benchmark. We perform extensive experiments with various popular multimodal retrievers based on our benchmark, whose results highlight the challenges of LVMR and limitations for existing methods. Our created resources will be shared with community to advance future research in this field.
Authors: Fengcheng Yu, Haoran Xu, Canming Xia, Ziyang Zong, Guang Tan
Abstract: Vision-based occupancy networks (VONs) provide an end-to-end solution for reconstructing 3D environments in autonomous driving. However, existing methods often suffer from temporal inconsistencies, manifesting as flickering effects that compromise visual experience and adversely affect decision-making. While recent approaches have incorporated historical data to mitigate the issue, they often incur high computational costs and introduce noisy information that interferes with object detection. We propose OccLinker, a novel plugin framework designed to seamlessly integrate with existing VONs for boosting performance. Our method employs a three-stage architecture that consolidates historical static and motion cues, correlates them with current features through a Motion-Static Integration (MSI) mechanism, and generates correction occupancy to refine base network predictions. Extensive experiments on two benchmarks demonstrate the efficiency and effectiveness of our method, outperforming the latest baseline models. The source code are available in the supplementary material.
Authors: Tuan-Anh Yang, Truong-Son Hy, Phuong D. Dao
Abstract: This paper introduces a novel multiscale object-based graph neural network called MOB-GCN for hyperspectral image (HSI) classification. The central aim of this study is to enhance feature extraction and classification performance by utilizing multiscale object-based image analysis (OBIA). Traditional pixel-based methods often suffer from low accuracy and speckle noise, while single-scale OBIA approaches may overlook crucial information of image objects at different levels of detail. MOB-GCN addresses this issue by extracting and integrating features from multiple segmentation scales to improve classification results using the Multiresolution Graph Network (MGN) architecture that can model fine-grained and global spatial patterns. By constructing a dynamic multiscale graph hierarchy, MOB-GCN offers a more comprehensive understanding of the intricate details and global context of HSIs. Experimental results demonstrate that MOB-GCN consistently outperforms single-scale graph convolutional networks (GCNs) in terms of classification accuracy, computational efficiency, and noise reduction, particularly when labeled data is limited. The implementation of MOB-GCN is publicly available at https://github.com/HySonLab/MultiscaleHSI
Authors: Zheling Meng, Bo Peng, Xiaochuan Jin, Yueming Lyu, Wei Wang, Jing Dong
Abstract: Text-to-image diffusion models have demonstrated the underlying risk of generating various unwanted content, such as sexual elements. To address this issue, the task of concept erasure has been introduced, aiming to erase any undesired concepts that the models can generate. Previous methods, whether training-based or training-free, have primarily focused on the input side, i.e., texts. However, they often suffer from incomplete erasure due to limitations in the generalization from limited prompts to diverse image content. In this paper, motivated by the notion that concept erasure on the output side, i.e., generated images, may be more direct and effective, we propose Concept Corrector. It checks target concepts based on visual features provided by final generated images predicted at certain time steps. Further, it incorporates Concept Removal Attention to erase generated concept features. It overcomes the limitations of existing methods, which are either unable to remove the concept features that have been generated in images or rely on the assumption that the related concept words are contained in input prompts. In the whole pipeline, our method changes no model parameters and only requires a given target concept as well as the corresponding replacement content, which is easy to implement. To the best of our knowledge, this is the first erasure method based on intermediate-generated images, achieving the ability to erase concepts on the fly. The experiments on various concepts demonstrate its impressive erasure performance.
Authors: Yunpeng Gao, Chenhui Li, Zhongrui You, Junli Liu, Zhen Li, Pengan Chen, Qizhi Chen, Zhonghan Tang, Liansheng Wang, Penghui Yang, Yiwen Tang, Yuhang Tang, Shuai Liang, Songyi Zhu, Ziqin Xiong, Yifei Su, Xinyi Ye, Jianan Li, Yan Ding, Dong Wang, Zhigang Wang, Bin Zhao, Xuelong Li
Abstract: Vision-Language Navigation (VLN) aims to guide agents through an environment by leveraging both language instructions and visual cues, playing a pivotal role in embodied AI. Indoor VLN has been extensively studied, whereas outdoor aerial VLN remains underexplored. The potential reason is that outdoor aerial view encompasses vast areas, making data collection more challenging, which results in a lack of benchmarks. To address this problem, we propose OpenFly, a platform comprising a versatile toolchain and large-scale benchmark for aerial VLN. Firstly, we develop a highly automated toolchain for data collection, enabling automatic point cloud acquisition, scene semantic segmentation, flight trajectory creation, and instruction generation. Secondly, based on the toolchain, we construct a large-scale aerial VLN dataset with 100k trajectories, covering diverse heights and lengths across 18 scenes. The corresponding visual data are generated using various rendering engines and advanced techniques, including Unreal Engine, GTA V, Google Earth, and 3D Gaussian Splatting (3D GS). All data exhibit high visual quality. Particularly, 3D GS supports real-to-sim rendering, further enhancing the realism of the dataset. Thirdly, we propose OpenFly-Agent, a keyframe-aware VLN model, which takes language instructions, current observations, and historical keyframes as input, and outputs flight actions directly. Extensive analyses and experiments are conducted, showcasing the superiority of our OpenFly platform and OpenFly-Agent. The toolchain, dataset, and codes will be open-sourced.
Authors: Cao Yuxuan, Wu Jiayang, Alistair Cheong Liang Chuen, Bryan Shan Guanrong, Theodore Lee Chong Jen, Sherman Chann Zhi Shen
Abstract: Traditional online content moderation systems struggle to classify modern multimodal means of communication, such as memes, a highly nuanced and information-dense medium. This task is especially hard in a culturally diverse society like Singapore, where low-resource languages are used and extensive knowledge on local context is needed to interpret online content. We curate a large collection of 112K memes labeled by GPT-4V for fine-tuning a VLM to classify offensive memes in Singapore context. We show the effectiveness of fine-tuned VLMs on our dataset, and propose a pipeline containing OCR, translation and a 7-billion parameter-class VLM. Our solutions reach 80.62% accuracy and 0.8192 AUROC on a held-out test set, and can greatly aid human in moderating online contents. The dataset, code, and model weights have been open-sourced at https://github.com/aliencaocao/vlm-for-memes-aisg.
Authors: Ayushi Dutta, Marco Pesavento, Marco Volino, Adrian Hilton, Armin Mustafa
Abstract: Recent approaches to jointly reconstruct 3D humans and objects from a single RGB image represent 3D shapes with template-based or coarse models, which fail to capture details of loose clothing on human bodies. In this paper, we introduce a novel implicit approach for jointly reconstructing realistic 3D clothed humans and objects from a monocular view. For the first time, we model both the human and the object with an implicit representation, allowing to capture more realistic details such as clothing. This task is extremely challenging due to human-object occlusions and the lack of 3D information in 2D images, often leading to poor detail reconstruction and depth ambiguity. To address these problems, we propose a novel attention-based neural implicit model that leverages image pixel alignment from both the input human-object image for a global understanding of the human-object scene and from local separate views of the human and object images to improve realism with, for example, clothing details. Additionally, the network is conditioned on semantic features derived from an estimated human-object pose prior, which provides 3D spatial information about the shared space of humans and objects. To handle human occlusion caused by objects, we use a generative diffusion model that inpaints the occluded regions, recovering otherwise lost details. For training and evaluation, we introduce a synthetic dataset featuring rendered scenes of inter-occluded 3D human scans and diverse objects. Extensive evaluation on both synthetic and real-world datasets demonstrates the superior quality of the proposed human-object reconstructions over competitive methods.
Authors: Huiqiang Wang, Mingchen Song, Guoqiang Zhong
Abstract: Currently, restoring clean images from a variety of degradation types using a single model is still a challenging task. Existing all-in-one image restoration approaches struggle with addressing complex and ambiguously defined degradation types. In this paper, we introduce a dynamic degradation decomposition network for all-in-one image restoration, named D$^3$Net. D$^3$Net achieves degradation-adaptive image restoration with guided prompt through cross-domain interaction and dynamic degradation decomposition. Concretely, in D$^3$Net, the proposed Cross-Domain Degradation Analyzer (CDDA) engages in deep interaction between frequency domain degradation characteristics and spatial domain image features to identify and model variations of different degradation types on the image manifold, generating degradation correction prompt and strategy prompt, which guide the following decomposition process. Furthermore, the prompt-based Dynamic Decomposition Mechanism (DDM) for progressive degradation decomposition, that encourages the network to adaptively select restoration strategies utilizing the two-level prompt generated by CDDA. Thanks to the synergistic cooperation between CDDA and DDM, D$^3$Net achieves superior flexibility and scalability in handling unknown degradation, while effectively reducing unnecessary computational overhead. Extensive experiments on multiple image restoration tasks demonstrate that D$^3$Net significantly outperforms the state-of-the-art approaches, especially improving PSNR by 5.47dB and 3.30dB on the SOTS-Outdoor and GoPro datasets, respectively.
Authors: Chunyang Cheng, Tianyang Xu, Zhenhua Feng, Xiaojun Wu, ZhangyongTang, Hui Li, Zeyang Zhang, Sara Atito, Muhammad Awais, Josef Kittler
Abstract: Advanced image fusion methods mostly prioritise high-level missions, where task interaction struggles with semantic gaps, requiring complex bridging mechanisms. In contrast, we propose to leverage low-level vision tasks from digital photography fusion, allowing for effective feature interaction through pixel-level supervision. This new paradigm provides strong guidance for unsupervised multimodal fusion without relying on abstract semantics, enhancing task-shared feature learning for broader applicability. Owning to the hybrid image features and enhanced universal representations, the proposed GIFNet supports diverse fusion tasks, achieving high performance across both seen and unseen scenarios with a single model. Uniquely, experimental results reveal that our framework also supports single-modality enhancement, offering superior flexibility for practical applications. Our code will be available at https://github.com/AWCXV/GIFNet.
Authors: Chao Wang, Luning Zhang, Zheng Wang, Yang Zhou
Abstract: Combining multiple perceptual inputs and performing combinatorial reasoning in complex scenarios is a sophisticated cognitive function in humans. With advancements in multi-modal large language models, recent benchmarks tend to evaluate visual understanding across multiple images. However, they often overlook the necessity of combinatorial reasoning across multiple perceptual information. To explore the ability of advanced models to integrate multiple perceptual inputs for combinatorial reasoning in complex scenarios, we introduce two benchmarks: Clue-Visual Question Answering (CVQA), with three task types to assess visual comprehension and synthesis, and Clue of Password-Visual Question Answering (CPVQA), with two task types focused on accurate interpretation and application of visual data. For our benchmarks, we present three plug-and-play approaches: utilizing model input for reasoning, enhancing reasoning through minimum margin decoding with randomness generation, and retrieving semantically relevant visual information for effective data integration. The combined results reveal current models' poor performance on combinatorial reasoning benchmarks, even the state-of-the-art (SOTA) closed-source model achieves only 33.04% accuracy on CVQA, and drops to 7.38% on CPVQA. Notably, our approach improves the performance of models on combinatorial reasoning, with a 22.17% boost on CVQA and 9.40% on CPVQA over the SOTA closed-source model, demonstrating its effectiveness in enhancing combinatorial reasoning with multiple perceptual inputs in complex scenarios. The code will be publicly available.
