Authors: Boxun Hu, Mingze Xia, Ding Zhao, Guanlin Wu
Abstract: Dynamic urban environments, characterized by moving cameras and objects, pose significant challenges for camera trajectory estimation by complicating the distinction between camera-induced and object motion. We introduce MONA, a novel framework designed for robust moving object detection and segmentation from videos shot by dynamic cameras. MONA comprises two key modules: Dynamic Points Extraction, which leverages optical flow and tracking any point to identify dynamic points, and Moving Object Segmentation, which employs adaptive bounding box filtering, and the Segment Anything for precise moving object segmentation. We validate MONA by integrating with the camera trajectory estimation method LEAP-VO, and it achieves state-of-the-art results on the MPI Sintel dataset comparing to existing methods. These results demonstrate MONA's effectiveness for moving object detection and its potential in many other applications in the urban planning field.
Authors: Daeun Jung, Jaehyeok Jang, Sooyoung Jang, Yu Rang Park
Abstract: Computed tomography (CT) and clinical numeric data are essential modalities for cancer evaluation, but building large-scale multimodal training datasets for developing medical foundation models remains challenging due to the structural complexity of multi-slice CT data and high cost of expert annotation. In this study, we propose MEDFORM, a multimodal pre-training strategy that guides CT image representation learning using complementary information from clinical data for medical foundation model development. MEDFORM efficiently processes CT slice through multiple instance learning (MIL) and adopts a dual pre-training strategy: first pretraining the CT slice feature extractor using SimCLR-based self-supervised learning, then aligning CT and clinical modalities through cross-modal contrastive learning. Our model was pre-trained on three different cancer types: lung cancer (141,171 slices), breast cancer (8,100 slices), colorectal cancer (10,393 slices). The experimental results demonstrated that this dual pre-training strategy improves cancer classification performance and maintains robust performance in few-shot learning scenarios. Code available at https://github.com/DigitalHealthcareLab/25MultiModalFoundationModel.git
URLs: https://github.com/DigitalHealthcareLab/25MultiModalFoundationModel.git
Authors: Mahdi Alehdaghi, Rajarshi Bhattacharya, Pourya Shamsolmoali, Rafael M. O. Cruz, Eric Granger
Abstract: Visible-infrared person re-identification (VI-ReID) aims to match individuals across different camera modalities, a critical task in modern surveillance systems. While current VI-ReID methods focus on cross-modality matching, real-world applications often involve mixed galleries containing both V and I images, where state-of-the-art methods show significant performance limitations due to large domain shifts and low discrimination across mixed modalities. This is because gallery images from the same modality may have lower domain gaps but correspond to different identities. This paper introduces a novel mixed-modal ReID setting, where galleries contain data from both modalities. To address the domain shift among inter-modal and low discrimination capacity in intra-modal matching, we propose the Mixed Modality-Erased and -Related (MixER) method. The MixER learning approach disentangles modality-specific and modality-shared identity information through orthogonal decomposition, modality-confusion, and ID-modality-related objectives. MixER enhances feature robustness across modalities, improving cross-modal and mixed-modal settings performance. Our extensive experiments on the SYSU-MM01, RegDB and LLMC datasets indicate that our approach can provide state-of-the-art results using a single backbone, and showcase the flexibility of our approach in mixed gallery applications.
Authors: Xianrui Luo, Juewen Peng, Zhongang Cai, Lei Yang, Fan Yang, Zhiguo Cao, Guosheng Lin
Abstract: We introduce Deblur-Avatar, a novel framework for modeling high-fidelity, animatable 3D human avatars from motion-blurred monocular video inputs. Motion blur is prevalent in real-world dynamic video capture, especially due to human movements in 3D human avatar modeling. Existing methods either (1) assume sharp image inputs, failing to address the detail loss introduced by motion blur, or (2) mainly consider blur by camera movements, neglecting the human motion blur which is more common in animatable avatars. Our proposed approach integrates a human movement-based motion blur model into 3D Gaussian Splatting (3DGS). By explicitly modeling human motion trajectories during exposure time, we jointly optimize the trajectories and 3D Gaussians to reconstruct sharp, high-quality human avatars. We employ a pose-dependent fusion mechanism to distinguish moving body regions, optimizing both blurred and sharp areas effectively. Extensive experiments on synthetic and real-world datasets demonstrate that Deblur-Avatar significantly outperforms existing methods in rendering quality and quantitative metrics, producing sharp avatar reconstructions and enabling real-time rendering under challenging motion blur conditions.
Authors: Xuelong Dai, Dong Wang, Duan Mingxing, Bin Xiao
Abstract: Adversarial training and adversarial purification are two effective and practical defense methods to enhance a model's robustness against adversarial attacks. However, adversarial training necessitates additional training, while adversarial purification suffers from low time efficiency. More critically, current defenses are designed under the perturbation-based adversarial threat model, which is ineffective against the recently proposed unrestricted adversarial attacks. In this paper, we propose an effective and efficient adversarial defense method that counters both perturbation-based and unrestricted adversarial attacks. Our defense is inspired by the observation that adversarial attacks are typically located near the decision boundary and are sensitive to pixel changes. To address this, we introduce adversarial anti-aliasing to mitigate adversarial modifications. Additionally, we propose adversarial super-resolution, which leverages prior knowledge from clean datasets to benignly recover images. These approaches do not require additional training and are computationally efficient without calculating gradients. Extensive experiments against both perturbation-based and unrestricted adversarial attacks demonstrate that our defense method outperforms state-of-the-art adversarial purification methods.
Authors: Hao Fang, Xiaohang Sui, Hongyao Yu, 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: Taegyeong Lee, Jinsik Bang, Soyeong Kwon, Taehwan Kim
Abstract: Recent advancements in deep learning have significantly improved performance on computer vision tasks. Previous image classification methods primarily modify model architectures or add features, and they optimize models using cross-entropy loss on class logits. Since they focus on classifying images with considering class labels, these methods may struggle to learn various \emph{aspects} of classes (e.g., natural positions and shape changes). Rethinking the previous approach from a novel view, we propose a multi-aspect knowledge distillation method using Multimodal Large Language Models (MLLMs). Our approach involves: 1) querying Large Language Model with multi-aspect questions relevant to the knowledge we want to transfer to the model, 2) extracting corresponding logits from MLLM, and 3) expanding the model's output dimensions to distill these multi-aspect logits. We then apply cross-entropy loss to class logits and binary cross-entropy loss to multi-aspect logits. Through our method, the model can learn not only the knowledge about visual aspects but also the abstract and complex aspects that require a deeper understanding. We primarily apply our method to image classification, and to explore the potential for extending our model, we expand it to other tasks, such as object detection. In all experimental results, our method improves the performance of the baselines. Additionally, we analyze the effect of multi-aspect knowledge distillation. These results demonstrate that our method can transfer knowledge about various aspects to the model and the aspect knowledge can enhance model performance in computer vision tasks. This paper demonstrates the great potential of multi-aspect knowledge distillation, and we believe it offers a promising direction for future research in computer vision and beyond.
Authors: Changhui Deng, Lieyang Chen, Shinan Liu
Abstract: Detecting objects in urban traffic images presents considerable difficulties because of the following reasons: 1) These images are typically immense in size, encompassing millions or even hundreds of millions of pixels, yet computational resources are constrained. 2) The small size of vehicles in certain scenarios leads to insufficient information for accurate detection. 3) The uneven distribution of vehicles causes inefficient use of computational resources. To address these issues, we propose YOLOSCM (You Only Look Once with Segmentation Clustering Module), an efficient and effective framework. To address the challenges of large-scale images and the non-uniform distribution of vehicles, we propose a Segmentation Clustering Module (SCM). This module adaptively identifies clustered regions, enabling the model to focus on these areas for more precise detection. Additionally, we propose a new training strategy to optimize the detection of small vehicles and densely packed targets in complex urban traffic scenes. We perform extensive experiments on urban traffic datasets to demonstrate the effectiveness and superiority of our proposed approach.
Authors: Haohang Xu, Longyu Chen, Shuangrui Ding, Yilin Gao, Dongsheng Jiang, Yin Li, Shugong Xu, Junqing Yu, Wei Yang
Abstract: Diffusion-based generative models have achieved remarkable progress in visual content generation. However, traditional diffusion models directly denoise the entire image from noisy inputs, disregarding the hierarchical structure present in visual signals. This method is computationally intensive, especially for high-resolution image generation. Signal processing often leverages hierarchical decompositions; for instance, Fourier analysis decomposes signals by frequency, while wavelet analysis captures localized frequency components, reflecting both spatial and frequency information simultaneously. Inspired by these principles, we propose a multiscale diffusion framework that generates hierarchical visual representations, which are subsequently integrated to form the final output. The diffusion model target, whether raw RGB pixels or latent features from a Variational Autoencoder, s divided into multiple components that each capture distinct spatial levels. The low-resolution component contains the primary informative signal, while higher-resolution components add high-frequency details, such as texture. This approach divides image generation into two stages: producing a low-resolution base signal, followed by a high-resolution residual signal. Both stages can be effectively modeled using simpler, lightweight transformer architectures compared to full-resolution generation. This decomposition is conceptually similar to wavelet decomposition but offers a more streamlined and intuitive design. Our method, termed MSF(short for Multi-Scale Factorization), achieves an FID of 2.2 and an IS of 255.4 on the ImageNet 256x256 benchmark, reducing computational costs by 50% compared to baseline methods.
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: Yongxiang Liu, Weijie Li, Li Liu, Jie Zhou, Xuying Xiong, Bowen Peng, Yafei Song, Wei Yang, Tianpeng Liu, Zhen Liu, Xiang Li
Abstract: Synthetic Aperture Radar (SAR) stands as an indispensable sensor for Earth observation, owing to its unique capability for all-day imaging. Nevertheless, in a data-driven era, the scarcity of large-scale datasets poses a significant bottleneck to advancing SAR automatic target recognition (ATR) technology. This paper introduces NUDT4MSTAR, a large-scale SAR dataset for vehicle target recognition in the wild, including 40 target types and a wide array of imaging conditions across 5 different scenes. NUDT4MSTAR represents a significant leap forward in dataset scale, containing over 190,000 images-tenfold the size of its predecessors. To enhance the utility of this dataset, we meticulously annotate each image with detailed target information and imaging conditions. We also provide data in both processed magnitude images and original complex formats. Then, we construct a comprehensive benchmark consisting of 7 experiments with 15 recognition methods focusing on the stable and effective ATR issues. Besides, we conduct transfer learning experiments utilizing various models trained on NUDT4MSTAR and applied to three other target datasets, thereby demonstrating its substantial potential to the broader field of ground objects ATR. Finally, we discuss this dataset's application value and ATR's significant challenges. To the best of our knowledge, this work marks the first-ever endeavor to create a large-scale dataset benchmark for fine-grained SAR recognition in the wild, featuring an extensive collection of exhaustively annotated vehicle images. We expect that the open source of NUDT4MSTAR will facilitate the development of SAR ATR and attract a wider community of researchers.
Authors: Saleh Sakib Ahmed, Saifur Rahman Jony, Md. Toufikuzzaman, Saifullah Sayed, Rashed Uz Zzaman, Sara Nowreen, M. Sohel Rahman
Abstract: The use of Sentinel-2 images to compute Normalized Difference Water Index (NDWI) has many applications, including water body area detection. However, cloud cover poses significant challenges in this regard, which hampers the effectiveness of Sentinel-2 images in this context. In this paper, we present a deep learning model that can generate NDWI given Sentinel-1 images, thereby overcoming this cloud barrier. We show the effectiveness of our model, where it demonstrates a high accuracy of 0.9134 and an AUC of 0.8656 to predict the NDWI. Additionally, we observe promising results with an R2 score of 0.4984 (for regressing the NDWI values) and a Mean IoU of 0.4139 (for the underlying segmentation task). In conclusion, our model offers a first and robust solution for generating NDWI images directly from Sentinel-1 images and subsequent use for various applications even under challenging conditions such as cloud cover and nighttime.
Authors: Hao Shu
Abstract: Recent advancements have demonstrated the effectiveness of the extractor-selector (E-S) framework in edge detection (ED) tasks, which achieves state-of-the-art (SOTA) performance in both quantitative metrics and perceptual quality. However, this method still falls short of fully exploiting the potential of feature extractors, as selectors only operate on highly compressed feature maps that lack diversity and suffer from substantial information loss. Additionally, while union training can improve perceptual quality, the highest evaluation scores are typically obtained without it, creating a trade-off between quantitative accuracy and perceptual fidelity. To address these limitations, we propose an enhanced E-S architecture, which utilizes richer, less-loss feature representations and incorporates auxiliary features during the selection process, thereby improving the effectiveness of the feature selection mechanism. Additionally, we introduce a novel loss function, the Symmetrization Weight Binary Cross-Entropy (SWBCE), which simultaneously emphasizes both the recall of edge pixels and the suppression of erroneous edge predictions, thereby enhancing the predictions both in the perceptual quality and the prediction accuracy. The effectiveness and superiority of our approaches over baseline models, the standard E-S framework, and the standard Weight Binary Cross-Entropy (WBCE) loss function are demonstrated by extensive experiments. For example, our enhanced E-S architecture trained with SWBCE loss function achieves average improvements of 8.25$\%$, 8.01$\%$, and 33.25$\%$ in ODS, OIS, and AP, measured on BIPED2 compared with the baseline models, significantly outperforming the standard E-S method. The results set new benchmarks for ED tasks, and highlight the potential of the methods in beyond.
Authors: Yuzhuo Li, Di Zhao, Yihao Wu, Yun Sing Koh
Abstract: Identifying individual animals within large wildlife populations is essential for effective wildlife monitoring and conservation efforts. Recent advancements in computer vision have shown promise in animal re-identification (Animal ReID) by leveraging data from camera traps. However, existing methods rely exclusively on visual data, neglecting environmental metadata that ecologists have identified as highly correlated with animal behavior and identity, such as temperature and circadian rhythms. To bridge this gap, we propose the Meta-Feature Adapter (MFA), a lightweight module designed to integrate environmental metadata into vision-language foundation models, such as CLIP, to enhance Animal ReID performance. Our approach translates environmental metadata into natural language descriptions, encodes them into metadata-aware text embeddings, and incorporates these embeddings into image features through a cross-attention mechanism. Furthermore, we introduce a Gated Cross-Attention mechanism that dynamically adjusts the weights of metadata contributions, further improving performance. To validate our approach, we constructed the Metadata Augmented Animal Re-identification (MAAR) dataset, encompassing six species from New Zealand and featuring paired image data and environmental metadata. Extensive experiments demonstrate that MFA consistently improves Animal ReID performance across multiple baseline models.