Authors: Sherlon Almeida da Silva, Davi Geiger, Luiz Velho, Moacir Antonelli Ponti
Abstract: We innovate in stereo vision by explicitly providing analytical 3D surface models as viewed by a cyclopean eye model that incorporate depth discontinuities and occlusions. This geometrical foundation combined with learned stereo features allows our system to benefit from the strengths of both approaches. We also invoke a prior monocular model of surfaces to fill in occlusion regions or texture-less regions where data matching is not sufficient. Our results already are on par with the state-of-the-art purely data-driven methods and are of much better visual quality, emphasizing the importance of the 3D geometrical model to capture critical visual information. Such qualitative improvements may find applicability in virtual reality, for a better human experience, as well as in robotics, for reducing critical errors. Our approach aims to demonstrate that understanding and modeling geometrical properties of 3D surfaces is beneficial to computer vision research.
Authors: Haoxin Li, Yingchen Yu, Qilong Wu, Hanwang Zhang, Song Bai, Boyang Li
Abstract: Despite recent progress, video generative models still struggle to animate static images into videos that portray delicate human actions, particularly when handling uncommon or novel actions whose training data are limited. In this paper, we explore the task of learning to animate images to portray delicate human actions using a small number of videos -- 16 or fewer -- which is highly valuable for real-world applications like video and movie production. Learning generalizable motion patterns that smoothly transition from user-provided reference images in a few-shot setting is highly challenging. We propose FLASH (Few-shot Learning to Animate and Steer Humans), which learns generalizable motion patterns by forcing the model to reconstruct a video using the motion features and cross-frame correspondences of another video with the same motion but different appearance. This encourages transferable motion learning and mitigates overfitting to limited training data. Additionally, FLASH extends the decoder with additional layers to propagate details from the reference image to generated frames, improving transition smoothness. Human judges overwhelmingly favor FLASH, with 65.78\% of 488 responses prefer FLASH over baselines. We strongly recommend watching the videos in the website: https://lihaoxin05.github.io/human_action_animation/, as motion artifacts are hard to notice from images.
Authors: Hui Liu, Chen Jia, Fan Shi, Xu Cheng, Shengyong Chen
Abstract: Pixel-level segmentation of structural cracks across various scenarios remains a considerable challenge. Current methods encounter challenges in effectively modeling crack morphology and texture, facing challenges in balancing segmentation quality with low computational resource usage. To overcome these limitations, we propose a lightweight Structure-Aware Vision Mamba Network (SCSegamba), capable of generating high-quality pixel-level segmentation maps by leveraging both the morphological information and texture cues of crack pixels with minimal computational cost. Specifically, we developed a Structure-Aware Visual State Space module (SAVSS), which incorporates a lightweight Gated Bottleneck Convolution (GBC) and a Structure-Aware Scanning Strategy (SASS). The key insight of GBC lies in its effectiveness in modeling the morphological information of cracks, while the SASS enhances the perception of crack topology and texture by strengthening the continuity of semantic information between crack pixels. Experiments on crack benchmark datasets demonstrate that our method outperforms other state-of-the-art (SOTA) methods, achieving the highest performance with only 2.8M parameters. On the multi-scenario dataset, our method reached 0.8390 in F1 score and 0.8479 in mIoU. The code is available at https://github.com/Karl1109/SCSegamba.
Authors: Zhipeng Huang, Shaobin Zhuang, Canmiao Fu, Binxin Yang, Ying Zhang, Chong Sun, Zhizheng Zhang, Yali Wang, Chen Li, Zheng-Jun Zha
Abstract: Existing multimodal generative models fall short as qualified design copilots, as they often struggle to generate imaginative outputs once instructions are less detailed or lack the ability to maintain consistency with the provided references. In this work, we introduce WeGen, a model that unifies multimodal generation and understanding, and promotes their interplay in iterative generation. It can generate diverse results with high creativity for less detailed instructions. And it can progressively refine prior generation results or integrating specific contents from references following the instructions in its chat with users. During this process, it is capable of preserving consistency in the parts that the user is already satisfied with. To this end, we curate a large-scale dataset, extracted from Internet videos, containing rich object dynamics and auto-labeled dynamics descriptions by advanced foundation models to date. These two information are interleaved into a single sequence to enable WeGen to learn consistency-aware generation where the specified dynamics are generated while the consistency of unspecified content is preserved aligned with instructions. Besides, we introduce a prompt self-rewriting mechanism to enhance generation diversity. Extensive experiments demonstrate the effectiveness of unifying multimodal understanding and generation in WeGen and show it achieves state-of-the-art performance across various visual generation benchmarks. These also demonstrate the potential of WeGen as a user-friendly design copilot as desired. The code and models will be available at https://github.com/hzphzp/WeGen.
Authors: Dengke Zhang, Quan Tang, Fagui Liu, Haiqing Mei, C. L. Philip Chen
Abstract: Semi-supervised semantic segmentation has witnessed remarkable advancements in recent years. However, existing algorithms are based on convolutional neural networks and directly applying them to Vision Transformers poses certain limitations due to conceptual disparities. To this end, we propose TokenMix, a data augmentation technique specifically designed for semi-supervised semantic segmentation with Vision Transformers. TokenMix aligns well with the global attention mechanism by mixing images at the token level, enhancing learning capability for contextual information among image patches. We further incorporate image augmentation and feature augmentation to promote the diversity of augmentation. Moreover, to enhance consistency regularization, we propose a dual-branch framework where each branch applies image and feature augmentation to the input image. We conduct extensive experiments across multiple benchmark datasets, including Pascal VOC 2012, Cityscapes, and COCO. Results suggest that the proposed method outperforms state-of-the-art algorithms with notably observed accuracy improvement, especially under limited fine annotations.
Authors: Biqiao Xin, Qianchen Mao, Bingshu Wang, Jiangbin Zheng, Yong Zhao, C. L. Philip Chen
Abstract: The widespread deployment of Infrared Small-Target Detection (IRSTD) algorithms on edge devices necessitates the exploration of model compression techniques. Binarized neural networks (BNNs) are distinguished by their exceptional efficiency in model compression. However, the small size of infrared targets introduces stringent precision requirements for the IRSTD task, while the inherent precision loss during binarization presents a significant challenge. To address this, we propose the Binarized Infrared Small-Target Detection Network (BiisNet), which preserves the core operations of binarized convolutions while integrating full-precision features into the network's information flow. Specifically, we propose the Dot Binary Convolution, which retains fine-grained semantic information in feature maps while still leveraging the binarized convolution operations. In addition, we introduce a smooth and adaptive Dynamic Softsign function, which provides more comprehensive and progressively finer gradient during backpropagation, enhancing model stability and promoting an optimal weight distribution. Experimental results demonstrate that BiisNet not only significantly outperforms other binary architectures but also has strong competitiveness among state-of-the-art full-precision models.
Authors: Nianzu Yang, Pandeng Li, Liming Zhao, Yang Li, Chen-Wei Xie, Yehui Tang, Xudong Lu, Zhihang Liu, Yun Zheng, Yu Liu, Junchi Yan
Abstract: Existing video tokenizers typically use the traditional Variational Autoencoder (VAE) architecture for video compression and reconstruction. However, to achieve good performance, its training process often relies on complex multi-stage training tricks that go beyond basic reconstruction loss and KL regularization. Among these tricks, the most challenging is the precise tuning of adversarial training with additional Generative Adversarial Networks (GANs) in the final stage, which can hinder stable convergence. In contrast to GANs, diffusion models offer more stable training processes and can generate higher-quality results. Inspired by these advantages, we propose CDT, a novel Conditioned Diffusion-based video Tokenizer, that replaces the GAN-based decoder with a conditional causal diffusion model. The encoder compresses spatio-temporal information into compact latents, while the decoder reconstructs videos through a reverse diffusion process conditioned on these latents. During inference, we incorporate a feature cache mechanism to generate videos of arbitrary length while maintaining temporal continuity and adopt sampling acceleration technique to enhance efficiency. Trained using only a basic MSE diffusion loss for reconstruction, along with KL term and LPIPS perceptual loss from scratch, extensive experiments demonstrate that CDT achieves state-of-the-art performance in video reconstruction tasks with just a single-step sampling. Even a scaled-down version of CDT (3$\times$ inference speedup) still performs comparably with top baselines. Moreover, the latent video generation model trained with CDT also exhibits superior performance. The source code and pretrained weights will be released shortly, so please stay tuned for updates!
Authors: Feng Ni, Kui Huang, Yao Lu, Wenyu Lv, Guanzhong Wang, Zeyu Chen, Yi Liu
Abstract: With the rapid advancement of digitalization, various document images are being applied more extensively in production and daily life, and there is an increasingly urgent need for fast and accurate parsing of the content in document images. Therefore, this report presents PP-DocBee, a novel multimodal large language model designed for end-to-end document image understanding. First, we develop a data synthesis strategy tailored to document scenarios in which we build a diverse dataset to improve the model generalization. Then, we apply a few training techniques, including dynamic proportional sampling, data preprocessing, and OCR postprocessing strategies. Extensive evaluations demonstrate the superior performance of PP-DocBee, achieving state-of-the-art results on English document understanding benchmarks and even outperforming existing open source and commercial models in Chinese document understanding. The source code and pre-trained models are publicly available at \href{https://github.com/PaddlePaddle/PaddleMIX}{https://github.com/PaddlePaddle/PaddleMIX}.
URLs: https://github.com/PaddlePaddle/PaddleMIX, https://github.com/PaddlePaddle/PaddleMIX
Authors: Yu-Hsi Chen, Chin-Tien Wu
Abstract: Optical flow is a fundamental technique for motion estimation, widely applied in video stabilization, interpolation, and object tracking. Traditional optical flow estimation methods rely on restrictive assumptions like brightness constancy and slow motion constraints. Recent deep learning-based flow estimations require extensive training on large domain-specific datasets, making them computationally demanding. Also, artificial intelligence (AI) advances have enabled deep learning models to take advantage of optical flow as an important feature for object tracking and motion analysis. Since optical flow is commonly encoded in HSV for visualization, its conversion to RGB for neural network processing is nonlinear and may introduce perceptual distortions. These transformations amplify the sensitivity to estimation errors, potentially affecting the predictive accuracy of the networks. To address these challenges that are influential to the performance of downstream network models, we propose Reynolds flow, a novel training-free flow estimation inspired by the Reynolds transport theorem, offering a principled approach to modeling complex motion dynamics. In addition to conventional HSV-based visualization of Reynolds flow, we also introduce an RGB-encoded representation of Reynolds flow designed to improve flow visualization and feature enhancement for neural networks. We evaluated the effectiveness of Reynolds flow in video-based tasks. Experimental results on three benchmarks, tiny object detection on UAVDB, infrared object detection on Anti-UAV, and pose estimation on GolfDB, demonstrate that networks trained with RGB-encoded Reynolds flow achieve SOTA performance, exhibiting improved robustness and efficiency across all tasks.
Authors: Aoxiong Yin, Kai Shen, Yichong Leng, Xu Tan, Xinyu Zhou, Juncheng Li, Siliang Tang
Abstract: Recent advancements in text-to-video (T2V) generation have been driven by two competing paradigms: autoregressive language models and diffusion models. However, each paradigm has intrinsic limitations: language models struggle with visual quality and error accumulation, while diffusion models lack semantic understanding and causal modeling. In this work, we propose LanDiff, a hybrid framework that synergizes the strengths of both paradigms through coarse-to-fine generation. Our architecture introduces three key innovations: (1) a semantic tokenizer that compresses 3D visual features into compact 1D discrete representations through efficient semantic compression, achieving a $\sim$14,000$\times$ compression ratio; (2) a language model that generates semantic tokens with high-level semantic relationships; (3) a streaming diffusion model that refines coarse semantics into high-fidelity videos. Experiments show that LanDiff, a 5B model, achieves a score of 85.43 on the VBench T2V benchmark, surpassing the state-of-the-art open-source models Hunyuan Video (13B) and other commercial models such as Sora, Kling, and Hailuo. Furthermore, our model also achieves state-of-the-art performance in long video generation, surpassing other open-source models in this field. Our demo can be viewed at https://landiff.github.io/.
Authors: Alaa Mazouz, Sumanta Chaudhuri, Marco Cagnanzzo, Mihai Mitrea, Enzo Tartaglione, Attilio Fiandrotti
Abstract: Learnable Image Compression (LIC) has shown the potential to outperform standardized video codecs in RD efficiency, prompting the research for hardware-friendly implementations. Most existing LIC hardware implementations prioritize latency to RD-efficiency and through an extensive exploration of the hardware design space. We present a novel design paradigm where the burden of tuning the design for a specific hardware platform is shifted towards model dimensioning and without compromising on RD-efficiency. First, we design a framework for distilling a leaner student LIC model from a reference teacher: by tuning a single model hyperparameters, we can meet the constraints of different hardware platforms without a complex hardware design exploration. Second, we propose a hardware-friendly implementation of the Generalized Divisive Normalization (GDN) activation that preserves RD efficiency even post parameter quantization. Third, we design a pipelined FPGA configuration which takes full advantage of available FPGA resources by leveraging parallel processing and optimizing resource allocation. Our experiments with a state of the art LIC model show that we outperform all existing FPGA implementations while performing very close to the original model in terms of RD efficiency.
Authors: Chan hur, Jeong-hun Hong, Dong-hun Lee, Dabin Kang, Semin Myeong, Sang-hyo Park, Hyeyoung Park
Abstract: In recent text-video retrieval, the use of additional captions from vision-language models has shown promising effects on the performance. However, existing models using additional captions often have struggled to capture the rich semantics, including temporal changes, inherent in the video. In addition, incorrect information caused by generative models can lead to inaccurate retrieval. To address these issues, we propose a new framework, Narrating the Video (NarVid), which strategically leverages the comprehensive information available from frame-level captions, the narration. The proposed NarVid exploits narration in multiple ways: 1) feature enhancement through cross-modal interactions between narration and video, 2) query-aware adaptive filtering to suppress irrelevant or incorrect information, 3) dual-modal matching score by adding query-video similarity and query-narration similarity, and 4) hard-negative loss to learn discriminative features from multiple perspectives using the two similarities from different views. Experimental results demonstrate that NarVid achieves state-of-the-art performance on various benchmark datasets.
Authors: Zebin Xing, Xingyu Zhang, Yang Hu, Bo Jiang, Tong He, Qian Zhang, Xiaoxiao Long, Wei Yin
Abstract: We propose GoalFlow, an end-to-end autonomous driving method for generating high-quality multimodal trajectories. In autonomous driving scenarios, there is rarely a single suitable trajectory. Recent methods have increasingly focused on modeling multimodal trajectory distributions. However, they suffer from trajectory selection complexity and reduced trajectory quality due to high trajectory divergence and inconsistencies between guidance and scene information. To address these issues, we introduce GoalFlow, a novel method that effectively constrains the generative process to produce high-quality, multimodal trajectories. To resolve the trajectory divergence problem inherent in diffusion-based methods, GoalFlow constrains the generated trajectories by introducing a goal point. GoalFlow establishes a novel scoring mechanism that selects the most appropriate goal point from the candidate points based on scene information. Furthermore, GoalFlow employs an efficient generative method, Flow Matching, to generate multimodal trajectories, and incorporates a refined scoring mechanism to select the optimal trajectory from the candidates. Our experimental results, validated on the Navsim\cite{Dauner2024_navsim}, demonstrate that GoalFlow achieves state-of-the-art performance, delivering robust multimodal trajectories for autonomous driving. GoalFlow achieved PDMS of 90.3, significantly surpassing other methods. Compared with other diffusion-policy-based methods, our approach requires only a single denoising step to obtain excellent performance. The code is available at https://github.com/YvanYin/GoalFlow.
Authors: Martijn P. A. Starmans (for the Alzheimers Disease Neuroimaging Initiative), Sebastian R. van der Voort (for the Alzheimers Disease Neuroimaging Initiative), Thomas Phil (for the Alzheimers Disease Neuroimaging Initiative), Milea J. M. Timbergen (for the Alzheimers Disease Neuroimaging Initiative), Melissa Vos (for the Alzheimers Disease Neuroimaging Initiative), Guillaume A. Padmos (for the Alzheimers Disease Neuroimaging Initiative), Wouter Kessels (for the Alzheimers Disease Neuroimaging Initiative), David Hanff (for the Alzheimers Disease Neuroimaging Initiative), Dirk J. Grunhagen (for the Alzheimers Disease Neuroimaging Initiative), Cornelis Verhoef (for the Alzheimers Disease Neuroimaging Initiative), Stefan Sleijfer (for the Alzheimers Disease Neuroimaging Initiative), Martin J. van den Bent (for the Alzheimers Disease Neuroimaging Initiative), Marion Smits (for the Alzheimers Disease Neuroimaging Initiative), Roy S. Dwarkasing (for the Alzheimers Disease Neuroimaging Initiative), Christopher J. Els (for the Alzheimers Disease Neuroimaging Initiative), Federico Fiduzi (for the Alzheimers Disease Neuroimaging Initiative), Geert J. L. H. van Leenders (for the Alzheimers Disease Neuroimaging Initiative), Anela Blazevic (for the Alzheimers Disease Neuroimaging Initiative), Johannes Hofland (for the Alzheimers Disease Neuroimaging Initiative), Tessa Brabander (for the Alzheimers Disease Neuroimaging Initiative), Renza A. H. van Gils (for the Alzheimers Disease Neuroimaging Initiative), Gaston J. H. Franssen (for the Alzheimers Disease Neuroimaging Initiative), Richard A. Feelders (for the Alzheimers Disease Neuroimaging Initiative), Wouter W. de Herder (for the Alzheimers Disease Neuroimaging Initiative), Florian E. Buisman (for the Alzheimers Disease Neuroimaging Initiative), Francois E. J. A. Willemssen (for the Alzheimers Disease Neuroimaging Initiative), Bas Groot Koerkamp (for the Alzheimers Disease Neuroimaging Initiative), Lindsay Angus (for the Alzheimers Disease Neuroimaging Initiative), Astrid A. M. van der Veldt (for the Alzheimers Disease Neuroimaging Initiative), Ana Rajicic (for the Alzheimers Disease Neuroimaging Initiative), Arlette E. Odink (for the Alzheimers Disease Neuroimaging Initiative), Mitchell Deen (for the Alzheimers Disease Neuroimaging Initiative), Jose M. Castillo T. (for the Alzheimers Disease Neuroimaging Initiative), Jifke Veenland (for the Alzheimers Disease Neuroimaging Initiative), Ivo Schoots (for the Alzheimers Disease Neuroimaging Initiative), Michel Renckens (for the Alzheimers Disease Neuroimaging Initiative), Michail Doukas (for the Alzheimers Disease Neuroimaging Initiative), Rob A. de Man (for the Alzheimers Disease Neuroimaging Initiative), Jan N. M. IJzermans (for the Alzheimers Disease Neuroimaging Initiative), Razvan L. Miclea (for the Alzheimers Disease Neuroimaging Initiative), Peter B. Vermeulen (for the Alzheimers Disease Neuroimaging Initiative), Esther E. Bron (for the Alzheimers Disease Neuroimaging Initiative), Maarten G. Thomeer (for the Alzheimers Disease Neuroimaging Initiative), Jacob J. Visser (for the Alzheimers Disease Neuroimaging Initiative), Wiro J. Niessen (for the Alzheimers Disease Neuroimaging Initiative), Stefan Klein (for the Alzheimers Disease Neuroimaging Initiative)
Abstract: Predicting clinical outcomes from medical images using quantitative features (``radiomics'') requires many method design choices, Currently, in new clinical applications, finding the optimal radiomics method out of the wide range of methods relies on a manual, heuristic trial-and-error process. We introduce a novel automated framework that optimizes radiomics workflow construction per application by standardizing the radiomics workflow in modular components, including a large collection of algorithms for each component, and formulating a combined algorithm selection and hyperparameter optimization problem. To solve it, we employ automated machine learning through two strategies (random search and Bayesian optimization) and three ensembling approaches. Results show that a medium-sized random search and straight-forward ensembling perform similar to more advanced methods while being more efficient. Validated across twelve clinical applications, our approach outperforms both a radiomics baseline and human experts. Concluding, our framework improves and streamlines radiomics research by fully automatically optimizing radiomics workflow construction. To facilitate reproducibility, we publicly release six datasets, software of the method, and code to reproduce this study.
Authors: Hai-Long Sun, Da-Wei Zhou, De-Chuan Zhan, Han-Jia Ye
Abstract: While traditional machine learning can effectively tackle a wide range of problems, it primarily operates within a closed-world setting, which presents limitations when dealing with streaming data. As a solution, incremental learning emerges to address real-world scenarios involving new data's arrival. Recently, pre-training has made significant advancements and garnered the attention of numerous researchers. The strong performance of these pre-trained models (PTMs) presents a promising avenue for developing continual learning algorithms that can effectively adapt to real-world scenarios. Consequently, exploring the utilization of PTMs in incremental learning has become essential. This paper introduces a pre-trained model-based continual learning toolbox known as PILOT. On the one hand, PILOT implements some state-of-the-art class-incremental learning algorithms based on pre-trained models, such as L2P, DualPrompt, and CODA-Prompt. On the other hand, PILOT also fits typical class-incremental learning algorithms (e.g., DER, FOSTER, and MEMO) within the context of pre-trained models to evaluate their effectiveness.
Authors: Yilei Wu, Zijian Dong, Chongyao Chen, Wangchunshu Zhou, Juan Helen Zhou
Abstract: In medical image analysis, regression plays a critical role in computer-aided diagnosis. It enables quantitative measurements such as age prediction from structural imaging, cardiac function quantification, and molecular measurement from PET scans. While deep learning has shown promise for these tasks, most approaches focus solely on optimizing regression loss or model architecture, neglecting the quality of learned feature representations which are crucial for robust clinical predictions. Directly applying representation learning techniques designed for classification to regression often results in fragmented representations in the latent space, yielding sub-optimal performance. In this paper, we argue that the potential of contrastive learning for medical image regression has been overshadowed due to the neglect of two crucial aspects: ordinality-awareness and hardness. To address these challenges, we propose Supervised Contrastive Learning for Medical Imaging Regression with Mixup (SupReMix). It takes anchor-inclusive mixtures (mixup of the anchor and a distinct negative sample) as hard negative pairs and anchor-exclusive mixtures (mixup of two distinct negative samples) as hard positive pairs at the embedding level. This strategy formulates harder contrastive pairs by integrating richer ordinal information. Through theoretical analysis and extensive experiments on six datasets spanning MRI, X-ray, ultrasound, and PET modalities, we demonstrate that SupReMix fosters continuous ordered representations, significantly improving regression performance.
Authors: Emanuele Sansone, Robin Manhaeve
Abstract: Self-supervised learning excels at learning representations from large amounts of data. At the same time, generative models offer the complementary property of learning information about the underlying data generation process. In this study, we aim at establishing a principled connection between these two paradigms and highlight the benefits of their complementarity. In particular, we perform an analysis of self-supervised learning objectives, elucidating the underlying probabilistic graphical models and presenting a standardized methodology for their derivation from first principles. The analysis suggests a natural means of integrating self-supervised learning with likelihood-based generative models. We instantiate this concept within the realm of cluster-based self-supervised learning and energy models, introducing a lower bound proven to reliably penalize the most important failure modes. Our theoretical findings are substantiated through experiments on synthetic and real-world data, including SVHN, CIFAR10, and CIFAR100, demonstrating that our objective function allows to jointly train a backbone network in a discriminative and generative fashion, consequently outperforming existing self-supervised learning strategies in terms of clustering, generation and out-of-distribution detection performance by a wide margin. We also demonstrate that the solution can be integrated into a neuro-symbolic framework to tackle a simple yet non-trivial instantiation of the symbol grounding problem.