Authors: Yipeng Liu, Qi Yang, Yujie Zhang, Yiling Xu, Le Yang, Zhu Li
Abstract: We present a novel quality assessment method which can predict the perceptual quality of point clouds from new scenes without available annotations by leveraging the rich prior knowledge in images, called the Distribution-Weighted Image-Transferred Point Cloud Quality Assessment (DWIT-PCQA). Recognizing the human visual system (HVS) as the decision-maker in quality assessment regardless of media types, we can emulate the evaluation criteria for human perception via neural networks and further transfer the capability of quality prediction from images to point clouds by leveraging the prior knowledge in the images. Specifically, domain adaptation (DA) can be leveraged to bridge the images and point clouds by aligning feature distributions of the two media in the same feature space. However, the different manifestations of distortions in images and point clouds make feature alignment a difficult task. To reduce the alignment difficulty and consider the different distortion distribution during alignment, we have derived formulas to decompose the optimization objective of the conventional DA into two suboptimization functions with distortion as a transition. Specifically, through network implementation, we propose the distortion-guided biased feature alignment which integrates existing/estimated distortion distribution into the adversarial DA framework, emphasizing common distortion patterns during feature alignment. Besides, we propose the quality-aware feature disentanglement to mitigate the destruction of the mapping from features to quality during alignment with biased distortions. Experimental results demonstrate that our proposed method exhibits reliable performance compared to general blind PCQA methods without needing point cloud annotations.
Authors: Arpit Garg, Cuong Nguyen, Rafael Felix, Yuyuan Liu, Thanh-Toan Do, Gustavo Carneiro
Abstract: Robust training with noisy labels is a critical challenge in image classification, offering the potential to reduce reliance on costly clean-label datasets. Real-world datasets often contain a mix of in-distribution (ID) and out-of-distribution (OOD) instance-dependent label noise, a challenge that is rarely addressed simultaneously by existing methods and is further compounded by the lack of comprehensive benchmarking datasets. Furthermore, even though current noisy-label learning approaches attempt to find noisy-label samples during training, these methods do not aim to estimate ID and OOD noise rates to promote their effectiveness in the selection of such noisy-label samples, and they are often represented by inefficient multi-stage learning algorithms. We propose the Adaptive Estimation of Instance-Dependent In-Distribution and Out-of-Distribution Label Noise (AEON) approach to address these research gaps. AEON is an efficient one-stage noisy-label learning methodology that dynamically estimates instance-dependent ID and OOD label noise rates to enhance robustness to complex noise settings. Additionally, we introduce a new benchmark reflecting real-world ID and OOD noise scenarios. Experiments demonstrate that AEON achieves state-of-the-art performance on both synthetic and real-world datasets
Authors: Minglong Dong, Dongliang Zhou, Jianghong Ma, Haijun Zhang
Abstract: Collocated clothing synthesis (CCS) has emerged as a pivotal topic in fashion technology, primarily concerned with the generation of a clothing item that harmoniously matches a given item. However, previous investigations have relied on using paired outfits, such as a pair of matching upper and lower clothing, to train a generative model for achieving this task. This reliance on the expertise of fashion professionals in the construction of such paired outfits has engendered a laborious and time-intensive process. In this paper, we introduce a new self-driven framework, named style- and texture-guided generative network (ST-Net), to synthesize collocated clothing without the necessity for paired outfits, leveraging self-supervised learning. ST-Net is designed to extrapolate fashion compatibility rules from the style and texture attributes of clothing, using a generative adversarial network. To facilitate the training and evaluation of our model, we have constructed a large-scale dataset specifically tailored for unsupervised CCS. Extensive experiments substantiate that our proposed method outperforms the state-of-the-art baselines in terms of both visual authenticity and fashion compatibility.
Authors: Priyanto Hidayatullah, Nurjannah Syakrani, Muhammad Rizqi Sholahuddin, Trisna Gelar, Refdinal Tubagus
Abstract: In the field of deep learning-based computer vision, YOLO is revolutionary. With respect to deep learning models, YOLO is also the one that is evolving the most rapidly. Unfortunately, not every YOLO model possesses scholarly publications. Moreover, there exists a YOLO model that lacks a publicly accessible official architectural diagram. Naturally, this engenders challenges, such as complicating the understanding of how the model operates in practice. Furthermore, the review articles that are presently available do not delve into the specifics of each model. The objective of this study is to present a comprehensive and in-depth architecture comparison of the four most recent YOLO models, specifically YOLOv8 through YOLO11, thereby enabling readers to quickly grasp not only how each model functions, but also the distinctions between them. To analyze each YOLO version's architecture, we meticulously examined the relevant academic papers, documentation, and scrutinized the source code. The analysis reveals that while each version of YOLO has improvements in architecture and feature extraction, certain blocks remain unchanged. The lack of scholarly publications and official diagrams presents challenges for understanding the model's functionality and future enhancement. Future developers are encouraged to provide these resources.
Authors: Guowei Yin, Sheng Huang, Luwen Huangfu, Yi Zhang, Xiaohong Zhang
Abstract: Feature quality is paramount for classification performance, particularly in few-shot scenarios. Contrastive learning, a widely adopted technique for enhancing feature quality, leverages sample relations to extract intrinsic features that capture semantic information and has achieved remarkable success in Few-Shot Learning (FSL). Nevertheless, current few-shot contrastive learning approaches often overlook the semantic similarity discrepancies at different granularities when employing the same modeling approach for different sample relations, which limits the potential of few-shot contrastive learning. In this paper, we introduce a straightforward yet effective contrastive learning approach, Multi-Grained Relation Contrastive Learning (MGRCL), as a pre-training feature learning model to boost few-shot learning by meticulously modeling sample relations at different granularities. MGRCL categorizes sample relations into three types: intra-sample relation of the same sample under different transformations, intra-class relation of homogenous samples, and inter-class relation of inhomogeneous samples. In MGRCL, we design Transformation Consistency Learning (TCL) to ensure the rigorous semantic consistency of a sample under different transformations by aligning predictions of input pairs. Furthermore, to preserve discriminative information, we employ Class Contrastive Learning (CCL) to ensure that a sample is always closer to its homogenous samples than its inhomogeneous ones, as homogenous samples share similar semantic content while inhomogeneous samples have different semantic content. Our method is assessed across four popular FSL benchmarks, showing that such a simple pre-training feature learning method surpasses a majority of leading FSL methods. Moreover, our method can be incorporated into other FSL methods as the pre-trained model and help them obtain significant performance gains.
Authors: Jinghan You, Yuanrui Sun, Mingyu Guo, Chao Feng, Jiao Ran
Abstract: Recently, large vision models have demonstrated powerful representation capabilities in the field of computer vision. However, we unexpectedly found that face recognition research is still mainly focused on CNN-based model architectures, which may lead to suboptimal state-of-the-art (SOTA) performance in face recognition. Therefore, we study how to use various loss functions from historical research orthogonally to train a new state-of-the-art face recognition model based on large vision models, called LVFace. On the largest public face database, WebFace42M, we demonstrated the superiority of LVFace over other advanced face recognition methods and achieved first place in the ICCV21 MFR-Ongoing challenge, until the submission of this work (December 30, 2024, academic track).
Authors: Shlok Mehendale, Jajati Keshari Sahoo, Rajendra Kumar Roul
Abstract: Parking space occupation detection using deep learning frameworks has seen significant advancements over the past few years. While these approaches effectively detect partial obstructions and adapt to varying lighting conditions, their performance significantly diminishes when haze is present. This paper proposes a novel hybrid model with a pre-trained feature extractor and a Pinball Generalized Twin Support Vector Machine (Pin-GTSVM) classifier, which removes the need for a dehazing system from the current State-of-The-Art hazy parking slot classification systems and is also insensitive to any atmospheric noise. The proposed system can seamlessly integrate with conventional smart parking infrastructures, leveraging a minimal number of cameras to monitor and manage hundreds of parking spaces efficiently. Its effectiveness has been evaluated against established parking space detection methods using the CNRPark Patches, PKLot, and a custom dataset specific to hazy parking scenarios. Furthermore, empirical results indicate a significant improvement in accuracy on a hazy parking system, thus emphasizing efficient atmospheric noise handling.
Authors: Jian Wang, Xiaokang Zhang, Xianping Ma, Weikang Yu, Pedram Ghamisi
Abstract: Weakly supervised landslide extraction aims to identify landslide regions from remote sensing data using models trained with weak labels, particularly image-level labels. However, it is often challenged by the imprecise boundaries of the extracted objects due to the lack of pixel-wise supervision and the properties of landslide objects. To tackle these issues, we propose a simple yet effective method by auto-prompting the Segment Anything Model (SAM), i.e., APSAM. Instead of depending on high-quality class activation maps (CAMs) for pseudo-labeling or fine-tuning SAM, our method directly yields fine-grained segmentation masks from SAM inference through prompt engineering. Specifically, it adaptively generates hybrid prompts from the CAMs obtained by an object localization network. To provide sufficient information for SAM prompting, an adaptive prompt generation (APG) algorithm is designed to fully leverage the visual patterns of CAMs, enabling the efficient generation of pseudo-masks for landslide extraction. These informative prompts are able to identify the extent of landslide areas (box prompts) and denote the centers of landslide objects (point prompts), guiding SAM in landslide segmentation. Experimental results on high-resolution aerial and satellite datasets demonstrate the effectiveness of our method, achieving improvements of at least 3.0\% in F1 score and 3.69\% in IoU compared to other state-of-the-art methods. The source codes and datasets will be available at https://github.com/zxk688.
Authors: Samer Attrah
Abstract: Emotion estimation in general is a field that has been studied for a long time, and several approaches exist using machine learning. in this paper, we present an LSTM model, that processes the blend-shapes produced by the library MediaPipe, for a face detected in a live stream of a camera, to estimate the main emotion from the facial expressions, this model is trained on the FER2013 dataset and delivers a result of 71% accuracy and 62% f1-score which meets the accuracy benchmark of the FER2013 dataset, with significantly reduced computation costs. https://github.com/ Samir-atra/Emotion_estimation_from_video_footage_with_LSTM_ML_algorithm
URLs: https://github.com/
Authors: Jiaxin Chen, Miao Hu, Dengyong Zhang, Jingyang Meng
Abstract: The rapid development of Deepfake technology has enabled the generation of highly realistic manipulated videos, posing severe social and ethical challenges. Existing Deepfake detection methods primarily focused on either spatial or temporal inconsistencies, often neglecting the interplay between the two or suffering from interference caused by natural facial motions. To address these challenges, we propose the global context consistency flow (GC-ConsFlow), a novel dual-stream framework that effectively integrates spatial and temporal features for robust Deepfake detection. The global grouped context aggregation module (GGCA), integrated into the global context-aware frame flow stream (GCAF), enhances spatial feature extraction by aggregating grouped global context information, enabling the detection of subtle, spatial artifacts within frames. The flow-gradient temporal consistency stream (FGTC), rather than directly modeling the residuals, it is used to improve the robustness of temporal feature extraction against the inconsistency introduced by unnatural facial motion using optical flow residuals and gradient-based features. By combining these two streams, GC-ConsFlow demonstrates the effectiveness and robustness in capturing complementary spatiotemporal forgery traces. Extensive experiments show that GC-ConsFlow outperforms existing state-of-the-art methods in detecting Deepfake videos under various compression scenarios.
Authors: Deepak Ghimire, Dayoung Kil, Seonghwan Jeong, Jaesik Park, Seong-heum Kim
Abstract: Existing structured pruning typically involves multi-stage training procedures that often demand heavy computation. Pruning at initialization, which aims to address this limitation, reduces training costs but struggles with performance. To address these challenges, we propose an efficient framework for one-cycle structured pruning without compromising model performance. In this approach, we integrate pre-training, pruning, and fine-tuning into a single training cycle, referred to as the `one cycle approach'. The core idea is to search for the optimal sub-network during the early stages of network training, guided by norm-based group saliency criteria and structured sparsity regularization. We introduce a novel pruning indicator that determines the stable pruning epoch by assessing the similarity between evolving pruning sub-networks across consecutive training epochs. Also, group sparsity regularization helps to accelerate the pruning process and results in speeding up the entire process. Extensive experiments on datasets, including CIFAR-10/100, and ImageNet, using VGGNet, ResNet, MobileNet, and ViT architectures, demonstrate that our method achieves state-of-the-art accuracy while being one of the most efficient pruning frameworks in terms of training time. The source code will be made publicly available.
Authors: Wooseok Song, Seunggyu Chang, Jaejun Yoo
Abstract: While single-concept customization has been studied in 3D, multi-concept customization remains largely unexplored. To address this, we propose MultiDreamer3D that can generate coherent multi-concept 3D content in a divide-and-conquer manner. First, we generate 3D bounding boxes using an LLM-based layout controller. Next, a selective point cloud generator creates coarse point clouds for each concept. These point clouds are placed in the 3D bounding boxes and initialized into 3D Gaussian Splatting with concept labels, enabling precise identification of concept attributions in 2D projections. Finally, we refine 3D Gaussians via concept-aware interval score matching, guided by concept-aware diffusion. Our experimental results show that MultiDreamer3D not only ensures object presence and preserves the distinct identities of each concept but also successfully handles complex cases such as property change or interaction. To the best of our knowledge, we are the first to address the multi-concept customization in 3D.
Authors: Jiangchuan Wei, Shiyue Yan, Wenfeng Lin, Boyuan Liu, Renjie Chen, Mingyu Guo
Abstract: Recent advancements in video generation have significantly impacted various downstream applications, particularly in identity-preserving video generation (IPT2V). However, existing methods struggle with "copy-paste" artifacts and low similarity issues, primarily due to their reliance on low-level facial image information. This dependence can result in rigid facial appearances and artifacts reflecting irrelevant details. To address these challenges, we propose EchoVideo, which employs two key strategies: (1) an Identity Image-Text Fusion Module (IITF) that integrates high-level semantic features from text, capturing clean facial identity representations while discarding occlusions, poses, and lighting variations to avoid the introduction of artifacts; (2) a two-stage training strategy, incorporating a stochastic method in the second phase to randomly utilize shallow facial information. The objective is to balance the enhancements in fidelity provided by shallow features while mitigating excessive reliance on them. This strategy encourages the model to utilize high-level features during training, ultimately fostering a more robust representation of facial identities. EchoVideo effectively preserves facial identities and maintains full-body integrity. Extensive experiments demonstrate that it achieves excellent results in generating high-quality, controllability and fidelity videos.
Authors: Haomiao Xiong, Zongxin Yang, Jiazuo Yu, Yunzhi Zhuge, Lu Zhang, Jiawen Zhu, Huchuan Lu
Abstract: Recent advances in Large Language Models (LLMs) have enabled the development of Video-LLMs, advancing multimodal learning by bridging video data with language tasks. However, current video understanding models struggle with processing long video sequences, supporting multi-turn dialogues, and adapting to real-world dynamic scenarios. To address these issues, we propose StreamChat, a training-free framework for streaming video reasoning and conversational interaction. $\StreamChat$ leverages a novel hierarchical memory system to efficiently process and compress video features over extended sequences, enabling real-time, multi-turn dialogue. Our framework incorporates a parallel system scheduling strategy that enhances processing speed and reduces latency, ensuring robust performance in real-world applications. Furthermore, we introduce StreamBench, a versatile benchmark that evaluates streaming video understanding across diverse media types and interactive scenarios, including multi-turn interactions and complex reasoning tasks. Extensive evaluations on StreamBench and other public benchmarks demonstrate that StreamChat significantly outperforms existing state-of-the-art models in terms of accuracy and response times, confirming its effectiveness for streaming video understanding. Code is available at StreamChat: https://github.com/hmxiong/StreamChat.
Authors: Yuliang Gu, Weilun Tsao, Bo Du, Thierry G\'eraud, Yongchao Xu
Abstract: Annotating 3D medical images demands substantial time and expertise, driving the adoption of semi-supervised learning (SSL) for segmentation tasks. However, the complex anatomical structures of organs often lead to significant class imbalances, posing major challenges for deploying SSL in real-world scenarios. Despite the availability of valuable prior information, such as inter-organ relative positions and organ shape priors, existing SSL methods have yet to fully leverage these insights. To address this gap, we propose a novel approach that integrates textual anatomical knowledge (TAK) into the segmentation model. Specifically, we use GPT-4o to generate textual descriptions of anatomical priors, which are then encoded using a CLIP-based model. These encoded priors are injected into the segmentation model as parameters of the segmentation head. Additionally, contrastive learning is employed to enhance the alignment between textual priors and visual features. Extensive experiments demonstrate the superior performance of our method, significantly surpassing state-of-the-art approaches. The source code will be available at: https://github.com/Lunn88/TAK-Semi.
Authors: JiaXin Chen, Miao Hu, DengYong Zhang, Yun Song, Xin Liao
Abstract: With the rapid advancement of generative models, the visual quality of generated images has become nearly indistinguishable from the real ones, posing challenges to content authenticity verification. Existing methods for detecting AI-generated images primarily focus on specific forgery clues, which are often tailored to particular generative models like GANs or diffusion models. These approaches struggle to generalize across architectures. Building on the observation that generative images often exhibit local anomalies, such as excessive smoothness, blurred textures, and unnatural pixel variations in small regions, we propose the localized discrepancy representation network (LDR-Net), a novel approach for detecting AI-generated images. LDR-Net captures smoothing artifacts and texture irregularities, which are common but often overlooked. It integrates two complementary modules: local gradient autocorrelation (LGA) which models local smoothing anomalies to detect smoothing anomalies, and local variation pattern (LVP) which captures unnatural regularities by modeling the complexity of image patterns. By merging LGA and LVP features, a comprehensive representation of localized discrepancies can be provided. Extensive experiments demonstrate that our LDR-Net achieves state-of-the-art performance in detecting generated images and exhibits satisfactory generalization across unseen generative models. The code will be released upon acceptance of this paper.
Authors: Xuerui Qiu, Jieyuan Zhang, Wenjie Wei, Honglin Cao, Junsheng Guo, Rui-Jie Zhu, Yimeng Shan, Yang Yang, Malu Zhang, Haizhou Li
Abstract: Spiking neural networks are emerging as a promising energy-efficient alternative to traditional artificial neural networks due to their spike-driven paradigm. However, recent research in the SNN domain has mainly focused on enhancing accuracy by designing large-scale Transformer structures, which typically rely on substantial computational resources, limiting their deployment on resource-constrained devices. To overcome this challenge, we propose a quantized spike-driven Transformer baseline (QSD-Transformer), which achieves reduced resource demands by utilizing a low bit-width parameter. Regrettably, the QSD-Transformer often suffers from severe performance degradation. In this paper, we first conduct empirical analysis and find that the bimodal distribution of quantized spike-driven self-attention (Q-SDSA) leads to spike information distortion (SID) during quantization, causing significant performance degradation. To mitigate this issue, we take inspiration from mutual information entropy and propose a bi-level optimization strategy to rectify the information distribution in Q-SDSA. Specifically, at the lower level, we introduce an information-enhanced LIF to rectify the information distribution in Q-SDSA. At the upper level, we propose a fine-grained distillation scheme for the QSD-Transformer to align the distribution in Q-SDSA with that in the counterpart ANN. By integrating the bi-level optimization strategy, the QSD-Transformer can attain enhanced energy efficiency without sacrificing its high-performance advantage.For instance, when compared to the prior SNN benchmark on ImageNet, the QSD-Transformer achieves 80.3\% top-1 accuracy, accompanied by significant reductions of 6.0$\times$ and 8.1$\times$ in power consumption and model size, respectively. Code is available at https://github.com/bollossom/QSD-Transformer.
Authors: Zicheng Pan, Xiaohan Yu, Weichuan Zhang, Yongsheng Gao
Abstract: Source-free active domain adaptation (SFADA) addresses the challenge of adapting a pre-trained model to new domains without access to source data while minimizing the need for target domain annotations. This scenario is particularly relevant in real-world applications where data privacy, storage limitations, or labeling costs are significant concerns. Key challenges in SFADA include selecting the most informative samples from the target domain for labeling, effectively leveraging both labeled and unlabeled target data, and adapting the model without relying on source domain information. Additionally, existing methods often struggle with noisy or outlier samples and may require impractical progressive labeling during training. To effectively select more informative samples without frequently requesting human annotations, we propose the Propensity-driven Uncertainty Learning (ProULearn) framework. ProULearn utilizes a novel homogeneity propensity estimation mechanism combined with correlation index calculation to evaluate feature-level relationships. This approach enables the identification of representative and challenging samples while avoiding noisy outliers. Additionally, we develop a central correlation loss to refine pseudo-labels and create compact class distributions during adaptation. In this way, ProULearn effectively bridges the domain gap and maximizes adaptation performance. The principles of informative sample selection underlying ProULearn have broad implications beyond SFADA, offering benefits across various deep learning tasks where identifying key data points or features is crucial. Extensive experiments on four benchmark datasets demonstrate that ProULearn outperforms state-of-the-art methods in domain adaptation scenarios.
Authors: Manuel Benavent-Lledo, David Mulero-P\'erez, David Ortiz-Perez, Jose Garcia-Rodriguez
Abstract: Detecting actions as they occur is essential for applications like video surveillance, autonomous driving, and human-robot interaction. Known as online action detection, this task requires classifying actions in streaming videos, handling background noise, and coping with incomplete actions. Transformer architectures are the current state-of-the-art, yet the potential of recent advancements in computer vision, particularly vision-language models (VLMs), remains largely untapped for this problem, partly due to high computational costs. In this paper, we introduce TOAD: a Text-driven Online Action Detection architecture that supports zero-shot and few-shot learning. TOAD leverages CLIP (Contrastive Language-Image Pretraining) textual embeddings, enabling efficient use of VLMs without significant computational overhead. Our model achieves 82.46% mAP on the THUMOS14 dataset, outperforming existing methods, and sets new baselines for zero-shot and few-shot performance on the THUMOS14 and TVSeries datasets.
Authors: Wenzhuo Ma, Zhenzhong Chen
Abstract: Recently, foundational diffusion models have attracted considerable attention in image compression tasks, whereas their application to video compression remains largely unexplored. In this article, we introduce DiffVC, a diffusion-based perceptual neural video compression framework that effectively integrates foundational diffusion model with the video conditional coding paradigm. This framework uses temporal context from previously decoded frame and the reconstructed latent representation of the current frame to guide the diffusion model in generating high-quality results. To accelerate the iterative inference process of diffusion model, we propose the Temporal Diffusion Information Reuse (TDIR) strategy, which significantly enhances inference efficiency with minimal performance loss by reusing the diffusion information from previous frames. Additionally, to address the challenges posed by distortion differences across various bitrates, we propose the Quantization Parameter-based Prompting (QPP) mechanism, which utilizes quantization parameters as prompts fed into the foundational diffusion model to explicitly modulate intermediate features, thereby enabling a robust variable bitrate diffusion-based neural compression framework. Experimental results demonstrate that our proposed solution delivers excellent performance in both perception metrics and visual quality.
Authors: Wailing Tang, Biqi Yang, Pheng-Ann Heng, Yun-Hui Liu, Chi-Wing Fu
Abstract: Few-shot Semantic Segmentation (FSS) is a challenging task that utilizes limited support images to segment associated unseen objects in query images. However, recent FSS methods are observed to perform worse, when enlarging the number of shots. As the support set enlarges, existing FSS networks struggle to concentrate on the high-contributed supports and could easily be overwhelmed by the low-contributed supports that could severely impair the mask predictions. In this work, we study this challenging issue, called support dilution, our goal is to recognize, select, preserve, and enhance those high-contributed supports in the raw support pool. Technically, our method contains three novel parts. First, we propose a contribution index, to quantitatively estimate if a high-contributed support dilutes. Second, we develop the Symmetric Correlation (SC) module to preserve and enhance the high-contributed support features, minimizing the distraction by the low-contributed features. Third, we design the Support Image Pruning operation, to retrieve a compact and high quality subset by discarding low-contributed supports. We conduct extensive experiments on two FSS benchmarks, COCO-20i and PASCAL-5i, the segmentation results demonstrate the compelling performance of our solution over state-of-the-art FSS approaches. Besides, we apply our solution for online segmentation and real-world segmentation, convincing segmentation results showing the practical ability of our work for real-world demonstrations.
Authors: Jianxin Liang, Xiaojun Meng, Huishuai Zhang, Yueqian Wang, Jiansheng Wei, Dongyan Zhao
Abstract: Video Question Answering (VideoQA) is a challenging task that requires understanding complex visual and temporal relationships within videos to answer questions accurately. In this work, we introduce \textbf{ReasVQA} (Reasoning-enhanced Video Question Answering), a novel approach that leverages reasoning processes generated by Multimodal Large Language Models (MLLMs) to improve the performance of VideoQA models. Our approach consists of three phases: reasoning generation, reasoning refinement, and learning from reasoning. First, we generate detailed reasoning processes using additional MLLMs, and second refine them via a filtering step to ensure data quality. Finally, we use the reasoning data, which might be in an imperfect form, to guide the VideoQA model via multi-task learning, on how to interpret and answer questions based on a given video. We evaluate ReasVQA on three popular benchmarks, and our results establish new state-of-the-art performance with significant improvements of +2.9 on NExT-QA, +7.3 on STAR, and +5.9 on IntentQA. Our findings demonstrate the supervising benefits of integrating reasoning processes into VideoQA. Further studies validate each component of our method, also with different backbones and MLLMs, and again highlight the advantages of this simple but effective method. We offer a new perspective on enhancing VideoQA performance by utilizing advanced reasoning techniques, setting a new benchmark in this research field.
Authors: Tao Liu, Kai Wang, Senmao Li, Joost van de Weijer, Fahad Shahbaz Khan, Shiqi Yang, Yaxing Wang, Jian Yang, Ming-Ming Cheng
Abstract: Text-to-image generation models can create high-quality images from input prompts. However, they struggle to support the consistent generation of identity-preserving requirements for storytelling. Existing approaches to this problem typically require extensive training in large datasets or additional modifications to the original model architectures. This limits their applicability across different domains and diverse diffusion model configurations. In this paper, we first observe the inherent capability of language models, coined context consistency, to comprehend identity through context with a single prompt. Drawing inspiration from the inherent context consistency, we propose a novel training-free method for consistent text-to-image (T2I) generation, termed "One-Prompt-One-Story" (1Prompt1Story). Our approach 1Prompt1Story concatenates all prompts into a single input for T2I diffusion models, initially preserving character identities. We then refine the generation process using two novel techniques: Singular-Value Reweighting and Identity-Preserving Cross-Attention, ensuring better alignment with the input description for each frame. In our experiments, we compare our method against various existing consistent T2I generation approaches to demonstrate its effectiveness through quantitative metrics and qualitative assessments. Code is available at https://github.com/byliutao/1Prompt1Story.
Authors: Francesco Di Sario, Riccardo Renzulli, Marco Grangetto, Akihiro Sugimoto, Enzo Tartaglione
Abstract: 3D Gaussian Splatting enhances real-time performance in novel view synthesis by representing scenes with mixtures of Gaussians and utilizing differentiable rasterization. However, it typically requires large storage capacity and high VRAM, demanding the design of effective pruning and compression techniques. Existing methods, while effective in some scenarios, struggle with scalability and fail to adapt models based on critical factors such as computing capabilities or bandwidth, requiring to re-train the model under different configurations. In this work, we propose a novel, model-agnostic technique that organizes Gaussians into several hierarchical layers, enabling progressive Level of Detail (LoD) strategy. This method, combined with recent approach of compression of 3DGS, allows a single model to instantly scale across several compression ratios, with minimal to none impact to quality compared to a single non-scalable model and without requiring re-training. We validate our approach on typical datasets and benchmarks, showcasing low distortion and substantial gains in terms of scalability and adaptability.
Authors: Lu Wang, Tianyuan Zhang, Yang Qu, Siyuan Liang, Yuwei Chen, Aishan Liu, Xianglong Liu, Dacheng Tao
Abstract: Vision-language models (VLMs) have significantly advanced autonomous driving (AD) by enhancing reasoning capabilities; however, these models remain highly susceptible to adversarial attacks. While existing research has explored white-box attacks to some extent, the more practical and challenging black-box scenarios remain largely underexplored due to their inherent difficulty. In this paper, we take the first step toward designing black-box adversarial attacks specifically targeting VLMs in AD. We identify two key challenges for achieving effective black-box attacks in this context: the effectiveness across driving reasoning chains in AD systems and the dynamic nature of driving scenarios. To address this, we propose Cascading Adversarial Disruption (CAD). It first introduces Decision Chain Disruption, which targets low-level reasoning breakdown by generating and injecting deceptive semantics, ensuring the perturbations remain effective across the entire decision-making chain. Building on this, we present Risky Scene Induction, which addresses dynamic adaptation by leveraging a surrogate VLM to understand and construct high-level risky scenarios that are likely to result in critical errors in the current driving contexts. Extensive experiments conducted on multiple AD VLMs and benchmarks demonstrate that CAD achieves state-of-the-art attack effectiveness, significantly outperforming existing methods (+13.43% on average). Moreover, we validate its practical applicability through real-world attacks on AD vehicles powered by VLMs, where the route completion rate drops by 61.11% and the vehicle crashes directly into the obstacle vehicle with adversarial patches. Finally, we release CADA dataset, comprising 18,808 adversarial visual-question-answer pairs, to facilitate further evaluation and research in this critical domain. Our codes and dataset will be available after paper's acceptance.
Authors: Mohit Vaishnav, Tanel Tammet
Abstract: Evaluating the reasoning capabilities of Vision-Language Models (VLMs) in complex visual tasks provides valuable insights into their potential and limitations. In this work, we assess the performance of VLMs on the challenging Bongard Openworld Problems benchmark, which involves reasoning over natural images. We propose and evaluate three human-inspired paradigms: holistic analysis (global context processing), deductive rule learning (explicit rule derivation and application), and componential analysis (structured decomposition of images into components). Our results demonstrate that state-of-the-art models, including GPT-4o and Gemini, not only surpass human benchmarks but also excel in structured reasoning tasks, with componential analysis proving especially effective. However, ablation studies reveal key challenges, such as handling synthetic images, making fine-grained distinctions, and interpreting nuanced contextual information. These insights underscore the need for further advancements in model robustness and generalization, while highlighting the transformative potential of structured reasoning approaches in enhancing VLM capabilities.
Authors: Yinglong Li, Xiaoyu Liu, Jiacheng Li, Ruikang Xu, Yinda Chen, Zhiwei Xiong
Abstract: State Space Models (SSMs), as key components of Mamaba, have gained increasing attention for vision models recently, thanks to their efficient long sequence modeling capability. Given the computational cost of deploying SSMs on resource-limited edge devices, Post-Training Quantization (PTQ) is a technique with the potential for efficient deployment of SSMs. In this work, we propose QMamba, one of the first PTQ frameworks to our knowledge, designed for vision SSMs based on the analysis of the activation distributions in SSMs. We reveal that the distribution of discrete parameters exhibits long-tailed skewness and the distribution of the hidden state sequence exhibits highly dynamic variations. Correspondingly, we design Long-tailed Skewness Quantization (LtSQ) to quantize discrete parameters and Temporal Group Quantization (TGQ) to quantize hidden states, which reduces the quantization errors. Extensive experiments demonstrate that QMamba outperforms advanced PTQ methods on vision models across multiple model sizes and architectures. Notably, QMamba surpasses existing methods by 21.0% on ImageNet classification with 4-bit activations.