Authors: Nimrod Shabtay, Eli Schwartz, Raja Giryes
Abstract: Phase-coded imaging is a computational imaging method designed to tackle tasks such as passive depth estimation and extended depth of field (EDOF) using depth cues inserted during image capture. Most of the current deep learning-based methods for depth estimation or all-in-focus imaging require a training dataset with high-quality depth maps and an optimal focus point at infinity for all-in-focus images. Such datasets are difficult to create, usually synthetic, and require external graphic programs. We propose a new method named "Deep Phase Coded Image Prior" (DPCIP) for jointly recovering the depth map and all-in-focus image from a coded-phase image using solely the captured image and the optical information of the imaging system. Our approach does not depend on any specific dataset and surpasses prior supervised techniques utilizing the same imaging system. This improvement is achieved through the utilization of a problem formulation based on implicit neural representation (INR) and deep image prior (DIP). Due to our zero-shot method, we overcome the barrier of acquiring accurate ground-truth data of depth maps and all-in-focus images for each new phase-coded system introduced. This allows focusing mainly on developing the imaging system, and not on ground-truth data collection.
Authors: Peng Yin, Jianhao Jiao, Shiqi Zhao, Lingyun Xu, Guoquan Huang, Howie Choset, Sebastian Scherer, Jianda Han
Abstract: In the realm of robotics, the quest for achieving real-world autonomy, capable of executing large-scale and long-term operations, has positioned place recognition (PR) as a cornerstone technology. Despite the PR community's remarkable strides over the past two decades, garnering attention from fields like computer vision and robotics, the development of PR methods that sufficiently support real-world robotic systems remains a challenge. This paper aims to bridge this gap by highlighting the crucial role of PR within the framework of Simultaneous Localization and Mapping (SLAM) 2.0. This new phase in robotic navigation calls for scalable, adaptable, and efficient PR solutions by integrating advanced artificial intelligence (AI) technologies. For this goal, we provide a comprehensive review of the current state-of-the-art (SOTA) advancements in PR, alongside the remaining challenges, and underscore its broad applications in robotics. This paper begins with an exploration of PR's formulation and key research challenges. We extensively review literature, focusing on related methods on place representation and solutions to various PR challenges. Applications showcasing PR's potential in robotics, key PR datasets, and open-source libraries are discussed. We conclude with a discussion on PR's future directions and provide a summary of the literature covered at: https://github.com/MetaSLAM/GPRS.
Authors: Peng Kuang, Zhibo Wang, Zhixuan Chu, Jingyi Wang, Kui Ren
Abstract: Spurious correlations in training data significantly hinder the generalization capability of machine learning models when faced with distribution shifts in real-world scenarios.To tackle the problem, numerous debiasing approaches have been proposed and benchmarked on datasets intentionally designed with severe biases.However, it remains to be asked: \textit{1. Do existing benchmarks really capture biases in the real world? 2. Can existing debiasing methods handle biases in the real world?} To answer the questions, we revisit biased distributions in existing benchmarks and real-world datasets, and propose a fine-grained framework for analyzing dataset bias by disentangling it into the magnitude and prevalence of bias. We empirically and theoretically identify key characteristics of real-world biases poorly represented by existing benchmarks. We further introduce two novel biased distributions to bridge this gap, forming a systematic evaluation framework for real-world debiasing.With the evaluation framework, we focus on the practical setting of debiasing w/o bias supervision and find existing methods incapable of handling real-world biases.Through in-depth analysis, we propose a simple yet effective approach that can be easily applied to existing debiasing methods, named Debias in Destruction (DiD).Empirical results on real-world datasets in both image and language modalities demonstrate the superiority of DiD, improving the performance of existing methods on all types of biases.
Authors: Ngoc-Du Tran, Thi-Thao Tran, Quang-Huy Nguyen, Manh-Hung Vu, Van-Truong Pham
Abstract: The emergence of deep learning techniques has advanced the image segmentation task, especially for medical images. Many neural network models have been introduced in the last decade bringing the automated segmentation accuracy close to manual segmentation. However, cutting-edge models like Transformer-based architectures rely on large scale annotated training data, and are generally designed with densely consecutive layers in the encoder, decoder, and skip connections resulting in large number of parameters. Additionally, for better performance, they often be pretrained on a larger data, thus requiring large memory size and increasing resource expenses. In this study, we propose a new lightweight but efficient model, namely LiteNeXt, based on convolutions and mixing modules with simplified decoder, for medical image segmentation. The model is trained from scratch with small amount of parameters (0.71M) and Giga Floating Point Operations Per Second (0.42). To handle boundary fuzzy as well as occlusion or clutter in objects especially in medical image regions, we propose the Marginal Weight Loss that can help effectively determine the marginal boundary between object and background. Additionally, the Self-embedding Representation Parallel technique is proposed as an innovative data augmentation strategy that utilizes the network architecture itself for self-learning augmentation, enhancing feature extraction robustness without external data. Experiments on public datasets including Data Science Bowls, GlaS, ISIC2018, PH2, Sunnybrook, and Lung X-ray data show promising results compared to other state-of-the-art CNN-based and Transformer-based architectures. Our code is released at: https://github.com/tranngocduvnvp/LiteNeXt.
Authors: Jueqing Lu, Wray Buntine, Yuanyuan Qi, Joanna Dipnall, Belinda Gabbe, Lan Du
Abstract: Resolving conflicts is essential to make the decisions of multi-view classification more reliable. Much research has been conducted on learning consistent informative representations among different views, assuming that all views are identically important and strictly aligned. However, real-world multi-view data may not always conform to these assumptions, as some views may express distinct information. To address this issue, we develop a computational trust-based discounting method to enhance the existing trustworthy framework in scenarios where conflicts between different views may arise. Its belief fusion process considers the trustworthiness of predictions made by individual views via an instance-wise probability-sensitive trust discounting mechanism. We evaluate our method on six real-world datasets, using Top-1 Accuracy, AUC-ROC for Uncertainty-Aware Prediction, Fleiss' Kappa, and a new metric called Multi-View Agreement with Ground Truth that takes into consideration the ground truth labels. The experimental results show that computational trust can effectively resolve conflicts, paving the way for more reliable multi-view classification models in real-world applications.
Authors: Yu Zhe, Jun Sakuma
Abstract: The widespread use of publicly available datasets for training machine learning models raises significant concerns about data misuse. Availability attacks have emerged as a means for data owners to safeguard their data by designing imperceptible perturbations that degrade model performance when incorporated into training datasets. However, existing availability attacks are ineffective when only a portion of the data can be perturbed. To address this challenge, we propose a novel availability attack approach termed Parameter Matching Attack (PMA). PMA is the first availability attack capable of causing more than a 30\% performance drop when only a portion of data can be perturbed. PMA optimizes perturbations so that when the model is trained on a mixture of clean and perturbed data, the resulting model will approach a model designed to perform poorly. Experimental results across four datasets demonstrate that PMA outperforms existing methods, achieving significant model performance degradation when a part of the training data is perturbed. Our code is available in the supplementary materials.
Authors: Zeqing Wang, Fei Cheng, Kangye Ji, Bohu Huang
Abstract: Continual learning (CL) is a technique that enables neural networks to constantly adapt to their dynamic surroundings. Despite being overlooked for a long time, this technology can considerably address the customized needs of users in edge devices. Actually, most CL methods require huge resource consumption by the training behavior to acquire generalizability among all tasks for delaying forgetting regardless of edge scenarios. Therefore, this paper proposes a compact algorithm called LightCL, which evaluates and compresses the redundancy of already generalized components in structures of the neural network. Specifically, we consider two factors of generalizability, learning plasticity and memory stability, and design metrics of both to quantitatively assess generalizability of neural networks during CL. This evaluation shows that generalizability of different layers in a neural network exhibits a significant variation. Thus, we $\textit{Maintain Generalizability}$ by freezing generalized parts without the resource-intensive training process and $\textit{Memorize Feature Patterns}$ by stabilizing feature extracting of previous tasks to enhance generalizability for less-generalized parts with a little extra memory, which is far less than the reduction by freezing. Experiments illustrate that LightCL outperforms other state-of-the-art methods and reduces at most $\textbf{6.16$\times$}$ memory footprint. We also verify the effectiveness of LightCL on the edge device.
Authors: Xinyi Wang, Angeliki Katsenou, David Bull
Abstract: With the rapid growth of User-Generated Content (UGC) exchanged between users and sharing platforms, the need for video quality assessment in the wild is increasingly evident. UGC is typically acquired using consumer devices and undergoes multiple rounds of compression (transcoding) before reaching the end user. Therefore, traditional quality metrics that employ the original content as a reference are not suitable. In this paper, we propose ReLaX-VQA, a novel No-Reference Video Quality Assessment (NR-VQA) model that aims to address the challenges of evaluating the quality of diverse video content without reference to the original uncompressed videos. ReLaX-VQA uses frame differences to select spatio-temporal fragments intelligently together with different expressions of spatial features associated with the sampled frames. These are then used to better capture spatial and temporal variabilities in the quality of neighbouring frames. Furthermore, the model enhances abstraction by employing layer-stacking techniques in deep neural network features from Residual Networks and Vision Transformers. Extensive testing across four UGC datasets demonstrates that ReLaX-VQA consistently outperforms existing NR-VQA methods, achieving an average SRCC of 0.8658 and PLCC of 0.8873. Open-source code and trained models that will facilitate further research and applications of NR-VQA can be found at https://github.com/xinyiW915/ReLaX-VQA.
Authors: Juan Song, Jiaxiang He, Lijie Yang, Mingtao Feng, Keyan Wang
Abstract: Diffusion probabilistic models have recently achieved remarkable success in generating high-quality images. However, balancing high perceptual quality and low distortion remains challenging in image compression applications. To address this issue, we propose an efficient Uncertainty-Guided image compression approach with wavelet Diffusion (UGDiff). Our approach focuses on high frequency compression via the wavelet transform, since high frequency components are crucial for reconstructing image details. We introduce a wavelet conditional diffusion model for high frequency prediction, followed by a residual codec that compresses and transmits prediction residuals to the decoder. This diffusion prediction-then-residual compression paradigm effectively addresses the low fidelity issue common in direct reconstructions by existing diffusion models. Considering the uncertainty from the random sampling of the diffusion model, we further design an uncertainty-weighted rate-distortion (R-D) loss tailored for residual compression, providing a more rational trade-off between rate and distortion. Comprehensive experiments on two benchmark datasets validate the effectiveness of UGDiff, surpassing state-of-the-art image compression methods in R-D performance, perceptual quality, subjective quality, and inference time. Our code is available at: https://github.com/hejiaxiang1/Wavelet-Diffusion/tree/main
URLs: https://github.com/hejiaxiang1/Wavelet-Diffusion/tree/main
Authors: Fabian Schmidt, Constantin Blessing, Markus Enzweiler, Abhinav Valada
Abstract: Simultaneous Localization and Mapping (SLAM) is essential for mobile robotics, enabling autonomous navigation in dynamic, unstructured outdoor environments without relying on external positioning systems. These environments pose significant challenges due to variable lighting, weather conditions, and complex terrain. Visual-Inertial SLAM has emerged as a promising solution for robust localization under such conditions. This paper benchmarks several open-source Visual-Inertial SLAM systems, including traditional methods (ORB-SLAM3, VINS-Fusion, OpenVINS, Kimera, and SVO Pro) and learning-based approaches (HFNet-SLAM, AirSLAM), to evaluate their performance in unstructured natural outdoor settings. We focus on the impact of loop closing on localization accuracy and computational demands, providing a comprehensive analysis of these systems' effectiveness in real-world environments and especially their application to embedded systems in outdoor robotics. Our contributions further include an assessment of varying frame rates on localization accuracy and computational load. The findings highlight the importance of loop closing in improving localization accuracy while managing computational resources efficiently, offering valuable insights for optimizing Visual-Inertial SLAM systems for practical outdoor applications in mobile robotics. The dataset and the benchmark code are available under https://github.com/iis-esslingen/vi-slam_lc_benchmark.