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: Potito Aghilar, Vito Walter Anelli, Michelantonio Trizio, Tommaso Di Noia
Abstract: Recent advancements in diffusion models have significantly broadened the possibilities for editing images of real-world objects. However, performing non-rigid transformations, such as changing the pose of objects or image-based conditioning, remains challenging. Maintaining object identity during these edits is difficult, and current methods often fall short of the precision needed for industrial applications, where consistency is critical. Additionally, fine-tuning diffusion models requires custom training data, which is not always accessible in real-world scenarios. This work introduces FashionRepose, a training-free pipeline for non-rigid pose editing specifically designed for the fashion industry. The approach integrates off-the-shelf models to adjust poses of long-sleeve garments, maintaining identity and branding attributes. FashionRepose uses a zero-shot approach to perform these edits in near real-time, eliminating the need for specialized training. consistent image editing. The solution holds potential for applications in the fashion industry and other fields demanding identity preservation in image editing.
Authors: Pengteng Li, Yunfan Lu, Pinghao Song, Wuyang Li, Huizai Yao, Hui Xiong
Abstract: The event-based Vision-Language Model (VLM) recently has made good progress for practical vision tasks. However, most of these works just utilize CLIP for focusing on traditional perception tasks, which obstruct model understanding explicitly the sufficient semantics and context from event streams. To address the deficiency, we propose EventVL, the first generative event-based MLLM (Multimodal Large Language Model) framework for explicit semantic understanding. Specifically, to bridge the data gap for connecting different modalities semantics, we first annotate a large event-image/video-text dataset, containing almost 1.4 million high-quality pairs of data, which enables effective learning across various scenes, e.g., drive scene or human motion. After that, we design Event Spatiotemporal Representation to fully explore the comprehensive information by diversely aggregating and segmenting the event stream. To further promote a compact semantic space, Dynamic Semantic Alignment is introduced to improve and complete sparse semantic spaces of events. Extensive experiments show that our EventVL can significantly surpass existing MLLM baselines in event captioning and scene description generation tasks. We hope our research could contribute to the development of the event vision community.
Authors: Abdulrahman Oladipupo Ibraheem
Abstract: I introduce two novel loss functions for classification in deep learning. The two loss functions extend standard cross entropy loss by regularizing it with minimum entropy and Kullback-Leibler (K-L) divergence terms. The first of the two novel loss functions is termed mixed entropy loss (MIX-ENT for short), while the second one is termed minimum entropy regularized cross-entropy loss (MIN-ENT for short). The MIX-ENT function introduces a regularizer that can be shown to be equivalent to the sum of a minimum entropy term and a K-L divergence term. However, it should be noted that the K-L divergence term here is different from that in the standard cross-entropy loss function, in the sense that it swaps the roles of the target probability and the hypothesis probability. The MIN-ENT function simply adds a minimum entropy regularizer to the standard cross entropy loss function. In both MIX-ENT and MIN-ENT, the minimum entropy regularizer minimizes the entropy of the hypothesis probability distribution which is output by the neural network. Experiments on the EMNIST-Letters dataset shows that my implementation of MIX-ENT and MIN-ENT lets the VGG model climb from its previous 3rd position on the paperswithcode leaderboard to reach the 2nd position on the leaderboard, outperforming the Spinal-VGG model in so doing. Specifically, using standard cross-entropy, VGG achieves 95.86% while Spinal-VGG achieves 95.88% classification accuracies, whereas using VGG (without Spinal-VGG) our MIN-ENT achieved 95.933%, while our MIX-ENT achieved 95.927% accuracies. The pre-trained models for both MIX-ENT and MIN-ENT are at https://github.com/rahmanoladi/minimum entropy project.
Authors: I\~naki Erregue, Kamal Nasrollahi, Sergio Escalera
Abstract: We introduce YOLO11-JDE, a fast and accurate multi-object tracking (MOT) solution that combines real-time object detection with self-supervised Re-Identification (Re-ID). By incorporating a dedicated Re-ID branch into YOLO11s, our model performs Joint Detection and Embedding (JDE), generating appearance features for each detection. The Re-ID branch is trained in a fully self-supervised setting while simultaneously training for detection, eliminating the need for costly identity-labeled datasets. The triplet loss, with hard positive and semi-hard negative mining strategies, is used for learning discriminative embeddings. Data association is enhanced with a custom tracking implementation that successfully integrates motion, appearance, and location cues. YOLO11-JDE achieves competitive results on MOT17 and MOT20 benchmarks, surpassing existing JDE methods in terms of FPS and using up to ten times fewer parameters. Thus, making our method a highly attractive solution for real-world applications.
Authors: Fahud Ahmmed, Md. Zaheer Raihan, Kamnur Nahar, D. M. Asadujjaman, Md. Mahfujur Rahman, Abdullah Tamim
Abstract: Skin diseases can arise from infections, allergies, genetic factors, autoimmune disorders, hormonal imbalances, or environmental triggers such as sun damage and pollution. Some skin diseases, such as Actinic Keratosis and Psoriasis, can be fatal if not treated in time. Early identification is crucial, but the diagnostic methods for these conditions are often expensive and not widely accessible. In this study, we propose a novel and efficient method for diagnosing skin diseases using deep learning techniques. This approach employs a modified VGG16 Convolutional Neural Network (CNN) model. The model includes several convolutional layers and utilizes ImageNet weights with modified top layers. The top layer is updated with fully connected layers and a final softmax activation layer to classify skin diseases. The dataset used, titled "Skin Disease Dataset," is publicly available. While the VGG16 architecture does not include data augmentation by default, preprocessing techniques such as rotation, shifting, and zooming were applied to augment the data prior to model training. The proposed methodology achieved 90.67% accuracy using the modified VGG16 model, demonstrating its reliability in classifying skin diseases. The promising results highlight the potential of this approach for real-world applications.
Authors: Dario Serez, Marco Cristani, Alessio Del Bue, Vittorio Murino, Pietro Morerio
Abstract: In image generation, Multiple Latent Variable Generative Models (MLVGMs) employ multiple latent variables to gradually shape the final images, from global characteristics to finer and local details (e.g., StyleGAN, NVAE), emerging as powerful tools for diverse applications. Yet their generative dynamics and latent variable utilization remain only empirically observed. In this work, we propose a novel framework to systematically quantify the impact of each latent variable in MLVGMs, using Mutual Information (MI) as a guiding metric. Our analysis reveals underutilized variables and can guide the use of MLVGMs in downstream applications. With this foundation, we introduce a method for generating synthetic data for Self-Supervised Contrastive Representation Learning (SSCRL). By leveraging the hierarchical and disentangled variables of MLVGMs, and guided by the previous analysis, we apply tailored latent perturbations to produce diverse views for SSCRL, without relying on real data altogether. Additionally, we introduce a Continuous Sampling (CS) strategy, where the generator dynamically creates new samples during SSCRL training, greatly increasing data variability. Our comprehensive experiments demonstrate the effectiveness of these contributions, showing that MLVGMs' generated views compete on par with or even surpass views generated from real data. This work establishes a principled approach to understanding and exploiting MLVGMs, advancing both generative modeling and self-supervised learning.
Authors: Timothy Chase Jr, Christopher Wilson, Karthik Dantu
Abstract: The in-situ detection of planetary, lunar, and small-body surface terrain is crucial for autonomous spacecraft applications, where learning-based computer vision methods are increasingly employed to enable intelligence without prior information or human intervention. However, many of these methods remain computationally expensive for spacecraft processors and prevent real-time operation. Training of such algorithms is additionally complex due to the scarcity of labeled data and reliance on supervised learning approaches. Unsupervised Domain Adaptation (UDA) offers a promising solution by facilitating model training with disparate data sources such as simulations or synthetic scenes, although UDA is difficult to apply to celestial environments where challenging feature spaces are paramount. To alleviate such issues, You Only Crash Once (YOCOv1) has studied the integration of Visual Similarity-based Alignment (VSA) into lightweight one-stage object detection architectures to improve space terrain UDA. Although proven effective, the approach faces notable limitations, including performance degradations in multi-class and high-altitude scenarios. Building upon the foundation of YOCOv1, we propose novel additions to the VSA scheme that enhance terrain detection capabilities under UDA, and our approach is evaluated across both simulated and real-world data. Our second YOCO rendition, YOCOv2, is capable of achieving state-of-the-art UDA performance on surface terrain detection, where we showcase improvements upwards of 31% compared with YOCOv1 and terrestrial state-of-the-art. We demonstrate the practical utility of YOCOv2 with spacecraft flight hardware performance benchmarking and qualitative evaluation of NASA mission data.
Authors: Ziheng Wang, Toni Lassila, Sharib Ali
Abstract: In real-world data, long-tailed data distribution is common, making it challenging for models trained on empirical risk minimisation to learn and classify tail classes effectively. While many studies have sought to improve long tail recognition by altering the data distribution in the feature space and adjusting model decision boundaries, research on the synergy and corrective approach among various methods is limited. Our study delves into three long-tail recognition techniques: Supervised Contrastive Learning (SCL), Rare-Class Sample Generator (RSG), and Label-Distribution-Aware Margin Loss (LDAM). SCL enhances intra-class clusters based on feature similarity and promotes clear inter-class separability but tends to favour dominant classes only. When RSG is integrated into the model, we observed that the intra-class features further cluster towards the class centre, which demonstrates a synergistic effect together with SCL's principle of enhancing intra-class clustering. RSG generates new tail features and compensates for the tail feature space squeezed by SCL. Similarly, LDAM is known to introduce a larger margin specifically for tail classes; we demonstrate that LDAM further bolsters the model's performance on tail classes when combined with the more explicit decision boundaries achieved by SCL and RSG. Furthermore, SCL can compensate for the dominant class accuracy sacrificed by RSG and LDAM. Our research emphasises the synergy and balance among the three techniques, with each amplifying the strengths of the others and mitigating their shortcomings. Our experiment on long-tailed distribution datasets, using an end-to-end architecture, yields competitive results by enhancing tail class accuracy without compromising dominant class performance, achieving a balanced improvement across all classes.
Authors: Chaolei Han, Hongsong Wang, Jidong Kuang, Lei Zhang, Jie Gui
Abstract: Existing zero-shot temporal action detection (ZSTAD) methods predominantly use fully supervised or unsupervised strategies to recognize unseen activities. However, these training-based methods are prone to domain shifts and require high computational costs, which hinder their practical applicability in real-world scenarios. In this paper, unlike previous works, we propose a training-Free Zero-shot temporal Action Detection (FreeZAD) method, leveraging existing vision-language (ViL) models to directly classify and localize unseen activities within untrimmed videos without any additional fine-tuning or adaptation. We mitigate the need for explicit temporal modeling and reliance on pseudo-label quality by designing the LOGarithmic decay weighted Outer-Inner-Contrastive Score (LogOIC) and frequency-based Actionness Calibration. Furthermore, we introduce a test-time adaptation (TTA) strategy using Prototype-Centric Sampling (PCS) to expand FreeZAD, enabling ViL models to adapt more effectively for ZSTAD. Extensive experiments on the THUMOS14 and ActivityNet-1.3 datasets demonstrate that our training-free method outperforms state-of-the-art unsupervised methods while requiring only 1/13 of the runtime. When equipped with TTA, the enhanced method further narrows the gap with fully supervised methods.
Authors: Changhao Wang, Guanwen Zhang, Zhengyun Cheng, Wei Zhou
Abstract: Considerable efforts have been made to improve monocular depth estimation under ideal conditions. However, in challenging environments, monocular depth estimation still faces difficulties. In this paper, we introduce visual prompt learning for predicting depth across different environments within a unified model, and present a self-supervised learning framework called PromptMono. It employs a set of learnable parameters as visual prompts to capture domain-specific knowledge. To integrate prompting information into image representations, a novel gated cross prompting attention (GCPA) module is proposed, which enhances the depth estimation in diverse conditions. We evaluate the proposed PromptMono on the Oxford Robotcar dataset and the nuScenes dataset. Experimental results demonstrate the superior performance of the proposed method.
Authors: Yizhe Lv, Tingting Zhang, Yunpeng Song, Han Ding, Jinsong Han, Fei Wang
Abstract: Recent advanced Virtual Reality (VR) headsets, such as the Apple Vision Pro, employ bottom-facing cameras to detect hand gestures and inputs, which offers users significant convenience in VR interactions. However, these bottom-facing cameras can sometimes be inconvenient and pose a risk of unintentionally exposing sensitive information, such as private body parts or personal surroundings. To mitigate these issues, we introduce EgoHand. This system provides an alternative solution by integrating millimeter-wave radar and IMUs for hand gesture recognition, thereby offering users an additional option for gesture interaction that enhances privacy protection. To accurately recognize hand gestures, we devise a two-stage skeleton-based gesture recognition scheme. In the first stage, a novel end-to-end Transformer architecture is employed to estimate the coordinates of hand joints. Subsequently, these estimated joint coordinates are utilized for gesture recognition. Extensive experiments involving 10 subjects show that EgoHand can detect hand gestures with 90.8% accuracy. Furthermore, EgoHand demonstrates robust performance across a variety of cross-domain tests, including different users, dominant hands, body postures, and scenes.
Authors: Christopher Palazzolo, Oliver van Kaick, David Mould
Abstract: We propose a new method for synthesizing an arbitrarily sized novel vector texture given a single raster exemplar. Our method first segments the exemplar to extract the primary textons, and then clusters them based on visual similarity. We then compute a descriptor to capture each texton's neighborhood which contains the inter-category relationships that are used at synthesis time. Next, we use a simple procedure to both extract and place the secondary textons behind the primary polygons. Finally, our method constructs a gradient field for the background which is defined by a set of data points and colors. The color of the secondary polygons are also adjusted to better match the gradient field. To compare our work with other methods, we use a wide range of perceptual-based metrics.
Authors: Kairui Hu, Penghao Wu, Fanyi Pu, Wang Xiao, Yuanhan Zhang, Xiang Yue, Bo Li, Ziwei Liu
Abstract: Humans acquire knowledge through three cognitive stages: perceiving information, comprehending knowledge, and adapting knowledge to solve novel problems. Videos serve as an effective medium for this learning process, facilitating a progression through these cognitive stages. However, existing video benchmarks fail to systematically evaluate the knowledge acquisition capabilities in Large Multimodal Models (LMMs). To address this gap, we introduce Video-MMMU, a multi-modal, multi-disciplinary benchmark designed to assess LMMs' ability to acquire and utilize knowledge from videos. Video-MMMU features a curated collection of 300 expert-level videos and 900 human-annotated questions across six disciplines, evaluating knowledge acquisition through stage-aligned question-answer pairs: Perception, Comprehension, and Adaptation. A proposed knowledge gain metric, {\Delta}knowledge, quantifies improvement in performance after video viewing. Evaluation of LMMs reveals a steep decline in performance as cognitive demands increase and highlights a significant gap between human and model knowledge acquisition, underscoring the need for methods to enhance LMMs' capability to learn and adapt from videos.
Authors: Yuhui Lin, Jiaxuan Lu, Yue Yong, Jiahao Zhang
Abstract: Recent advancements in multi-view action recognition have largely relied on Transformer-based models. While effective and adaptable, these models often require substantial computational resources, especially in scenarios with multiple views and multiple temporal sequences. Addressing this limitation, this paper introduces the MV-GMN model, a state-space model specifically designed to efficiently aggregate multi-modal data (RGB and skeleton), multi-view perspectives, and multi-temporal information for action recognition with reduced computational complexity. The MV-GMN model employs an innovative Multi-View Graph Mamba network comprising a series of MV-GMN blocks. Each block includes a proposed Bidirectional State Space Block and a GCN module. The Bidirectional State Space Block introduces four scanning strategies, including view-prioritized and time-prioritized approaches. The GCN module leverages rule-based and KNN-based methods to construct the graph network, effectively integrating features from different viewpoints and temporal instances. Demonstrating its efficacy, MV-GMN outperforms the state-of-the-arts on several datasets, achieving notable accuracies of 97.3\% and 96.7\% on the NTU RGB+D 120 dataset in cross-subject and cross-view scenarios, respectively. MV-GMN also surpasses Transformer-based baselines while requiring only linear inference complexity, underscoring the model's ability to reduce computational load and enhance the scalability and applicability of multi-view action recognition technologies.
Authors: Mohammad Ali Rezaei, Fardin Ayar, Ehsan Javanmardi, Manabu Tsukada, Mahdi Javanmardi
Abstract: Accurate prediction of pedestrian trajectories is crucial for enhancing the safety of autonomous vehicles and reducing traffic fatalities involving pedestrians. While numerous studies have focused on modeling interactions among pedestrians to forecast their movements, the influence of environmental factors and scene-object placements has been comparatively underexplored. In this paper, we present a novel trajectory prediction model that integrates both pedestrian interactions and environmental context to improve prediction accuracy. Our approach captures spatial and temporal interactions among pedestrians within a sparse graph framework. To account for pedestrian-scene interactions, we employ advanced image enhancement and semantic segmentation techniques to extract detailed scene features. These scene and interaction features are then fused through a cross-attention mechanism, enabling the model to prioritize relevant environmental factors that influence pedestrian movements. Finally, a temporal convolutional network processes the fused features to predict future pedestrian trajectories. Experimental results demonstrate that our method significantly outperforms existing state-of-the-art approaches, achieving ADE and FDE values of 0.252 and 0.372 meters, respectively, underscoring the importance of incorporating both social interactions and environmental context in pedestrian trajectory prediction.
Authors: Timo Lange, Ajish Babu, Philipp Meyer, Matthis Keppner, Tim Tiedemann, Martin Wittmaier, Sebastian Wolff, Thomas V\"ogele
Abstract: Current disposal facilities for coarse-grained waste perform manual sorting of materials with heavy machinery. Large quantities of recyclable materials are lost to coarse waste, so more effective sorting processes must be developed to recover them. Two key aspects to automate the sorting process are object detection with material classification in mixed piles of waste, and autonomous control of hydraulic machinery. Because most objects in those accumulations of waste are damaged or destroyed, object detection alone is not feasible in the majority of cases. To address these challenges, we propose a classification of materials with multispectral images of ultraviolet (UV), visual (VIS), near infrared (NIR), and short-wave infrared (SWIR) spectrums. Solution for autonomous control of hydraulic heavy machines for sorting of bulky waste is being investigated using cost-effective cameras and artificial intelligence-based controllers.
Authors: Shiyu Zhang, Cheng Yan, Yang Liu, Chenchen Jing, Lei Zhou, Wenjun Wang
Abstract: Compositional Zero-Shot Learning (CZSL) aims to recognize novel compositions of attributes and objects by leveraging knowledge learned from seen compositions. Recent approaches have explored the use of Vision-Language Models (VLMs) to align textual and visual modalities. These methods typically employ prompt engineering, parameter-tuning, and modality fusion to generate rich textual prototypes that serve as class prototypes for CZSL. However, the modality gap results in textual prototypes being unable to fully capture the optimal representations of all class prototypes, particularly those with fine-grained features, which can be directly obtained from the visual modality. In this paper, we propose a novel Dual-Modal Prototype Joint Learning framework for the CZSL task. Our approach, based on VLMs, introduces prototypes in both the textual and visual modalities. The textual prototype is optimized to capture broad conceptual information, aiding the model's generalization across unseen compositions. Meanwhile, the visual prototype is used to mitigate the classification errors caused by the modality gap and capture fine-grained details to distinguish images with similar appearances. To effectively optimize these prototypes, we design specialized decomposition modules and a joint learning strategy that enrich the features from both modalities. These prototypes not only capture key category information during training but also serve as crucial reference targets during inference. Experimental results demonstrate that our approach achieves state-of-the-art performance in the closed-world setting and competitive performance in the open-world setting across three publicly available CZSL benchmarks. These findings validate the effectiveness of our method in advancing compositional generalization.
Authors: Abhishek Tandon, Geetanjali Sharma, Gaurav Jaswal, Aditya Nigam, Raghavendra Ramachandra
Abstract: We propose a trait-specific image generation method that models forehead creases geometrically using B-spline and B\'ezier curves. This approach ensures the realistic generation of both principal creases and non-prominent crease patterns, effectively constructing detailed and authentic forehead-crease images. These geometrically rendered images serve as visual prompts for a diffusion-based Edge-to-Image translation model, which generates corresponding mated samples. The resulting novel synthetic identities are then used to train a forehead-crease verification network. To enhance intra-subject diversity in the generated samples, we employ two strategies: (a) perturbing the control points of B-splines under defined constraints to maintain label consistency, and (b) applying image-level augmentations to the geometric visual prompts, such as dropout and elastic transformations, specifically tailored to crease patterns. By integrating the proposed synthetic dataset with real-world data, our method significantly improves the performance of forehead-crease verification systems under a cross-database verification protocol.
Authors: Zuyao You, Junke Wang, Lingyu Kong, Bo He, Zuxuan Wu
Abstract: We present Pix2Cap-COCO, the first panoptic pixel-level caption dataset designed to advance fine-grained visual understanding. To achieve this, we carefully design an automated annotation pipeline that prompts GPT-4V to generate pixel-aligned, instance-specific captions for individual objects within images, enabling models to learn more granular relationships between objects and their contexts. This approach results in 167,254 detailed captions, with an average of 22.94 words per caption. Building on Pix2Cap-COCO, we introduce a novel task, panoptic segmentation-captioning, which challenges models to recognize instances in an image and provide detailed descriptions for each simultaneously. To benchmark this task, we design a robust baseline based on X-Decoder. The experimental results demonstrate that Pix2Cap-COCO is a particularly challenging dataset, as it requires models to excel in both fine-grained visual understanding and detailed language generation. Furthermore, we leverage Pix2Cap-COCO for Supervised Fine-Tuning (SFT) on large multimodal models (LMMs) to enhance their performance. For example, training with Pix2Cap-COCO significantly improves the performance of GPT4RoI, yielding gains in CIDEr +1.4%, ROUGE +0.4%, and SPICE +0.5% on Visual Genome dataset, and strengthens its region understanding ability on the ViP-BENCH, with an overall improvement of +5.1%, including notable increases in recognition accuracy +11.2% and language generation quality +22.2%.
Authors: Peiyuan Zhang, Junwei Luo, Xue Yang, Yi Yu, Qingyun Li, Yue Zhou, Xiaosong Jia, Xudong Lu, Jingdong Chen, Xiang Li, Junchi Yan, Yansheng Li
Abstract: With the growing demand for oriented object detection (OOD), recent studies on point-supervised OOD have attracted significant interest. In this paper, we propose PointOBB-v3, a stronger single point-supervised OOD framework. Compared to existing methods, it generates pseudo rotated boxes without additional priors and incorporates support for the end-to-end paradigm. PointOBB-v3 functions by integrating three unique image views: the original view, a resized view, and a rotated/flipped (rot/flp) view. Based on the views, a scale augmentation module and an angle acquisition module are constructed. In the first module, a Scale-Sensitive Consistency (SSC) loss and a Scale-Sensitive Feature Fusion (SSFF) module are introduced to improve the model's ability to estimate object scale. To achieve precise angle predictions, the second module employs symmetry-based self-supervised learning. Additionally, we introduce an end-to-end version that eliminates the pseudo-label generation process by integrating a detector branch and introduces an Instance-Aware Weighting (IAW) strategy to focus on high-quality predictions. We conducted extensive experiments on the DIOR-R, DOTA-v1.0/v1.5/v2.0, FAIR1M, STAR, and RSAR datasets. Across all these datasets, our method achieves an average improvement in accuracy of 3.56% in comparison to previous state-of-the-art methods. The code will be available at https://github.com/ZpyWHU/PointOBB-v3.
Authors: Vitaliy Kinakh, Slava Voloshynovskiy
Abstract: We introduce the Binary Diffusion Probabilistic Model (BDPM), a novel generative model optimized for binary data representations. While denoising diffusion probabilistic models (DDPMs) have demonstrated notable success in tasks like image synthesis and restoration, traditional DDPMs rely on continuous data representations and mean squared error (MSE) loss for training, applying Gaussian noise models that may not be optimal for discrete or binary data structures. BDPM addresses this by decomposing images into bitplanes and employing XOR-based noise transformations, with a denoising model trained using binary cross-entropy loss. This approach enables precise noise control and computationally efficient inference, significantly lowering computational costs and improving model convergence. When evaluated on image restoration tasks such as image super-resolution, inpainting, and blind image restoration, BDPM outperforms state-of-the-art methods on the FFHQ, CelebA, and CelebA-HQ datasets. Notably, BDPM requires fewer inference steps than traditional DDPM models to reach optimal results, showcasing enhanced inference efficiency.
Authors: Jie Liu, Gongye Liu, Jiajun Liang, Ziyang Yuan, Xiaokun Liu, Mingwu Zheng, Xiele Wu, Qiulin Wang, Wenyu Qin, Menghan Xia, Xintao Wang, Xiaohong Liu, Fei Yang, Pengfei Wan, Di Zhang, Kun Gai, Yujiu Yang, Wanli Ouyang
Abstract: Video generation has achieved significant advances through rectified flow techniques, but issues like unsmooth motion and misalignment between videos and prompts persist. In this work, we develop a systematic pipeline that harnesses human feedback to mitigate these problems and refine the video generation model. Specifically, we begin by constructing a large-scale human preference dataset focused on modern video generation models, incorporating pairwise annotations across multi-dimensions. We then introduce VideoReward, a multi-dimensional video reward model, and examine how annotations and various design choices impact its rewarding efficacy. From a unified reinforcement learning perspective aimed at maximizing reward with KL regularization, we introduce three alignment algorithms for flow-based models by extending those from diffusion models. These include two training-time strategies: direct preference optimization for flow (Flow-DPO) and reward weighted regression for flow (Flow-RWR), and an inference-time technique, Flow-NRG, which applies reward guidance directly to noisy videos. Experimental results indicate that VideoReward significantly outperforms existing reward models, and Flow-DPO demonstrates superior performance compared to both Flow-RWR and standard supervised fine-tuning methods. Additionally, Flow-NRG lets users assign custom weights to multiple objectives during inference, meeting personalized video quality needs. Project page: https://gongyeliu.github.io/videoalign.
Authors: Rui Li, Xiaohan Wang, Yuhui Zhang, Zeyu Wang, Serena Yeung-Levy
Abstract: Despite significant advancements in video large multimodal models (video-LMMs), achieving effective temporal grounding in long-form videos remains a challenge for existing models. To address this limitation, we propose Temporal Preference Optimization (TPO), a novel post-training framework designed to enhance the temporal grounding capabilities of video-LMMs through preference learning. TPO adopts a self-training approach that enables models to differentiate between well-grounded and less accurate temporal responses by leveraging curated preference datasets at two granularities: localized temporal grounding, which focuses on specific video segments, and comprehensive temporal grounding, which captures extended temporal dependencies across entire video sequences. By optimizing on these preference datasets, TPO significantly enhances temporal understanding while reducing reliance on manually annotated data. Extensive experiments on three long-form video understanding benchmarks--LongVideoBench, MLVU, and Video-MME--demonstrate the effectiveness of TPO across two state-of-the-art video-LMMs. Notably, LLaVA-Video-TPO establishes itself as the leading 7B model on the Video-MME benchmark, underscoring the potential of TPO as a scalable and efficient solution for advancing temporal reasoning in long-form video understanding. Project page: https://ruili33.github.io/tpo_website.
Authors: Jiayi Lei, Renrui Zhang, Xiangfei Hu, Weifeng Lin, Zhen Li, Wenjian Sun, Ruoyi Du, Le Zhuo, Zhongyu Li, Xinyue Li, Shitian Zhao, Ziyu Guo, Yiting Lu, Peng Gao, Hongsheng Li
Abstract: With the rapid development of diffusion models, text-to-image(T2I) models have made significant progress, showcasing impressive abilities in prompt following and image generation. Recently launched models such as FLUX.1 and Ideogram2.0, along with others like Dall-E3 and Stable Diffusion 3, have demonstrated exceptional performance across various complex tasks, raising questions about whether T2I models are moving towards general-purpose applicability. Beyond traditional image generation, these models exhibit capabilities across a range of fields, including controllable generation, image editing, video, audio, 3D, and motion generation, as well as computer vision tasks like semantic segmentation and depth estimation. However, current evaluation frameworks are insufficient to comprehensively assess these models' performance across expanding domains. To thoroughly evaluate these models, we developed the IMAGINE-E and tested six prominent models: FLUX.1, Ideogram2.0, Midjourney, Dall-E3, Stable Diffusion 3, and Jimeng. Our evaluation is divided into five key domains: structured output generation, realism, and physical consistency, specific domain generation, challenging scenario generation, and multi-style creation tasks. This comprehensive assessment highlights each model's strengths and limitations, particularly the outstanding performance of FLUX.1 and Ideogram2.0 in structured and specific domain tasks, underscoring the expanding applications and potential of T2I models as foundational AI tools. This study provides valuable insights into the current state and future trajectory of T2I models as they evolve towards general-purpose usability. Evaluation scripts will be released at https://github.com/jylei16/Imagine-e.
Authors: Hao Dong, Eleni Chatzi, Olga Fink
Abstract: Test-time adaptation (TTA) has demonstrated significant potential in addressing distribution shifts between training and testing data. Open-set test-time adaptation (OSTTA) aims to adapt a source pre-trained model online to an unlabeled target domain that contains unknown classes. This task becomes more challenging when multiple modalities are involved. Existing methods have primarily focused on unimodal OSTTA, often filtering out low-confidence samples without addressing the complexities of multimodal data. In this work, we present Adaptive Entropy-aware Optimization (AEO), a novel framework specifically designed to tackle Multimodal Open-set Test-time Adaptation (MM-OSTTA) for the first time. Our analysis shows that the entropy difference between known and unknown samples in the target domain strongly correlates with MM-OSTTA performance. To leverage this, we propose two key components: Unknown-aware Adaptive Entropy Optimization (UAE) and Adaptive Modality Prediction Discrepancy Optimization (AMP). These components enhance the ability of model to distinguish unknown class samples during online adaptation by amplifying the entropy difference between known and unknown samples. To thoroughly evaluate our proposed methods in the MM-OSTTA setting, we establish a new benchmark derived from existing datasets. This benchmark includes two downstream tasks and incorporates five modalities. Extensive experiments across various domain shift situations demonstrate the efficacy and versatility of the AEO framework. Additionally, we highlight the strong performance of AEO in long-term and continual MM-OSTTA settings, both of which are challenging and highly relevant to real-world applications. Our source code is available at https://github.com/donghao51/AEO.