URLs: https://github.com/iis-esslingen/vi-slam_lc_benchmark.
Authors: Lin Zuo, Yongqi Ding, Wenwei Luo, Mengmeng Jing, Kunshan Yang
Abstract: Spiking neural networks (SNNs) have received widespread attention as an ultra-low power computing paradigm. Recent studies have shown that SNNs suffer from severe overfitting, which limits their generalization performance. In this paper, we propose a simple yet effective Temporal Reversal Regularization (TRR) to mitigate overfitting during training and facilitate generalization of SNNs. We exploit the inherent temporal properties of SNNs to perform input/feature temporal reversal perturbations, prompting the SNN to produce original-reversed consistent outputs and learn perturbation-invariant representations. To further enhance generalization, we utilize the lightweight ``star operation" (Hadamard product) for feature hybridization of original and temporally reversed spike firing rates, which expands the implicit dimensionality and acts as a spatio-temporal regularizer. We show theoretically that our method is able to tighten the upper bound of the generalization error, and extensive experiments on static/neuromorphic recognition as well as 3D point cloud classification tasks demonstrate its effectiveness, versatility, and adversarial robustness. In particular, our regularization significantly improves the recognition accuracy of low-latency SNN for neuromorphic objects, contributing to the real-world deployment of neuromorphic computational software-hardware integration.
Authors: Jiaze E, Srutarshi Banerjee, Tekin Bicer, Guannan Wang, Yanfu Zhang, Bin Ren
Abstract: Computed tomography (CT) is widely used in industrial and medical imaging, but sparse-view scanning reduces radiation exposure at the cost of incomplete sinograms and challenging reconstruction. Existing RGB-based inpainting models struggle with severe feature entanglement, while sinogram-specific methods often lack explicit physics constraints. We propose FCDM, a physics-guided, frequency-aware sinogram inpainting framework. It integrates bidirectional frequency-domain convolutions to disentangle overlapping features while enforcing total absorption and frequency-domain consistency via a physics-informed loss. To enhance diffusion-based restoration, we introduce a Fourier-enhanced mask embedding to encode angular dependencies and a frequency-adaptive noise scheduling strategy that incorporates a soft row-wise absorption constraint to maintain physical realism. Experiments on synthetic and real-world datasets show that FCDM outperforms existing methods, achieving SSIM over 0.95 and PSNR above 30 dB, with up to 33% and 29% improvements over baselines.
Authors: Tianao Li, Manxiu Cui, Cheng Ma, Emma Alexander
Abstract: Photoacoustic computed tomography (PACT) is a non-invasive imaging modality, similar to ultrasound, with wide-ranging medical applications. Conventional PACT images are degraded by wavefront distortion caused by the heterogeneous speed of sound (SOS) in tissue. Accounting for these effects can improve image quality and provide medically useful information, but measuring the SOS directly is burdensome and the existing joint reconstruction method is computationally expensive. Traditional supervised learning techniques are currently inaccessible in this data-starved domain. In this work, we introduce an efficient, self-supervised joint reconstruction method that recovers SOS and high-quality images using a differentiable physics model to solve the semi-blind inverse problem. The SOS, parametrized by either a pixel grid or a neural field (NF), is updated directly by backpropagation. Our method removes SOS aberrations more accurately and 35x faster than the current SOTA. We demonstrate the success of our method quantitatively in simulation and qualitatively on experimentally-collected and in-vivo data.
Authors: Jun Liu, Geng Yuan, Weihao Zeng, Hao Tang, Wenbin Zhang, Xue Lin, XiaoLin Xu, Dong Huang, Yanzhi Wang
Abstract: In research findings, co-deletion of the 1p/19q gene is associated with clinical outcomes in low-grade gliomas. The ability to predict 1p19q status is critical for treatment planning and patient follow-up. This study aims to utilize a specially MRI-based convolutional neural network for brain cancer detection. Although public networks such as RestNet and AlexNet can effectively diagnose brain cancers using transfer learning, the model includes quite a few weights that have nothing to do with medical images. As a result, the diagnostic results are unreliable by the transfer learning model. To deal with the problem of trustworthiness, we create the model from the ground up, rather than depending on a pre-trained model. To enable flexibility, we combined convolution stacking with a dropout and full connect operation, it improved performance by reducing overfitting. During model training, we also supplement the given dataset and inject Gaussian noise. We use three--fold cross-validation to train the best selection model. Comparing InceptionV3, VGG16, and MobileNetV2 fine-tuned with pre-trained models, our model produces better results. On an validation set of 125 codeletion vs. 31 not codeletion images, the proposed network achieves 96.37\% percent F1-score, 97.46\% percent precision, and 96.34\% percent recall when classifying 1p/19q codeletion and not codeletion images.
Authors: Wanpeng Zhang, Zilong Xie, Yicheng Feng, Yijiang Li, Xingrun Xing, Sipeng Zheng, Zongqing Lu
Abstract: Multimodal Large Language Models have made significant strides in integrating visual and textual information, yet they often struggle with effectively aligning these modalities. We introduce a novel image tokenizer that bridges this gap by applying the principle of Byte-Pair Encoding (BPE) to visual data. Unlike conventional approaches that rely on separate visual encoders, our method directly incorporates structural prior information into image tokens, mirroring the successful tokenization strategies used in text-only Large Language Models. This innovative approach enables Transformer models to more effectively learn and reason across modalities. Through theoretical analysis and extensive experiments, we demonstrate that our BPE Image Tokenizer significantly enhances MLLMs' multimodal understanding capabilities, even with limited training data. Leveraging this method, we develop Being-VL-0, a model that demonstrates superior performance across various benchmarks and shows promising scalability, potentially paving the way for more efficient and capable multimodal foundation models.
Authors: Peter Yichen Chen, Chao Liu, Pingchuan Ma, John Eastman, Daniela Rus, Dylan Randle, Yuri Ivanov, Wojciech Matusik
Abstract: Differentiable simulation has become a powerful tool for system identification. While prior work has focused on identifying robot properties using robot-specific data or object properties using object-specific data, our approach calibrates object properties by using information from the robot, without relying on data from the object itself. Specifically, we utilize robot joint encoder information, which is commonly available in standard robotic systems. Our key observation is that by analyzing the robot's reactions to manipulated objects, we can infer properties of those objects, such as inertia and softness. Leveraging this insight, we develop differentiable simulations of robot-object interactions to inversely identify the properties of the manipulated objects. Our approach relies solely on proprioception -- the robot's internal sensing capabilities -- and does not require external measurement tools or vision-based tracking systems. This general method is applicable to any articulated robot and requires only joint position information. We demonstrate the effectiveness of our method on a low-cost robotic platform, achieving accurate mass and elastic modulus estimations of manipulated objects with just a few seconds of computation on a laptop.
Authors: Melanie Rieff, Fabian Holzberger, Oksana Lapina, Geir Ringstad, Lars Magnus Valnes, Bogna Warsza, Kent-Andre Mardal, Per Kristian Eide, Barbara Wohlmuth
Abstract: Previous work indicates evidence that cerebrospinal fluid (CSF) plays a crucial role in brain waste clearance processes, and that altered flow patterns are associated with various diseases of the central nervous system. In this study, we investigate the potential of deep learning to predict the distribution in human brain of a gadolinium-based CSF contrast agent (tracer) administered intrathecal. For this, T1-weighted magnetic resonance imaging (MRI) scans taken at multiple time points before and after injection were utilized. We propose a U-net-based supervised learning model to predict pixel-wise signal increase at its peak after 24 hours. Performance is evaluated based on different tracer distribution stages provided during training, including predictions from baseline scans taken before injection. Our findings show that training with imaging data from only the first two hours post-injection yields tracer flow predictions comparable to models trained with additional later-stage scans. Validation against ventricular reflux gradings from neuroradiologists confirmed alignment with expert evaluations. These results demonstrate that deep learning-based methods for CSF flow prediction deserve more attention, as minimizing MR imaging without compromising clinical analysis could enhance efficiency, improve patient well-being, and lower healthcare costs.
Authors: Matthew Strong, Boshu Lei, Aiden Swann, Wen Jiang, Kostas Daniilidis, Monroe Kennedy III
Abstract: We propose a framework for active next best view and touch selection for robotic manipulators using 3D Gaussian Splatting (3DGS). 3DGS is emerging as a useful explicit 3D scene representation for robotics, as it has the ability to represent scenes in a both photorealistic and geometrically accurate manner. However, in real-world, online robotic scenes where the number of views is limited given efficiency requirements, random view selection for 3DGS becomes impractical as views are often overlapping and redundant. We address this issue by proposing an end-to-end online training and active view selection pipeline, which enhances the performance of 3DGS in few-view robotics settings. We first elevate the performance of few-shot 3DGS with a novel semantic depth alignment method using Segment Anything Model 2 (SAM2) that we supplement with Pearson depth and surface normal loss to improve color and depth reconstruction of real-world scenes. We then extend FisherRF, a next-best-view selection method for 3DGS, to select views and touch poses based on depth uncertainty. We perform online view selection on a real robot system during live 3DGS training. We motivate our improvements to few-shot GS scenes, and extend depth-based FisherRF to them, where we demonstrate both qualitative and quantitative improvements on challenging robot scenes. For more information, please see our project page at https://arm.stanford.edu/next-best-sense.
Authors: Florian Hahlbohm, Fabian Friederichs, Tim Weyrich, Linus Franke, Moritz Kappel, Susana Castillo, Marc Stamminger, Martin Eisemann, Marcus Magnor
Abstract: 3D Gaussian Splats (3DGS) have proven a versatile rendering primitive, both for inverse rendering as well as real-time exploration of scenes. In these applications, coherence across camera frames and multiple views is crucial, be it for robust convergence of a scene reconstruction or for artifact-free fly-throughs. Recent work started mitigating artifacts that break multi-view coherence, including popping artifacts due to inconsistent transparency sorting and perspective-correct outlines of (2D) splats. At the same time, real-time requirements forced such implementations to accept compromises in how transparency of large assemblies of 3D Gaussians is resolved, in turn breaking coherence in other ways. In our work, we aim at achieving maximum coherence, by rendering fully perspective-correct 3D Gaussians while using a high-quality approximation of accurate blending, hybrid transparency, on a per-pixel level, in order to retain real-time frame rates. Our fast and perspectively accurate approach for evaluation of 3D Gaussians does not require matrix inversions, thereby ensuring numerical stability and eliminating the need for special handling of degenerate splats, and the hybrid transparency formulation for blending maintains similar quality as fully resolved per-pixel transparencies at a fraction of the rendering costs. We further show that each of these two components can be independently integrated into Gaussian splatting systems. In combination, they achieve up to 2$\times$ higher frame rates, 2$\times$ faster optimization, and equal or better image quality with fewer rendering artifacts compared to traditional 3DGS on common benchmarks.
Authors: Yijiang Li, Qingying Gao, Tianwei Zhao, Bingyang Wang, Haoran Sun, Haiyun Lyu, Dezhi Luo, Hokin Deng
Abstract: While Multimodal Large Language Models (MLLMs) demonstrate impressive abilities over high level perception and reasoning, their robustness in the wild still lags behind humans and exhibits diminished efficacy on simple tasks that are intuitive for humans. We examine the hypothesis that these deficiencies stem from the absence of core knowledge, rudimentary cognitive abilities innate to humans from early childhood. To probe core knowledge representation in MLLMs, we draw from developmental cognitive sciences and develop a large-scale benchmark, CoreCognition dataset, encompassing 12 core cognitive concepts. We evaluate 219 models with 10 different prompts, leading to a total of 2409 data points for analysis. Our findings reveal core knowledge deficits in early developed core abilities while models demonstrate human comparable performance in high level cognition. Moreover, we find that low level abilities show little to no scaling, in stark contrast to high level abilities. Finally, we introduce an evaluation technique, Concept Hacking, through which we demonstrate that MLLMs do not genuinely advance toward core knowledge but instead rely on illusory understanding and shortcut learning as they scale. Website with this $\href{https://growing-ai-like-a-child.github.io/}{link}$.
Authors: Murtaza Dalal, Min Liu, Walter Talbott, Chen Chen, Deepak Pathak, Jian Zhang, Ruslan Salakhutdinov
Abstract: Sim2real for robotic manipulation is difficult due to the challenges of simulating complex contacts and generating realistic task distributions. To tackle the latter problem, we introduce ManipGen, which leverages a new class of policies for sim2real transfer: local policies. Locality enables a variety of appealing properties including invariances to absolute robot and object pose, skill ordering, and global scene configuration. We combine these policies with foundation models for vision, language and motion planning and demonstrate SOTA zero-shot performance of our method to Robosuite benchmark tasks in simulation (97%). We transfer our local policies from simulation to reality and observe they can solve unseen long-horizon manipulation tasks with up to 8 stages with significant pose, object and scene configuration variation. ManipGen outperforms SOTA approaches such as SayCan, OpenVLA, LLMTrajGen and VoxPoser across 50 real-world manipulation tasks by 36%, 76%, 62% and 60% respectively. Video results at https://mihdalal.github.io/manipgen/
Authors: Plabon Paul, Md. Nazmul Islam, Fazle Rafsani, Pegah Khorasani, Shovito Barua Soumma
Abstract: Uncontrolled cell division in the brain is what gives rise to brain tumors. If the tumor size increases by more than half, there is little hope for the patient's recovery. This emphasizes the need of rapid and precise brain tumor diagnosis. When it comes to analyzing, diagnosing, and planning therapy for brain tumors, MRI imaging plays a crucial role. A brain tumor's development history is crucial information for doctors to have. When it comes to distinguishing between human soft tissues, MRI scans are superior. In order to get reliable classification results from MRI scans quickly, deep learning is one of the most practical methods. Early human illness diagnosis has been demonstrated to be more accurate when deep learning methods are used. In the case of diagnosing a brain tumor, when even a little misdiagnosis might have serious consequences, accuracy is especially important. Disclosure of brain tumors in medical images is still a difficult task. Brain MRIs are notoriously imprecise in revealing the presence or absence of tumors. Using MRI scans of the brain, a CNN was trained to identify the presence of a tumor in this research. Results from the CNN model showed an accuracy of 99.17%. The CNN model's characteristics were also retrieved. The CNN model's characteristics were also retrieved and we also localized the tumor regions from the unannotated images using GradCAM, a deep learning explainability tool. In order to evaluate the CNN model's capability for processing images, we applied the features into different ML models. CNN and machine learning models were also evaluated using the standard metrics of Precision, Recall, Specificity, and F1 score. The significance of the doctor's diagnosis enhanced the accuracy of the CNN model's assistance in identifying the existence of tumor and treating the patient.
Authors: Bin Chen, Gehui Li, Rongyuan Wu, Xindong Zhang, Jie Chen, Jian Zhang, Lei Zhang
Abstract: Real-world image super-resolution (Real-ISR) aims to reconstruct high-resolution images from low-resolution inputs degraded by complex, unknown processes. While many Stable Diffusion (SD)-based Real-ISR methods have achieved remarkable success, their slow, multi-step inference hinders practical deployment. Recent SD-based one-step networks like OSEDiff and S3Diff alleviate this issue but still incur high computational costs due to their reliance on large pretrained SD models. This paper proposes a novel Real-ISR method, AdcSR, by distilling the one-step diffusion network OSEDiff into a streamlined diffusion-GAN model under our Adversarial Diffusion Compression (ADC) framework. We meticulously examine the modules of OSEDiff, categorizing them into two types: (1) Removable (VAE encoder, prompt extractor, text encoder, etc.) and (2) Prunable (denoising UNet and VAE decoder). Since direct removal and pruning can degrade the model's generation capability, we pretrain our pruned VAE decoder to restore its ability to decode images and employ adversarial distillation to compensate for performance loss. This ADC-based diffusion-GAN hybrid design effectively reduces complexity by 73% in inference time, 78% in computation, and 74% in parameters, while preserving the model's generation capability. Experiments manifest that our proposed AdcSR achieves competitive recovery quality on both synthetic and real-world datasets, offering up to 9.3$\times$ speedup over previous one-step diffusion-based methods. Code and models are available at https://github.com/Guaishou74851/AdcSR.
Authors: Binqian Xu, Xiangbo Shu, Haiyang Mei, Guosen Xie, Basura Fernando, Jinhui Tang
Abstract: Multimodal Large Language Models (MLLMs) have made significant advancements, demonstrating powerful capabilities in processing and understanding multimodal data. Fine-tuning MLLMs with Federated Learning (FL) allows for expanding the training data scope by including private data sources, thereby enhancing their practical applicability in privacy-sensitive domains. However, current research remains in the early stage, particularly in addressing the \textbf{multimodal heterogeneities} in real-world applications. In this paper, we introduce a benchmark to evaluate the performance of federated fine-tuning of MLLMs across various multimodal heterogeneous scenarios, laying the groundwork for future research in the field. Our benchmark includes two lightweight MLLMs, two downstream tasks, three evaluation metrics, and five datasets across three domains, along with six comparison baselines, covering over ten types of modality heterogeneities across four multimodal scenarios. To address the challenges posed by multimodal heterogeneity, we develop a general FedMLLM framework that integrates classic FL methods alongside two modality-agnostic strategies. Extensive experimental results show that our proposed FL paradigm improves the performance of MLLMs by broadening the range of training data and mitigating multimodal heterogeneity. Code is available in supplementary materials.
Authors: Ri-Zhao Qiu, Yuchen Song, Xuanbin Peng, Sai Aneesh Suryadevara, Ge Yang, Minghuan Liu, Mazeyu Ji, Chengzhe Jia, Ruihan Yang, Xueyan Zou, Xiaolong Wang
Abstract: 'In-the-wild' mobile manipulation aims to deploy robots in diverse real-world environments, which requires the robot to (1) have skills that generalize across object configurations; (2) be capable of long-horizon task execution in diverse environments; and (3) perform complex manipulation beyond pick-and-place. Quadruped robots with manipulators hold promise for extending the workspace and enabling robust locomotion, but existing results do not investigate such a capability. This paper proposes WildLMa with three components to address these issues: (1) adaptation of learned low-level controller for VR-enabled whole-body teleoperation and traversability; (2) WildLMa-Skill -- a library of generalizable visuomotor skills acquired via imitation learning or heuristics and (3) WildLMa-Planner -- an interface of learned skills that allow LLM planners to coordinate skills for long-horizon tasks. We demonstrate the importance of high-quality training data by achieving higher grasping success rate over existing RL baselines using only tens of demonstrations. WildLMa exploits CLIP for language-conditioned imitation learning that empirically generalizes to objects unseen in training demonstrations. Besides extensive quantitative evaluation, we qualitatively demonstrate practical robot applications, such as cleaning up trash in university hallways or outdoor terrains, operating articulated objects, and rearranging items on a bookshelf.
Authors: Yong Xie, Weijie Zheng, Hanxun Huang, Guangnan Ye, Xingjun Ma
Abstract: As deep learning models are increasingly deployed in safety-critical applications, evaluating their vulnerabilities to adversarial perturbations is essential for ensuring their reliability and trustworthiness. Over the past decade, a large number of white-box adversarial robustness evaluation methods (i.e., attacks) have been proposed, ranging from single-step to multi-step methods and from individual to ensemble methods. Despite these advances, challenges remain in conducting meaningful and comprehensive robustness evaluations, particularly when it comes to large-scale testing and ensuring evaluations reflect real-world adversarial risks. In this work, we focus on image classification models and propose a novel individual attack method, Probability Margin Attack (PMA), which defines the adversarial margin in the probability space rather than the logits space. We analyze the relationship between PMA and existing cross-entropy or logits-margin-based attacks, and show that PMA can outperform the current state-of-the-art individual methods. Building on PMA, we propose two types of ensemble attacks that balance effectiveness and efficiency. Furthermore, we create a million-scale dataset, CC1M, derived from the existing CC3M dataset, and use it to conduct the first million-scale white-box adversarial robustness evaluation of adversarially-trained ImageNet models. Our findings provide valuable insights into the robustness gaps between individual versus ensemble attacks and small-scale versus million-scale evaluations.
Authors: Marco Paul E. Apolinario, Sakshi Choudhary, Kaushik Roy
Abstract: Continual learning (CL) - the ability to progressively acquire and integrate new concepts - is essential to intelligent systems to adapt to dynamic environments. However, deep neural networks struggle with catastrophic forgetting (CF) when learning tasks sequentially, as training for new tasks often overwrites previously learned knowledge. To address this, recent approaches constrain updates to orthogonal subspaces using gradient projection, effectively preserving important gradient directions for previous tasks. While effective in reducing forgetting, these approaches inadvertently hinder forward knowledge transfer (FWT), particularly when tasks are highly correlated. In this work, we propose Conceptor-based gradient projection for Deep Continual Learning (CODE-CL), a novel method that leverages conceptor matrix representations, a form of regularized reconstruction, to adaptively handle highly correlated tasks. CODE-CL mitigates CF by projecting gradients onto pseudo-orthogonal subspaces of previous task feature spaces while simultaneously promoting FWT. It achieves this by learning a linear combination of shared basis directions, allowing efficient balance between stability and plasticity and transfer of knowledge between overlapping input feature representations. Extensive experiments on continual learning benchmarks validate CODE-CL's efficacy, demonstrating superior performance, reduced forgetting, and improved FWT as compared to state-of-the-art methods.
Authors: Wei Guo, Heng Wang, Jianbo Ma, Weidong Cai
Abstract: Vision-to-audio (V2A) synthesis has broad applications in multimedia. Recent advancements of V2A methods have made it possible to generate relevant audios from inputs of videos or still images. However, the immersiveness and expressiveness of the generation are limited. One possible problem is that existing methods solely rely on the global scene and overlook details of local sounding objects (i.e., sound sources). To address this issue, we propose a Sound Source-Aware V2A (SSV2A) generator. SSV2A is able to locally perceive multimodal sound sources from a scene with visual detection and cross-modality translation. It then contrastively learns a Cross-Modal Sound Source (CMSS) Manifold to semantically disambiguate each source. Finally, we attentively mix their CMSS semantics into a rich audio representation, from which a pretrained audio generator outputs the sound. To model the CMSS manifold, we curate a novel single-sound-source visual-audio dataset VGGS3 from VGGSound. We also design a Sound Source Matching Score to measure localized audio relevance. By addressing V2A generation at the sound-source level, SSV2A surpasses state-of-the-art methods in both generation fidelity and relevance as evidenced by extensive experiments. We further demonstrate SSV2A's ability to achieve intuitive V2A control by compositing vision, text, and audio conditions. Our generation can be tried and heard at https://ssv2a.github.io/SSV2A-demo .