Authors: Akashah Shabbir, Mohammed Zumri, Mohammed Bennamoun, Fahad S. Khan, Salman Khan
Abstract: Recent advances in large multimodal models (LMMs) have recognized fine-grained grounding as an imperative factor of visual understanding and dialogue. However, the benefits of such representation in LMMs are limited to the natural image domain, and these models perform poorly for remote sensing (RS). The distinct overhead viewpoint, scale variation, and presence of small objects in high-resolution RS imagery present a unique challenge in region-level comprehension. Moreover, the development of the grounding conversation capability of LMMs within RS is hindered by the lack of granular, RS domain-specific grounded data. Addressing these limitations, we propose GeoPixel - the first end-to-end high resolution RS-LMM that supports pixel-level grounding. This capability allows fine-grained visual perception by generating interleaved masks in conversation. GeoPixel supports up to 4K HD resolution in any aspect ratio, ideal for high-precision RS image analysis. To support the grounded conversation generation (GCG) in RS imagery, we curate a visually grounded dataset GeoPixelD through a semi-automated pipeline that utilizes set-of-marks prompting and spatial priors tailored for RS data to methodically control the data generation process. GeoPixel demonstrates superior performance in pixel-level comprehension, surpassing existing LMMs in both single-target and multi-target segmentation tasks. Our methodological ablation studies validate the effectiveness of each component in the overall architecture. Our code and data will be publicly released.
Authors: Ziyu Guo, Renrui Zhang, Chengzhuo Tong, Zhizheng Zhao, Peng Gao, Hongsheng Li, Pheng-Ann Heng
Abstract: Chain-of-Thought (CoT) reasoning has been extensively explored in large models to tackle complex understanding tasks. However, it still remains an open question whether such strategies can be applied to verifying and reinforcing image generation scenarios. In this paper, we provide the first comprehensive investigation of the potential of CoT reasoning to enhance autoregressive image generation. We focus on three techniques: scaling test-time computation for verification, aligning model preferences with Direct Preference Optimization (DPO), and integrating these techniques for complementary effects. Our results demonstrate that these approaches can be effectively adapted and combined to significantly improve image generation performance. Furthermore, given the pivotal role of reward models in our findings, we propose the Potential Assessment Reward Model (PARM) and PARM++, specialized for autoregressive image generation. PARM adaptively assesses each generation step through a potential assessment approach, merging the strengths of existing reward models, and PARM++ further introduces a reflection mechanism to self-correct the generated unsatisfactory image. Using our investigated reasoning strategies, we enhance a baseline model, Show-o, to achieve superior results, with a significant +24% improvement on the GenEval benchmark, surpassing Stable Diffusion 3 by +15%. We hope our study provides unique insights and paves a new path for integrating CoT reasoning with autoregressive image generation. Code and models are released at https://github.com/ZiyuGuo99/Image-Generation-CoT
Authors: Jianing Yang, Alexander Sax, Kevin J. Liang, Mikael Henaff, Hao Tang, Ang Cao, Joyce Chai, Franziska Meier, Matt Feiszli
Abstract: Multi-view 3D reconstruction remains a core challenge in computer vision, particularly in applications requiring accurate and scalable representations across diverse perspectives. Current leading methods such as DUSt3R employ a fundamentally pairwise approach, processing images in pairs and necessitating costly global alignment procedures to reconstruct from multiple views. In this work, we propose Fast 3D Reconstruction (Fast3R), a novel multi-view generalization to DUSt3R that achieves efficient and scalable 3D reconstruction by processing many views in parallel. Fast3R's Transformer-based architecture forwards N images in a single forward pass, bypassing the need for iterative alignment. Through extensive experiments on camera pose estimation and 3D reconstruction, Fast3R demonstrates state-of-the-art performance, with significant improvements in inference speed and reduced error accumulation. These results establish Fast3R as a robust alternative for multi-view applications, offering enhanced scalability without compromising reconstruction accuracy.
Authors: Ella Koresh, Ronit D. Gross, Yuval Meir, Yarden Tzach, Tal Halevi, Ido Kanter
Abstract: Convolutional neural networks (CNNs) evaluate short-range correlations in input images which progress along the layers, whereas vision transformer (ViT) architectures evaluate long-range correlations, using repeated transformer encoders composed of fully connected layers. Both are designed to solve complex classification tasks but from different perspectives. This study demonstrates that CNNs and ViT architectures stem from a unified underlying learning mechanism, which quantitatively measures the single-nodal performance (SNP) of each node in feedforward (FF) and multi-head attention (MHA) subblocks. Each node identifies small clusters of possible output labels, with additional noise represented as labels outside these clusters. These features are progressively sharpened along the transformer encoders, enhancing the signal-to-noise ratio. This unified underlying learning mechanism leads to two main findings. First, it enables an efficient applied nodal diagonal connection (ANDC) pruning technique without affecting the accuracy. Second, based on the SNP, spontaneous symmetry breaking occurs among the MHA heads, such that each head focuses its attention on a subset of labels through cooperation among its SNPs. Consequently, each head becomes an expert in recognizing its designated labels, representing a quantitative MHA modus vivendi mechanism. These results are based on a compact convolutional transformer architecture trained on the CIFAR-100 and Flowers-102 datasets and call for their extension to other architectures and applications, such as natural language processing.
Authors: Alexander Spinos, Bradley Woosley, Justin Rokisky, Christopher Korpela, John G. Rogers III, Brian A. Bittner
Abstract: Traditionally, autonomous reconnaissance applications have acted on explicit sets of historical observations. Aided by recent breakthroughs in generative technologies, this work enables robot teams to act beyond what is currently known about the environment by inferring a distribution of reasonable interpretations of the scene. We developed a map predictor that inpaints the unknown space in a multi-agent 2D occupancy map during an exploration mission. From a comparison of several inpainting methods, we found that a fine-tuned latent diffusion inpainting model could provide rich and coherent interpretations of simulated urban environments with relatively little computation time. By iteratively inferring interpretations of the scene throughout an exploration run, we are able to identify areas that exhibit high uncertainty in the prediction, which we formalize with the concept of generative entropy. We prioritize tasks in regions of high generative entropy, hypothesizing that this will expedite convergence on an accurate predicted map of the scene. In our study we juxtapose this new paradigm of task ranking with the state of the art, which ranks regions to explore by those which maximize expected information recovery. We compare both of these methods in a simulated urban environment with three vehicles. Our results demonstrate that by using our new task ranking method, we can predict a correct scene significantly faster than with a traditional information-guided method.
Authors: Adam Tupper, Christian Gagn\'e
Abstract: Data augmentation is a widely used and effective technique to improve the generalization performance of deep neural networks. Yet, despite often facing limited data availability when working with medical images, it is frequently underutilized. This appears to come from a gap in our collective understanding of the efficacy of different augmentation techniques across different tasks and modalities. One modality where this is especially true is ultrasound imaging. This work addresses this gap by analyzing the effectiveness of different augmentation techniques at improving model performance across a wide range of ultrasound image analysis tasks. To achieve this, we introduce a new standardized benchmark of 14 ultrasound image classification and semantic segmentation tasks from 10 different sources and covering 11 body regions. Our results demonstrate that many of the augmentations commonly used for tasks on natural images are also effective on ultrasound images, even more so than augmentations developed specifically for ultrasound images in some cases. We also show that diverse augmentation using TrivialAugment, which is widely used for natural images, is also effective for ultrasound images. Moreover, our proposed methodology represents a structured approach for assessing various data augmentations that can be applied to other contexts and modalities.
Authors: Reza Saadati Fard, Emmanuel Agu, Palawat Busaranuvong, Deepak Kumar, Shefalika Gautam, Bengisu Tulu, Diane Strong
Abstract: Chronic wounds affect 8.5 million Americans, particularly the elderly and patients with diabetes. These wounds can take up to nine months to heal, making regular care essential to ensure healing and prevent severe outcomes like limb amputations. Many patients receive care at home from visiting nurses with varying levels of wound expertise, leading to inconsistent care. Problematic, non-healing wounds should be referred to wound specialists, but referral decisions in non-clinical settings are often erroneous, delayed, or unnecessary. This paper introduces the Deep Multimodal Wound Assessment Tool (DM-WAT), a machine learning framework designed to assist visiting nurses in deciding whether to refer chronic wound patients. DM-WAT analyzes smartphone-captured wound images and clinical notes from Electronic Health Records (EHRs). It uses DeiT-Base-Distilled, a Vision Transformer (ViT), to extract visual features from images and DeBERTa-base to extract text features from clinical notes. DM-WAT combines visual and text features using an intermediate fusion approach. To address challenges posed by a small and imbalanced dataset, it integrates image and text augmentation with transfer learning to achieve high performance. In evaluations, DM-WAT achieved 77% with std 3% accuracy and a 70% with std 2% F1 score, outperforming prior approaches. Score-CAM and Captum interpretation algorithms provide insights into specific parts of image and text inputs that influence recommendations, enhancing interpretability and trust.
Authors: Yixuan Wang, Leonor Fermoselle, Tarik Kelestemur, Jiuguang Wang, Yunzhu Li
Abstract: Mobile exploration is a longstanding challenge in robotics, yet current methods primarily focus on active perception instead of active interaction, limiting the robot's ability to interact with and fully explore its environment. Existing robotic exploration approaches via active interaction are often restricted to tabletop scenes, neglecting the unique challenges posed by mobile exploration, such as large exploration spaces, complex action spaces, and diverse object relations. In this work, we introduce a 3D relational object graph that encodes diverse object relations and enables exploration through active interaction. We develop a system based on this representation and evaluate it across diverse scenes. Our qualitative and quantitative results demonstrate the system's effectiveness and generalization capabilities, outperforming methods that rely solely on vision-language models (VLMs).
Authors: Tianyuan Yao, Zhiyuan Li, Praitayini Kanakaraj, Derek B. Archer, Kurt Schilling, Lori Beason-Held, Susan Resnick, Bennett A. Landman, Yuankai Huo
Abstract: Diffusion-weighted Magnetic Resonance Imaging (dMRI) is an essential tool in neuroimaging. It is arguably the sole noninvasive technique for examining the microstructural properties and structural connectivity of the brain. Recent years have seen the emergence of machine learning and data-driven approaches that enhance the speed, accuracy, and consistency of dMRI data analysis. However, traditional deep learning models often fell short, as they typically utilize pixel-level or volumetric patch-level embeddings similar to those used in structural MRI, and do not account for the unique distribution of various gradient encodings. In this paper, we propose a novel method called Polyhedra Encoding Transformer (PE-Transformer) for dMRI, designed specifically to handle spherical signals. Our approach involves projecting an icosahedral polygon onto a unit sphere to resample signals from predetermined directions. These resampled signals are then transformed into embeddings, which are processed by a transformer encoder that incorporates orientational information reflective of the icosahedral structure. Through experimental validation with various gradient encoding protocols, our method demonstrates superior accuracy in estimating multi-compartment models and Fiber Orientation Distributions (FOD), outperforming both conventional CNN architectures and standard transformers.
Authors: Peirong Liu, Ana Lawry Aguila, Juan E. Iglesias
Abstract: Data-driven machine learning has made significant strides in medical image analysis. However, most existing methods are tailored to specific modalities and assume a particular resolution (often isotropic). This limits their generalizability in clinical settings, where variations in scan appearance arise from differences in sequence parameters, resolution, and orientation. Furthermore, most general-purpose models are designed for healthy subjects and suffer from performance degradation when pathology is present. We introduce UNA (Unraveling Normal Anatomy), the first modality-agnostic learning approach for normal brain anatomy reconstruction that can handle both healthy scans and cases with pathology. We propose a fluid-driven anomaly randomization method that generates an unlimited number of realistic pathology profiles on-the-fly. UNA is trained on a combination of synthetic and real data, and can be applied directly to real images with potential pathology without the need for fine-tuning. We demonstrate UNA's effectiveness in reconstructing healthy brain anatomy and showcase its direct application to anomaly detection, using both simulated and real images from 3D healthy and stroke datasets, including CT and MRI scans. By bridging the gap between healthy and diseased images, UNA enables the use of general-purpose models on diseased images, opening up new opportunities for large-scale analysis of uncurated clinical images in the presence of pathology. Code is available at https://github.com/peirong26/UNA.
Authors: Gabrielle Hoyer, Michelle W Tong, Rupsa Bhattacharjee, Valentina Pedoia, Sharmila Majumdar
Abstract: Foundation models hold transformative potential for medical imaging, but their clinical utility requires rigorous evaluation to address their strengths and limitations. This study introduces an evaluation framework for assessing the clinical impact and translatability of SAM, MedSAM, and SAM2, using musculoskeletal MRI as a case study. We tested these models across zero-shot and finetuned paradigms to assess their ability to process diverse anatomical structures and effectuate clinically reliable biomarkers, including cartilage thickness, muscle volume, and disc height. We engineered a modular pipeline emphasizing scalability, clinical relevance, and workflow integration, reducing manual effort and aligning validation with end-user expectations. Hierarchical modeling revealed how dataset mixing, anatomical complexity, and MRI acquisition parameters influence performance, providing insights into the role of imaging refinements in improving segmentation accuracy. This work demonstrates how clinically focused evaluations can connect computational advancements with tangible applications, creating a pathway for foundation models to address medical challenges. By emphasizing interdisciplinary collaboration and aligning technical innovation with clinical priorities, our framework provides a roadmap for advancing machine learning technologies into scalable and impactful biomedical solutions.
Authors: Gyuhyeon Pak, Euntai Kim
Abstract: Recently, map representations based on radiance fields such as 3D Gaussian Splatting and NeRF, which excellent for realistic depiction, have attracted considerable attention, leading to attempts to combine them with SLAM. While these approaches can build highly realistic maps, large-scale SLAM still remains a challenge because they require a large number of Gaussian images for mapping and adjacent images as keyframes for tracking. We propose a novel 3D Gaussian Splatting SLAM method, VIGS SLAM, that utilizes sensor fusion of RGB-D and IMU sensors for large-scale indoor environments. To reduce the computational load of 3DGS-based tracking, we adopt an ICP-based tracking framework that combines IMU preintegration to provide a good initial guess for accurate pose estimation. Our proposed method is the first to propose that Gaussian Splatting-based SLAM can be effectively performed in large-scale environments by integrating IMU sensor measurements. This proposal not only enhances the performance of Gaussian Splatting SLAM beyond room-scale scenarios but also achieves SLAM performance comparable to state-of-the-art methods in large-scale indoor environments.