Authors: Haosheng Li, Weixin Mao, Weipeng Deng, Chenyu Meng, Haoqiang Fan, Tiancai Wang, Ping Tan, Hongan Wang, Xiaoming Deng
Abstract: Multi-hand semantic grasp generation aims to generate feasible and semantically appropriate grasp poses for different robotic hands based on natural language instructions. Although the task is highly valuable, due to the lack of multihand grasp datasets with fine-grained contact description between robotic hands and objects, it is still a long-standing difficult task. In this paper, we present Multi-GraspSet, the first large-scale multi-hand grasp dataset with automatically contact annotations. Based on Multi-GraspSet, we propose Multi-GraspLLM, a unified language-guided grasp generation framework, which leverages large language models (LLM) to handle variable-length sequences, generating grasp poses for diverse robotic hands in a single unified architecture. Multi-GraspLLM first aligns the encoded point cloud features and text features into a unified semantic space. It then generates grasp bin tokens that are subsequently converted into grasp pose for each robotic hand via hand-aware linear mapping. The experimental results demonstrate that our approach significantly outperforms existing methods in both real-world experiments and simulator. More information can be found on our project page https://multi-graspllm.github.io.
Authors: Leonardo Barcellona, Andrii Zadaianchuk, Davide Allegro, Samuele Papa, Stefano Ghidoni, Efstratios Gavves
Abstract: A world model provides an agent with a representation of its environment, enabling it to predict the causal consequences of its actions. Current world models typically cannot directly and explicitly imitate the actual environment in front of a robot, often resulting in unrealistic behaviors and hallucinations that make them unsuitable for real-world robotics applications. To overcome those challenges, we propose to rethink robot world models as learnable digital twins. We introduce DreMa, a new approach for constructing digital twins automatically using learned explicit representations of the real world and its dynamics, bridging the gap between traditional digital twins and world models. DreMa replicates the observed world and its structure by integrating Gaussian Splatting and physics simulators, allowing robots to imagine novel configurations of objects and to predict the future consequences of robot actions thanks to its compositionality. We leverage this capability to generate new data for imitation learning by applying equivariant transformations to a small set of demonstrations. Our evaluations across various settings demonstrate significant improvements in accuracy and robustness by incrementing actions and object distributions, reducing the data needed to learn a policy and improving the generalization of the agents. As a highlight, we show that a real Franka Emika Panda robot, powered by DreMa's imagination, can successfully learn novel physical tasks from just a single example per task variation (one-shot policy learning). Our project page can be found in: https://dreamtomanipulate.github.io/.
Authors: Jinliang Zheng, Jianxiong Li, Dongxiu Liu, Yinan Zheng, Zhihao Wang, Zhonghong Ou, Yu Liu, Jingjing Liu, Ya-Qin Zhang, Xianyuan Zhan
Abstract: Training on diverse, internet-scale data is a key factor in the success of recent large foundation models. Yet, using the same recipe for building embodied agents has faced noticeable difficulties. Despite the availability of many crowd-sourced embodied datasets, their action spaces often exhibit significant heterogeneity due to distinct physical embodiment and control interfaces for different robots, causing substantial challenges in developing embodied foundation models using cross-domain data. In this paper, we introduce UniAct, a new embodied foundation modeling framework operating in a Universal Action Space. Our learned universal actions capture the generic atomic behaviors across diverse robots by exploiting their shared structural features, and enable enhanced cross-domain data utilization and cross-embodiment generalizations by eliminating the notorious heterogeneity. The universal actions can be efficiently translated back to heterogeneous actionable commands by simply adding embodiment-specific details, from which fast adaptation to new robots becomes simple and straightforward. Our 0.5B instantiation of UniAct outperforms 14X larger SOTA embodied foundation models in extensive evaluations on various real-world and simulation robots, showcasing exceptional cross-embodiment control and adaptation capability, highlighting the crucial benefit of adopting universal actions. Project page: https://github.com/2toinf/UniAct
Authors: Chenxu Wu, Qingpeng Kong, Zihang Jiang, S. Kevin Zhou
Abstract: Magnetic Resonance Imaging (MRI), including diffusion MRI (dMRI), serves as a ``microscope'' for anatomical structures and routinely mitigates the influence of low signal-to-noise ratio scans by compromising temporal or spatial resolution. However, these compromises fail to meet clinical demands for both efficiency and precision. Consequently, denoising is a vital preprocessing step, particularly for dMRI, where clean data is unavailable. In this paper, we introduce Di-Fusion, a fully self-supervised denoising method that leverages the latter diffusion steps and an adaptive sampling process. Unlike previous approaches, our single-stage framework achieves efficient and stable training without extra noise model training and offers adaptive and controllable results in the sampling process. Our thorough experiments on real and simulated data demonstrate that Di-Fusion achieves state-of-the-art performance in microstructure modeling, tractography tracking, and other downstream tasks. Code is available at https://github.com/FouierL/Di-Fusion.
Authors: Mahshid Shiri, Chandra Bortolotto, Alessandro Bruno, Alessio Consonni, Daniela Maria Grasso, Leonardo Brizzi, Daniele Loiacono, Lorenzo Preda
Abstract: Introduction: Generative Adversarial Networks (GANs) are increasingly used to generate synthetic medical images, addressing the critical shortage of annotated data for training Artificial Intelligence (AI) systems. This study introduces a novel memory-efficient GAN architecture, incorporating Conditional Random Fields (CRFs) to generate high-resolution 3D medical images and evaluates its performance against the state-of-the-art hierarchical (HA)-GAN model. Materials and Methods: The CRF-GAN was trained using the open-source lung CT LUNA16 dataset. The architecture was compared to HA-GAN through a quantitative evaluation, using Frechet Inception Distance (FID) and Maximum Mean Discrepancy (MMD) metrics, and a qualitative evaluation, through a two-alternative forced choice (2AFC) test completed by a pool of 12 resident radiologists, in order to assess the realism of the generated images. Results: CRF-GAN outperformed HA-GAN with lower FID (0.047 vs. 0.061) and MMD (0.084 vs. 0.086) scores, indicating better image fidelity. The 2AFC test showed a significant preference for images generated by CRF-Gan over those generated by HA-GAN with a p-value of 1.93e-05. Additionally, CRF-GAN demonstrated 9.34% lower memory usage at 256 resolution and achieved up to 14.6% faster training speeds, offering substantial computational savings. Discussion: CRF-GAN model successfully generates high-resolution 3D medical images with non-inferior quality to conventional models, while being more memory-efficient and faster. Computational power and time saved can be used to improve the spatial resolution and anatomical accuracy of generated images, which is still a critical factor limiting their direct clinical applicability.
Authors: Bo Wen, Anna Heinke, Akshay Agnihotri, Dirk-Uwe Bartsch, William Freeman, Truong Nguyen, Cheolhong An
Abstract: We identify two major limitations in the existing studies on retinal vessel segmentation: (1) Most existing works are restricted to one modality, i.e, the Color Fundus (CF). However, multi-modality retinal images are used every day in the study of retina and retinal diseases, and the study of vessel segmentation on the other modalities is scarce; (2) Even though a small amount of works extended their experiments to limited new modalities such as the Multi-Color Scanning Laser Ophthalmoscopy (MC), these works still require finetuning a separate model for the new modality. The finetuning will require extra training data, which is difficult to acquire. In this work, we present a foundational universal vessel segmentation model (UVSM) for multi-modality retinal images. Not only do we perform the study on a much wider range of modalities, but we also propose a universal model to segment the vessels in all these commonly-used modalities. Despite being much more versatile comparing with existing methods, our universal model still demonstrates comparable performance with the state-of-the-art finetuned methods. To the best of our knowledge, this is the first work that achieves cross-modality retinal vessel segmentation and also the first work to study retinal vessel segmentation in some novel modalities.
Authors: Duong Anh Kiet
Abstract: We present our solution to the AAAI-25 VRD-IU challenge, achieving first place in the competition. Our approach integrates large margin loss for improved feature discrimination and employs heuristic rules to refine hierarchical relationships. By combining a deep learning-based matching strategy with greedy algorithms, we achieve a significant boost in accuracy while maintaining computational efficiency. Our method attains an accuracy of 0.98904 on the private leaderboard, demonstrating its effectiveness in document structure parsing. Source codes are publicly available at https://github.com/ffyyytt/VRUID-AAAI-DAKiet
Authors: Massimiliano Ciranni, Vito Paolo Pastore, Roberto Di Via, Enzo Tartaglione, Francesca Odone, Vittorio Murino
Abstract: Deep learning model effectiveness in classification tasks is often challenged by the quality and quantity of training data whenever they are affected by strong spurious correlations between specific attributes and target labels. This results in a form of bias affecting training data, which typically leads to unrecoverable weak generalization in prediction. This paper aims at facing this problem by leveraging bias amplification with generated synthetic data: we introduce Diffusing DeBias (DDB), a novel approach acting as a plug-in for common methods of unsupervised model debiasing exploiting the inherent bias-learning tendency of diffusion models in data generation. Specifically, our approach adopts conditional diffusion models to generate synthetic bias-aligned images, which replace the original training set for learning an effective bias amplifier model that we subsequently incorporate into an end-to-end and a two-step unsupervised debiasing approach. By tackling the fundamental issue of bias-conflicting training samples memorization in learning auxiliary models, typical of this type of techniques, our proposed method beats current state-of-the-art in multiple benchmark datasets, demonstrating its potential as a versatile and effective tool for tackling bias in deep learning models.
Authors: Kung-Hsiang Huang, Can Qin, Haoyi Qiu, Philippe Laban, Shafiq Joty, Caiming Xiong, Chien-Sheng Wu
Abstract: Vision Language Models (VLMs) have achieved remarkable progress in multimodal tasks, yet they often struggle with visual arithmetic, seemingly simple capabilities like object counting or length comparison, which are essential for relevant complex tasks like chart understanding and geometric reasoning. In this work, we first investigate the root causes of this deficiency through a suite of probing tasks focusing on basic visual arithmetic. Our analysis reveals that while pre-trained vision encoders typically capture sufficient information, the text decoder often fails to decode it correctly for arithmetic reasoning. To address this, we propose CogAlign, a novel post-training strategy inspired by Piaget's theory of cognitive development. CogAlign trains VLMs to recognize invariant properties under visual transformations. We demonstrate that this approach significantly improves the performance of three diverse VLMs on our proposed probing tasks. Furthermore, CogAlign enhances performance by an average of 4.6% on CHOCOLATE and 2.9% on MATH-VISION, outperforming or matching supervised fine-tuning methods while requiring only 60% less training data. These results highlight the effectiveness and generalizability of CogAlign in improving fundamental visual arithmetic capabilities and their transfer to downstream tasks.