Authors: Jaewon Lee, Mangyu Kong, Minseong Park, Euntai Kim
Abstract: Mapping and localization are crucial problems in robotics and autonomous driving. Recent advances in 3D Gaussian Splatting (3DGS) have enabled precise 3D mapping and scene understanding by rendering photo-realistic images. However, existing 3DGS methods often struggle to accurately reconstruct a 3D map that reflects the actual scale and geometry of the real world, which degrades localization performance. To address these limitations, we propose a novel 3DGS method called Geometry-Aware Gaussian Splatting (GeomGS). This method fully integrates LiDAR data into 3D Gaussian primitives via a probabilistic approach, as opposed to approaches that only use LiDAR as initial points or introduce simple constraints for Gaussian points. To this end, we introduce a Geometric Confidence Score (GCS), which identifies the structural reliability of each Gaussian point. The GCS is optimized simultaneously with Gaussians under probabilistic distance constraints to construct a precise structure. Furthermore, we propose a novel localization method that fully utilizes both the geometric and photometric properties of GeomGS. Our GeomGS demonstrates state-of-the-art geometric and localization performance across several benchmarks, while also improving photometric performance.
Authors: Huilin Yin, Yangwenhui Xu, Jiaxiang Li, Hao Zhang, Gerhard Rigoll
Abstract: Multi-agent trajectory prediction at signalized intersections is crucial for developing efficient intelligent transportation systems and safe autonomous driving systems. Due to the complexity of intersection scenarios and the limitations of single-vehicle perception, the performance of vehicle-centric prediction methods has reached a plateau. Furthermore, most works underutilize critical intersection information, including traffic signals, and behavior patterns induced by road structures. Therefore, we propose a multi-agent trajectory prediction framework at signalized intersections dedicated to Infrastructure-to-Everything (I2XTraj). Our framework leverages dynamic graph attention to integrate knowledge from traffic signals and driving behaviors. A continuous signal-informed mechanism is proposed to adaptively process real-time traffic signals from infrastructure devices. Additionally, leveraging the prior knowledge of the intersection topology, we propose a driving strategy awareness mechanism to model the joint distribution of goal intentions and maneuvers. To the best of our knowledge, I2XTraj represents the first multi-agent trajectory prediction framework explicitly designed for infrastructure deployment, supplying subscribable prediction services to all vehicles at intersections. I2XTraj demonstrates state-of-the-art performance on both the Vehicle-to-Infrastructure dataset V2X-Seq and the aerial-view dataset SinD for signalized intersections. Quantitative evaluations show that our approach outperforms existing methods by more than 30% in both multi-agent and single-agent scenarios.
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.
Authors: Khadija Rais, Mohamed Amroune, Mohamed Yassine Haouam
Abstract: Medical image analysis suffers from a lack of labeled data due to several challenges including patient privacy and lack of experts. Although some AI models only perform well with large amounts of data, we will move to data augmentation where there is a solution to improve the performance of our models and increase the dataset size through traditional or advanced techniques. In this paper, we evaluate the effectiveness of data augmentation techniques on two different medical image datasets. In the first step, we applied some transformation techniques to the skin cancer dataset containing benign and malignant classes. Then, we trained the convolutional neural network (CNN) on the dataset before and after augmentation, which significantly improved test accuracy from 90.74% to 96.88% and decreased test loss from 0.7921 to 0.1468 after augmentation. In the second step, we used the Mixup technique by mixing two random images and their corresponding masks using the retina and blood vessels dataset, then we trained the U-net model and obtained the Dice coefficient which increased from 0 before augmentation to 0.4163 after augmentation. The result shows the effect of using data augmentation to increase the dataset size on the classification and segmentation performance.
Authors: Muhammad Shahkar Khan, Haider Ali, Laura Villazan Garcia, Noor Badshah, Siegfried Trattnig, Florian Schwarzhans, Ramona Woitek, Olgica Zaric
Abstract: Background: Multiparametric breast MRI data might improve tumor diagnostics, characterization, and treatment planning. Accurate alignment and delineation of images acquired at different field strengths such as 3T and 7T, remain challenging research tasks. Purpose: To address alignment challenges and enable consistent tumor segmentation across different MRI field strengths. Study type: Retrospective. Subjects: Nine female subjects with breast tumors were involved: six histologically proven invasive ductal carcinomas (IDC) and three fibroadenomas. Field strength/sequence: Imaging was performed at 3T and 7T scanners using post-contrast T1-weighted three-dimensional time-resolved angiography with stochastic trajectories (TWIST) sequence. Assessments: The method's performance for joint image registration and tumor segmentation was evaluated using several quantitative metrics, including signal-to-noise ratio (PSNR), structural similarity index (SSIM), normalized cross-correlation (NCC), Dice coefficient, F1 score, and relative sum of squared differences (rel SSD). Statistical tests: The Pearson correlation coefficient was used to test the relationship between the registration and segmentation metrics. Results: When calculated for each subject individually, the PSNR was in a range from 27.5 to 34.5 dB, and the SSIM was from 82.6 to 92.8%. The model achieved an NCC from 96.4 to 99.3% and a Dice coefficient of 62.9 to 95.3%. The F1 score was between 55.4 and 93.2% and the rel SSD was in the range of 2.0 and 7.5%. The segmentation metrics Dice and F1 Score are highly correlated (0.995), while a moderate correlation between NCC and SSIM (0.681) was found for registration. Data conclusion: Initial results demonstrate that the proposed method may be feasible in providing joint tumor segmentation and registration of MRI data acquired at different field strengths.
Authors: Han Li, Shaohui Li, Wenrui Dai, Maida Cao, Nuowen Kan, Chenglin Li, Junni Zou, Hongkai Xiong
Abstract: Learned image compression (LIC) has demonstrated superior rate-distortion (R-D) performance compared to traditional codecs, but is challenged by training inefficiency that could incur more than two weeks to train a state-of-the-art model from scratch. Existing LIC methods overlook the slow convergence caused by compacting energy in learning nonlinear transforms. In this paper, we first reveal that such energy compaction consists of two components, i.e., feature decorrelation and uneven energy modulation. On such basis, we propose a linear auxiliary transform (AuxT) to disentangle energy compaction in training nonlinear transforms. The proposed AuxT obtains coarse approximation to achieve efficient energy compaction such that distribution fitting with the nonlinear transforms can be simplified to fine details. We then develop wavelet-based linear shortcuts (WLSs) for AuxT that leverages wavelet-based downsampling and orthogonal linear projection for feature decorrelation and subband-aware scaling for uneven energy modulation. AuxT is lightweight and plug-and-play to be integrated into diverse LIC models to address the slow convergence issue. Experimental results demonstrate that the proposed approach can accelerate training of LIC models by 2 times and simultaneously achieves an average 1\% BD-rate reduction. To our best knowledge, this is one of the first successful attempt that can significantly improve the convergence of LIC with comparable or superior rate-distortion performance. Code will be released at \url{https://github.com/qingshi9974/AuxT}
Authors: Frederik Pahde, Thomas Wiegand, Sebastian Lapuschkin, Wojciech Samek
Abstract: Deep neural networks are increasingly employed in high-stakes medical applications, despite their tendency for shortcut learning in the presence of spurious correlations, which can have potentially fatal consequences in practice. Detecting and mitigating shortcut behavior is a challenging task that often requires significant labeling efforts from domain experts. To alleviate this problem, we introduce a semi-automated framework for the identification of spurious behavior from both data and model perspective by leveraging insights from eXplainable Artificial Intelligence (XAI). This allows the retrieval of spurious data points and the detection of model circuits that encode the associated prediction rules. Moreover, we demonstrate how these shortcut encodings can be used for XAI-based sample- and pixel-level data annotation, providing valuable information for bias mitigation methods to unlearn the undesired shortcut behavior. We show the applicability of our framework using four medical datasets across two modalities, featuring controlled and real-world spurious correlations caused by data artifacts. We successfully identify and mitigate these biases in VGG16, ResNet50, and contemporary Vision Transformer models, ultimately increasing their robustness and applicability for real-world medical tasks.
Authors: Ethan Wilson, Naveen Sendhilnathan, Charlie S. Burlingham, Yusuf Mansour, Robert Cavin, Sai Deep Tetali, Ajoy Savio Fernandes, Michael J. Proulx
Abstract: Advanced multimodal AI agents can now collaborate with users to solve challenges in the world. We explore eye tracking's role in such interaction to convey a user's attention relative to the physical environment. We hypothesize that this knowledge improves contextual understanding for AI agents. By observing hours of human-object interactions, we first measure the relationship between an eye tracker's signal quality and its ability to reliably place gaze on nearby physical objects. We then conduct experiments which relay the user's scanpath history as additional context querying multimodal agents. Our results show that eye tracking provides high value as a user attention signal and can convey information about the user's current task and interests to the agent.
Authors: Ali Abedi, Charlene H. Chu, Shehroz S. Khan
Abstract: Lower-Limb Fractures (LLF) are a major health concern for older adults, often leading to reduced mobility and prolonged recovery, potentially impairing daily activities and independence. During recovery, older adults frequently face social isolation and functional decline, complicating rehabilitation and adversely affecting physical and mental health. Multi-modal sensor platforms that continuously collect data and analyze it using machine-learning algorithms can remotely monitor this population and infer health outcomes. They can also alert clinicians to individuals at risk of isolation and decline. This paper presents a new publicly available multi-modal sensor dataset, MAISON-LLF, collected from older adults recovering from LLF in community settings. The dataset includes data from smartphone and smartwatch sensors, motion detectors, sleep-tracking mattresses, and clinical questionnaires on isolation and decline. The dataset was collected from ten older adults living alone at home for eight weeks each, totaling 560 days of 24-hour sensor data. For technical validation, supervised machine-learning and deep-learning models were developed using the sensor and clinical questionnaire data, providing a foundational comparison for the research community.
Authors: Yue Fan, Handong Zhao, Ruiyi Zhang, Yu Shen, Xin Eric Wang, Gang Wu
Abstract: Graphical User Interface (GUI) action grounding is a critical step in GUI automation that maps language instructions to actionable elements on GUI screens. Most recent works of GUI action grounding leverage large GUI datasets to fine-tune MLLMs. However, the fine-tuning data always covers limited GUI environments, and we find the performance of the resulting model deteriorates in novel environments. We argue that the GUI grounding models should be further aligned to the novel environments to reveal their full potential, when the inference is known to involve novel environments, i.e., environments not used during the previous fine-tuning. To realize this, we first propose GUI-Bee, an MLLM-based autonomous agent, to collect high-quality, environment-specific data through exploration and then continuously fine-tune GUI grounding models with the collected data. Our agent leverages a novel Q-value-Incentive In-Context Reinforcement Learning (Q-ICRL) method to optimize exploration efficiency and data quality. Additionally, we introduce NovelScreenSpot, a benchmark for testing how well the data can help align GUI action grounding models to novel environments and demonstrate the effectiveness of data collected by GUI-Bee in the experiments. Furthermore, we conduct an ablation study to validate the Q-ICRL method in enhancing the efficiency of GUI-Bee. Project page: https://gui-bee.github.io
Authors: Guofeng Cui, Pichao Wang, Yang Liu, Zemian Ke, Zhu Liu, Vimal Bhat
Abstract: Large language models (LLMs) have shown great potential in natural language processing tasks, but their application to machine translation (MT) remains challenging due to pretraining on English-centric data and the complexity of reinforcement learning from human feedback (RLHF). Direct Preference Optimization (DPO) has emerged as a simpler and more efficient alternative, but its performance depends heavily on the quality of preference data. To address this, we propose Confidence-Reward driven Preference Optimization (CRPO), a novel method that combines reward scores with model confidence to improve data selection for fine-tuning. CRPO selects challenging sentence pairs where the model is uncertain or underperforms, leading to more effective learning. While primarily designed for LLMs, CRPO also generalizes to encoder-decoder models like NLLB, demonstrating its versatility. Empirical results show that CRPO outperforms existing methods such as RS-DPO, RSO and MBR score in both translation accuracy and data efficiency.
Authors: Vaibhav Vavilala, Faaris Shaik, David Forsyth
Abstract: We demonstrate an image dequantizing diffusion model that enables novel edits on natural images. We propose operating on quantized images because they offer easy abstraction for patch-based edits and palette transfer. In particular, we show that color palettes can make the output of the diffusion model easier to control and interpret. We first establish that existing image restoration methods are not sufficient, such as JPEG noise reduction models. We then demonstrate that our model can generate natural images that respect the color palette the user asked for. For palette transfer, we propose a method based on weighted bipartite matching. We then show that our model generates plausible images even after extreme palette transfers, respecting user query. Our method can optionally condition on the source texture in part or all of the image. In doing so, we overcome a common problem in existing image colorization methods that are unable to produce colors with a different luminance than the input. We evaluate several possibilities for texture conditioning and their trade-offs, including luminance, image gradients, and thresholded gradients, the latter of which performed best in maintaining texture and color control simultaneously. Our method can be usefully extended to another practical edit: recoloring patches of an image while respecting the source texture. Our procedure is supported by several qualitative and quantitative evaluations.
Authors: Yu-Huan Wu, Shi-Chen Zhang, Yun Liu, Le Zhang, Xin Zhan, Daquan Zhou, Jiashi Feng, Ming-Ming Cheng, Liangli Zhen
Abstract: Semantic segmentation tasks naturally require high-resolution information for pixel-wise segmentation and global context information for class prediction. While existing vision transformers demonstrate promising performance, they often utilize high-resolution context modeling, resulting in a computational bottleneck. In this work, we challenge conventional wisdom and introduce the Low-Resolution Self-Attention (LRSA) mechanism to capture global context at a significantly reduced computational cost, i.e., FLOPs. Our approach involves computing self-attention in a fixed low-resolution space regardless of the input image's resolution, with additional 3x3 depth-wise convolutions to capture fine details in the high-resolution space. We demonstrate the effectiveness of our LRSA approach by building the LRFormer, a vision transformer with an encoder-decoder structure. Extensive experiments on the ADE20K, COCO-Stuff, and Cityscapes datasets demonstrate that LRFormer outperforms state-of-the-art models. he code is available at https://github.com/yuhuan-wu/LRFormer.
Authors: Bohai Gu, Yongsheng Yu, Heng Fan, Libo Zhang
Abstract: Video inpainting has been challenged by complex scenarios like large movements and low-light conditions. Current methods, including emerging diffusion models, face limitations in quality and efficiency. This paper introduces the Flow-Guided Diffusion model for Video Inpainting (FGDVI), a novel approach that significantly enhances temporal consistency and inpainting quality via reusing an off-the-shelf image generation diffusion model. We employ optical flow for precise one-step latent propagation and introduces a model-agnostic flow-guided latent interpolation technique. This technique expedites denoising, seamlessly integrating with any Video Diffusion Model (VDM) without additional training. Our FGDVI demonstrates a remarkable 10% improvement in flow warping error E_warp over existing state-of-the-art methods. Our comprehensive experiments validate superior performance of FGDVI, offering a promising direction for advanced video inpainting. The code and detailed results will be publicly available in https://github.com/NevSNev/FGDVI.