Authors: Yi Fang, Bowen Jin, Jiacheng Shen, Sirui Ding, Qiaoyu Tan, Jiawei Han
Abstract: The rapid development of Multimodal Large Language Models (MLLMs) has enabled the integration of multiple modalities, including texts and images, within the large language model (LLM) framework. However, texts and images are usually interconnected, forming a multimodal attributed graph (MMAG). It is underexplored how MLLMs can incorporate the relational information (\textit{i.e.}, graph structure) and semantic information (\textit{i.e.,} texts and images) on such graphs for multimodal comprehension and generation. In this paper, we propose GraphGPT-o, which supports omni-multimodal understanding and creation on MMAGs. We first comprehensively study linearization variants to transform semantic and structural information as input for MLLMs. Then, we propose a hierarchical aligner that enables deep graph encoding, bridging the gap between MMAGs and MLLMs. Finally, we explore the inference choices, adapting MLLM to interleaved text and image generation in graph scenarios. Extensive experiments on three datasets from different domains demonstrate the effectiveness of our proposed method. Datasets and codes will be open-sourced upon acceptance.
Authors: Henry Hengyuan Zhao, Wenqi Pei, Yifei Tao, Haiyang Mei, Mike Zheng Shou
Abstract: Existing benchmarks do not test Large Multimodal Models (LMMs) on their interactive intelligence with human users, which is vital for developing general-purpose AI assistants. We design InterFeedback, an interactive framework, which can be applied to any LMM and dataset to assess this ability autonomously. On top of this, we introduce InterFeedback-Bench which evaluates interactive intelligence using two representative datasets, MMMU-Pro and MathVerse, to test 10 different open-source LMMs. Additionally, we present InterFeedback-Human, a newly collected dataset of 120 cases designed for manually testing interactive performance in leading models such as OpenAI-o1 and Claude-3.5-Sonnet. Our evaluation results indicate that even the state-of-the-art LMM, OpenAI-o1, struggles to refine its responses based on human feedback, achieving an average score of less than 50%. Our findings point to the need for methods that can enhance LMMs' capabilities to interpret and benefit from feedback.
Authors: Yilun Xu, Weili Nie, Arash Vahdat
Abstract: Sampling from diffusion models involves a slow iterative process that hinders their practical deployment, especially for interactive applications. To accelerate generation speed, recent approaches distill a multi-step diffusion model into a single-step student generator via variational score distillation, which matches the distribution of samples generated by the student to the teacher's distribution. However, these approaches use the reverse Kullback-Leibler (KL) divergence for distribution matching which is known to be mode seeking. In this paper, we generalize the distribution matching approach using a novel $f$-divergence minimization framework, termed $f$-distill, that covers different divergences with different trade-offs in terms of mode coverage and training variance. We derive the gradient of the $f$-divergence between the teacher and student distributions and show that it is expressed as the product of their score differences and a weighting function determined by their density ratio. This weighting function naturally emphasizes samples with higher density in the teacher distribution, when using a less mode-seeking divergence. We observe that the popular variational score distillation approach using the reverse-KL divergence is a special case within our framework. Empirically, we demonstrate that alternative $f$-divergences, such as forward-KL and Jensen-Shannon divergences, outperform the current best variational score distillation methods across image generation tasks. In particular, when using Jensen-Shannon divergence, $f$-distill achieves current state-of-the-art one-step generation performance on ImageNet64 and zero-shot text-to-image generation on MS-COCO. Project page: https://research.nvidia.com/labs/genair/f-distill
Authors: Hyeonjeong Ha, Qiusi Zhan, Jeonghwan Kim, Dimitrios Bralios, Saikrishna Sanniboina, Nanyun Peng, Kai-Wei Chang, Daniel Kang, Heng Ji
Abstract: Multimodal large language models (MLLMs) equipped with Retrieval Augmented Generation (RAG) leverage both their rich parametric knowledge and the dynamic, external knowledge to excel in tasks such as Question Answering. While RAG enhances MLLMs by grounding responses in query-relevant external knowledge, this reliance poses a critical yet underexplored safety risk: knowledge poisoning attacks, where misinformation or irrelevant knowledge is intentionally injected into external knowledge bases to manipulate model outputs to be incorrect and even harmful. To expose such vulnerabilities in multimodal RAG, we propose MM-PoisonRAG, a novel knowledge poisoning attack framework with two attack strategies: Localized Poisoning Attack (LPA), which injects query-specific misinformation in both text and images for targeted manipulation, and Globalized Poisoning Attack (GPA) to provide false guidance during MLLM generation to elicit nonsensical responses across all queries. We evaluate our attacks across multiple tasks, models, and access settings, demonstrating that LPA successfully manipulates the MLLM to generate attacker-controlled answers, with a success rate of up to 56% on MultiModalQA. Moreover, GPA completely disrupts model generation to 0% accuracy with just a single irrelevant knowledge injection. Our results highlight the urgent need for robust defenses against knowledge poisoning to safeguard multimodal RAG frameworks.
Authors: Adnan Iltaf, Rayan Merghani Ahmed, Zhenxi Zhang, Bin Li, Shoujun Zhou
Abstract: Medical image segmentation is crucial for clinical diagnosis and treatment planning, especially when dealing with complex anatomical structures such as vessels. However, accurately segmenting vessels remains challenging due to their small size, intricate edge structures, and susceptibility to artifacts and imaging noise. In this work, we propose VesselSAM, an enhanced version of the Segment Anything Model (SAM), specifically tailored for aortic vessel segmentation. VesselSAM incorporates AtrousLoRA, a novel module integrating Atrous Attention and Low-Rank Adaptation (LoRA), to enhance segmentation performance. Atrous Attention enables the model to capture multi-scale contextual information, preserving both fine-grained local details and broader global context. Additionally, LoRA facilitates efficient fine-tuning of the frozen SAM image encoder, reducing the number of trainable parameters and thereby enhancing computational efficiency. We evaluate VesselSAM using two challenging datasets: the Aortic Vessel Tree (AVT) dataset and the Type-B Aortic Dissection (TBAD) dataset. VesselSAM achieves state-of-the-art performance, attaining DSC scores of 93.50\%, 93.25\%, 93.02\%, and 93.26\% across multi-center datasets. Our results demonstrate that VesselSAM delivers high segmentation accuracy while significantly reducing computational overhead compared to existing large-scale models. This development paves the way for enhanced AI-based aortic vessel segmentation in clinical environments. The code and models will be released at https://github.com/Adnan-CAS/AtrousLora.
Authors: Guanzhou Ke, Shengfeng He, Xiao Li Wang, Bo Wang, Guoqing Chao, Yuanyang Zhang, Yi Xie, HeXing Su
Abstract: Previous successful approaches to missing modality completion rely on carefully designed fusion techniques and extensive pre-training on complete data, which can limit their generalizability in out-of-domain (OOD) scenarios. In this study, we pose a new challenge: can we develop a missing modality completion model that is both resource-efficient and robust to OOD generalization? To address this, we present a training-free framework for missing modality completion that leverages large multimodal models (LMMs). Our approach, termed the "Knowledge Bridger", is modality-agnostic and integrates generation and ranking of missing modalities. By defining domain-specific priors, our method automatically extracts structured information from available modalities to construct knowledge graphs. These extracted graphs connect the missing modality generation and ranking modules through the LMM, resulting in high-quality imputations of missing modalities. Experimental results across both general and medical domains show that our approach consistently outperforms competing methods, including in OOD generalization. Additionally, our knowledge-driven generation and ranking techniques demonstrate superiority over variants that directly employ LMMs for generation and ranking, offering insights that may be valuable for applications in other domains.
Authors: Nesryne Mejri, Enjie Ghorbel, Anis Kacem, Pavel Chernakov, Niki Foteinopoulou, Djamila Aouada
Abstract: This paper introduces the first fully unsupervised domain adaptation (UDA) framework for unsupervised anomaly detection (UAD). The performance of UAD techniques degrades significantly in the presence of a domain shift, difficult to avoid in a real-world setting. While UDA has contributed to solving this issue in binary and multi-class classification, such a strategy is ill-posed in UAD. This might be explained by the unsupervised nature of the two tasks, namely, domain adaptation and anomaly detection. Herein, we first formulate this problem that we call the two-fold unsupervised curse. Then, we propose a pioneering solution to this curse, considered intractable so far, by assuming that anomalies are rare. Specifically, we leverage clustering techniques to identify a dominant cluster in the target feature space. Posed as the normal cluster, the latter is aligned with the source normal features. Concretely, given a one-class source set and an unlabeled target set composed mostly of normal data and some anomalies, we fit the source features within a hypersphere while jointly aligning them with the features of the dominant cluster from the target set. The paper provides extensive experiments and analysis on common adaptation benchmarks for anomaly detection, demonstrating the relevance of both the newly introduced paradigm and the proposed approach. The code will be made publicly available.
Authors: Rohit Menon, Nils Dengler, Sicong Pan, Gokul Krishna Chenchani, Maren Bennewitz
Abstract: For scene understanding in unstructured environments, an accurate and uncertainty-aware metric-semantic mapping is required to enable informed action selection by autonomous systems. Existing mapping methods often suffer from overconfident semantic predictions, and sparse and noisy depth sensing, leading to inconsistent map representations. In this paper, we therefore introduce EvidMTL, a multi-task learning framework that uses evidential heads for depth estimation and semantic segmentation, enabling uncertainty-aware inference from monocular RGB images. To enable uncertainty-calibrated evidential multi-task learning, we propose a novel evidential depth loss function that jointly optimizes the belief strength of the depth prediction in conjunction with evidential segmentation loss. Building on this, we present EvidKimera, an uncertainty-aware semantic surface mapping framework, which uses evidential depth and semantics prediction for improved 3D metric-semantic consistency. We train and evaluate EvidMTL on the NYUDepthV2 and assess its zero-shot performance on ScanNetV2, demonstrating superior uncertainty estimation compared to conventional approaches while maintaining comparable depth estimation and semantic segmentation. In zero-shot mapping tests on ScanNetV2, EvidKimera outperforms Kimera in semantic surface mapping accuracy and consistency, highlighting the benefits of uncertainty-aware mapping and underscoring its potential for real-world robotic applications.
Authors: Hengguang Zhou, Xirui Li, Ruochen Wang, Minhao Cheng, Tianyi Zhou, Cho-Jui Hsieh
Abstract: Recently DeepSeek R1 demonstrated how reinforcement learning with simple rule-based incentives can enable autonomous development of complex reasoning in large language models, characterized by the "aha moment", in which the model manifest self-reflection and increased response length during training. However, attempts to extend this success to multimodal reasoning often failed to reproduce these key characteristics. In this report, we present the first successful replication of these emergent characteristics for multimodal reasoning on only a non-SFT 2B model. Starting with Qwen2-VL-2B and applying reinforcement learning directly on the SAT dataset, our model achieves 59.47% accuracy on CVBench, outperforming the base model by approximately ~30% and exceeding both SFT setting by ~2%. In addition, we share our failed attempts and insights in attempting to achieve R1-like reasoning using RL with instruct models. aiming to shed light on the challenges involved. Our key observations include: (1) applying RL on instruct model often results in trivial reasoning trajectories, and (2) naive length reward are ineffective in eliciting reasoning capabilities. The project code is available at https://github.com/turningpoint-ai/VisualThinker-R1-Zero
URLs: https://github.com/turningpoint-ai/VisualThinker-R1-Zero
Authors: Jiaxing Zhao, Xihan Wei, Liefeng Bo
Abstract: In this work, we present the first application of Reinforcement Learning with Verifiable Reward (RLVR) to an Omni-multimodal large language model in the context of emotion recognition, a task where both visual and audio modalities play crucial roles. We leverage RLVR to optimize the Omni model, significantly enhancing its performance in three key aspects: reasoning capability, emotion recognition accuracy, and generalization ability. The introduction of RLVR not only improves the model's overall performance on in-distribution data but also demonstrates superior robustness when evaluated on out-of-distribution datasets. More importantly, the improved reasoning capability enables clear analysis of the contributions of different modalities, particularly visual and audio information, in the emotion recognition process. This provides valuable insights into the optimization of multimodal large language models.