Authors: Yixin Zhang, Shen Zhao, Hanxue Gu, Maciej A. Mazurowski
Abstract: Deep neural networks (DNNs) have demonstrated exceptional performance across various image segmentation tasks. However, the process of preparing datasets for training segmentation DNNs is both labor-intensive and costly, as it typically requires pixel-level annotations for each object of interest. To mitigate this challenge, alternative approaches such as using weak labels (e.g., bounding boxes or scribbles) or less precise (noisy) annotations can be employed. Noisy and weak labels are significantly quicker to generate, allowing for more annotated images within the same time frame. However, the potential decrease in annotation quality may adversely impact the segmentation performance of the resulting model. In this study, we conducted a comprehensive cost-effectiveness evaluation on six variants of annotation strategies (9~10 sub-variants in total) across 4 datasets and conclude that the common practice of precisely outlining objects of interest is virtually never the optimal approach when annotation budget is limited. Both noisy and weak annotations showed usage cases that yield similar performance to the perfectly annotated counterpart, yet had significantly better cost-effectiveness. We hope our findings will help researchers be aware of the different available options and use their annotation budgets more efficiently, especially in cases where accurately acquiring labels for target objects is particularly costly. Our code will be made available on https://github.com/yzluka/AnnotationEfficiency2D.
Authors: Riccardo Majellaro, Jonathan Collu, Aske Plaat, Thomas M. Moerland
Abstract: Extracting structured representations from raw visual data is an important and long-standing challenge in machine learning. Recently, techniques for unsupervised learning of object-centric representations have raised growing interest. In this context, enhancing the robustness of the latent features can improve the efficiency and effectiveness of the training of downstream tasks. A promising step in this direction is to disentangle the factors that cause variation in the data. Previously, Invariant Slot Attention disentangled position, scale, and orientation from the remaining features. Extending this approach, we focus on separating the shape and texture components. In particular, we propose a novel architecture that biases object-centric models toward disentangling shape and texture components into two non-overlapping subsets of the latent space dimensions. These subsets are known a priori, hence before the training process. Experiments on a range of object-centric benchmarks reveal that our approach achieves the desired disentanglement while also numerically improving baseline performance in most cases. In addition, we show that our method can generate novel textures for a specific object or transfer textures between objects with distinct shapes.
Authors: Yunusa Haruna, Shiyin Qin, Abdulrahman Hamman Adama Chukkol, Isah Bello, Adamu Lawan
Abstract: Detecting objects across varying scales is still a challenge in computer vision, particularly in agricultural applications like Rice Leaf Disease (RLD) detection, where objects exhibit significant scale variations (SV). Conventional object detection (OD) like Faster R-CNN, SSD, and YOLO methods often fail to effectively address SV, leading to reduced accuracy and missed detections. To tackle this, we propose SaRPFF (Self-Attention with Register-based Pyramid Feature Fusion), a novel module designed to enhance multi-scale object detection. SaRPFF integrates 2D-Multi-Head Self-Attention (MHSA) with Register tokens, improving feature interpretability by mitigating artifacts within MHSA. Additionally, it integrates efficient attention atrous convolutions into the pyramid feature fusion and introduce a deconvolutional layer for refined up-sampling. We evaluate SaRPFF on YOLOv7 using the MRLD and COCO datasets. Our approach demonstrates a +2.61% improvement in Average Precision (AP) on the MRLD dataset compared to the baseline FPN method in YOLOv7. Furthermore, SaRPFF outperforms other FPN variants, including BiFPN, NAS-FPN, and PANET, showcasing its versatility and potential to advance OD techniques. This study highlights SaRPFF effectiveness in addressing SV challenges and its adaptability across FPN-based OD models.
Authors: Xiaoyang Lyu, Yang-Tian Sun, Yi-Hua Huang, Xiuzhe Wu, Ziyi Yang, Yilun Chen, Jiangmiao Pang, Xiaojuan Qi
Abstract: In this paper, we present an implicit surface reconstruction method with 3D Gaussian Splatting (3DGS), namely 3DGSR, that allows for accurate 3D reconstruction with intricate details while inheriting the high efficiency and rendering quality of 3DGS. The key insight is incorporating an implicit signed distance field (SDF) within 3D Gaussians to enable them to be aligned and jointly optimized. First, we introduce a differentiable SDF-to-opacity transformation function that converts SDF values into corresponding Gaussians' opacities. This function connects the SDF and 3D Gaussians, allowing for unified optimization and enforcing surface constraints on the 3D Gaussians. During learning, optimizing the 3D Gaussians provides supervisory signals for SDF learning, enabling the reconstruction of intricate details. However, this only provides sparse supervisory signals to the SDF at locations occupied by Gaussians, which is insufficient for learning a continuous SDF. Then, to address this limitation, we incorporate volumetric rendering and align the rendered geometric attributes (depth, normal) with those derived from 3D Gaussians. This consistency regularization introduces supervisory signals to locations not covered by discrete 3D Gaussians, effectively eliminating redundant surfaces outside the Gaussian sampling range. Our extensive experimental results demonstrate that our 3DGSR method enables high-quality 3D surface reconstruction while preserving the efficiency and rendering quality of 3DGS. Besides, our method competes favorably with leading surface reconstruction techniques while offering a more efficient learning process and much better rendering qualities. The code will be available at https://github.com/CVMI-Lab/3DGSR.
Authors: Xinyue Wei, Kai Zhang, Sai Bi, Hao Tan, Fujun Luan, Valentin Deschaintre, Kalyan Sunkavalli, Hao Su, Zexiang Xu
Abstract: We propose MeshLRM, a novel LRM-based approach that can reconstruct a high-quality mesh from merely four input images in less than one second. Different from previous large reconstruction models (LRMs) that focus on NeRF-based reconstruction, MeshLRM incorporates differentiable mesh extraction and rendering within the LRM framework. This allows for end-to-end mesh reconstruction by fine-tuning a pre-trained NeRF LRM with mesh rendering. Moreover, we improve the LRM architecture by simplifying several complex designs in previous LRMs. MeshLRM's NeRF initialization is sequentially trained with low- and high-resolution images; this new LRM training strategy enables significantly faster convergence and thereby leads to better quality with less compute. Our approach achieves state-of-the-art mesh reconstruction from sparse-view inputs and also allows for many downstream applications, including text-to-3D and single-image-to-3D generation. Project page: https://sarahweiii.github.io/meshlrm/
Authors: Abhishek Aich, Yumin Suh, Samuel Schulter, Manmohan Chandraker
Abstract: A powerful architecture for universal segmentation relies on transformers that encode multi-scale image features and decode object queries into mask predictions. With efficiency being a high priority for scaling such models, we observed that the state-of-the-art method Mask2Former uses 50% of its compute only on the transformer encoder. This is due to the retention of a full-length token-level representation of all backbone feature scales at each encoder layer. With this observation, we propose a strategy termed PROgressive Token Length SCALing for Efficient transformer encoders (PRO-SCALE) that can be plugged-in to the Mask2Former segmentation architecture to significantly reduce the computational cost. The underlying principle of PRO-SCALE is: progressively scale the length of the tokens with the layers of the encoder. This allows PRO-SCALE to reduce computations by a large margin with minimal sacrifice in performance (~52% encoder and ~27% overall GFLOPs reduction with no drop in performance on COCO dataset). Experiments conducted on public benchmarks demonstrates PRO-SCALE's flexibility in architectural configurations, and exhibits potential for extension beyond the settings of segmentation tasks to encompass object detection. Code here: https://github.com/abhishekaich27/proscale-pytorch
Authors: Haoran Chen, Micah Goldblum, Zuxuan Wu, Yu-Gang Jiang
Abstract: Continual learning, also known as lifelong learning or incremental learning, refers to the process by which a model learns from a stream of incoming data over time. A common problem in continual learning is the classification layer's bias towards the most recent task. Traditionally, methods have relied on incorporating data from past tasks during training to mitigate this issue. However, the recent shift in continual learning to memory-free environments has rendered these approaches infeasible. In this study, we propose a solution focused on the testing phase. We first introduce a simple Out-of-Task Detection method, OTD, designed to accurately identify samples from past tasks during testing. Leveraging OTD, we then propose: (1) an Adaptive Retention mechanism for dynamically tuning the classifier layer on past task data; (2) an Adaptive Correction mechanism for revising predictions when the model classifies data from previous tasks into classes from the current task. We name our approach Adaptive Retention & Correction (ARC). While designed for memory-free environments, ARC also proves effective in memory-based settings. Extensive experiments show that our proposed method can be plugged in to virtually any existing continual learning approach without requiring any modifications to its training procedure. Specifically, when integrated with state-of-the-art approaches, ARC achieves an average performance increase of 2.7% and 2.6% on the CIFAR-100 and Imagenet-R datasets, respectively.
Authors: Ibne Farabi Shihab, Sudesh Ramesh Bhagat, Anuj Sharma
Abstract: This study aims to compare the effectiveness of a robust ensemble model with the state-of-the-art ONE-PEACE Large Language Model (LLM) for accurate detection of sidewalks. Accurate sidewalk detection is crucial in improving road safety and urban planning. The study evaluated the model's performance on Cityscapes, Ade20k, and the Boston Dataset. The results showed that the ensemble model performed better than the individual models, achieving mean Intersection Over Union (mIOU) scores of 93.1\%, 90.3\%, and 90.6\% on these datasets under ideal conditions. Additionally, the ensemble model maintained a consistent level of performance even in challenging conditions such as Salt-and-Pepper and Speckle noise, with only a gradual decrease in efficiency observed. On the other hand, the ONE-PEACE LLM performed slightly better than the ensemble model in ideal scenarios but experienced a significant decline in performance under noisy conditions. These findings demonstrate the robustness and reliability of the ensemble model, making it a valuable asset for improving urban infrastructure related to road safety and curb space management. This study contributes positively to the broader context of urban health and mobility.
Authors: Tianren Ma, Lingxi Xie, Yunjie Tian, Boyu Yang, Qixiang Ye
Abstract: Aligning vision and language concepts at a finer level remains an essential topic of multimodal large language models (MLLMs), particularly for tasks such as referring and grounding. Existing methods, such as proxy encoding and geometry encoding, incorporate additional syntax to encode spatial information, imposing extra burdens when communicating between language and vision modules. In this study, we propose ClawMachine, offering a new methodology that explicitly notates each entity using token collectives groups of visual tokens that collaboratively represent higher level semantics. A hybrid perception mechanism is also explored to perceive and understand scenes from both discrete and continuous spaces. Our method unifies the prompt and answer of visual referential tasks without using additional syntax. By leveraging a joint vision-language vocabulary, ClawMachine further integrates referring and grounding in an auto-regressive manner, demonstrating great potential with scaled-up pre-training data. Experiments show that ClawMachine achieves superior performance on scene-level and referential understanding tasks with higher efficiency. It also exhibits the potential to integrate multi-source information for complex visual reasoning, which is beyond the capability of many MLLMs. Our code is available at github.com/martian422/ClawMachine.
Authors: Haoru Wang, Wentao Zhu, Luyi Miao, Yishu Xu, Feng Gao, Qi Tian, Yizhou Wang
Abstract: Human motion generation is a critical task with a wide range of applications. Achieving high realism in generated motions requires naturalness, smoothness, and plausibility. Despite rapid advancements in the field, current generation methods often fall short of these goals. Furthermore, existing evaluation metrics typically rely on ground-truth-based errors, simple heuristics, or distribution distances, which do not align well with human perceptions of motion quality. In this work, we propose a data-driven approach to bridge this gap by introducing a large-scale human perceptual evaluation dataset, MotionPercept, and a human motion critic model, MotionCritic, that capture human perceptual preferences. Our critic model offers a more accurate metric for assessing motion quality and could be readily integrated into the motion generation pipeline to enhance generation quality. Extensive experiments demonstrate the effectiveness of our approach in both evaluating and improving the quality of generated human motions by aligning with human perceptions. Code and data are publicly available at https://motioncritic.github.io/.
Authors: Yaoyao Zhu, Xiuding Cai, Yingkai Wang, Dong Miao, Zhongliang Fu, Xu Luo
Abstract: Deep learning models excel in computer vision tasks but often fail to generalize to out-of-distribution (OOD) domains. Invariant Risk Minimization (IRM) aims to address OOD generalization by learning domain-invariant features. However, IRM struggles with datasets exhibiting significant diversity shifts. While data augmentation methods like Mixup and Semantic Data Augmentation (SDA) enhance diversity, they risk over-augmentation and label instability. To address these challenges, we propose a domain-shared Semantic Data Augmentation (SDA) module, a novel implementation of Variance Risk Minimization (VRM) designed to enhance dataset diversity while maintaining label consistency. We further provide a Rademacher complexity analysis, establishing a tighter generalization error bound compared to baseline methods. Extensive evaluations on OOD benchmarks, including PACS, VLCS, OfficeHome, and TerraIncognita, demonstrate consistent performance improvements over state-of-the-art domain generalization methods.
Authors: Joel Sol, Jamil Fayyad, Shadi Alijani, Homayoun Najjaran
Abstract: Deformation detection is vital for enabling accurate assessment and prediction of structural changes in materials, ensuring timely and effective interventions to maintain safety and integrity. Automating deformation detection through computer vision is crucial for efficient monitoring, but it faces significant challenges in creating a comprehensive dataset of both deformed and non-deformed objects, which can be difficult to obtain in many scenarios. In this paper, we introduce a novel framework for generating controlled synthetic data that simulates deformed objects. This approach allows for the realistic modeling of object deformations under various conditions. Our framework integrates an intelligent adapter network that facilitates sim-to-real domain adaptation, enhancing classification results without requiring real data from deformed objects. We conduct experiments on domain adaptation and classification tasks and demonstrate that our framework improves sim-to-real classification results compared to simulation baseline.
Authors: Bryan Wong, Mun Yong Yi
Abstract: Multiple instance learning (MIL) has become a preferred method for gigapixel whole slide image (WSI) classification without requiring patch-level annotations. Current MIL research primarily relies on embedding-based approaches, which extract patch features using a pre-trained feature extractor and aggregate them for slide-level prediction. Despite the critical role of feature extraction, there is limited guidance on selecting optimal feature extractors to maximize WSI performance. This study addresses this gap by systematically evaluating MIL feature extractors across three dimensions: pre-training dataset, backbone model, and pre-training method. Extensive experiments were conducted on two public WSI datasets (TCGA-NSCLC and Camelyon16) using four state-of-the-art (SOTA) MIL models. Our findings reveal that: 1) selecting a robust self-supervised learning (SSL) method has a greater impact on performance than relying solely on an in-domain pre-training dataset; 2) prioritizing Transformer-based backbones with deeper architectures over CNN-based models; and 3) using larger, more diverse pre-training datasets significantly enhances classification outcomes. We hope that these insights can provide practical guidance for optimizing WSI classification and explain the reasons behind the performance advantages of the current SOTA pathology foundation models. Furthermore, this work may inform the development of more effective pathology foundation models. Our code is publicly available at https://github.com/bryanwong17/MIL-Feature-Extractor-Selection
URLs: https://github.com/bryanwong17/MIL-Feature-Extractor-Selection
Authors: Qizhou Chen, Taolin Zhang, Chengyu Wang, Xiaofeng He, Dakan Wang, Tingting Liu
Abstract: Model editing aims to correct outdated or erroneous knowledge in large models without costly retraining. Recent research discovered that the mid-layer representation of the subject's final token in a prompt has a strong influence on factual predictions, and developed Large Language Model (LLM) editing techniques based on this observation. However, for Vision-LLMs (VLLMs), how visual representations impact the predictions from a decoder-only language model remains largely unexplored. To the best of our knowledge, model editing for VLLMs has not been extensively studied in the literature. In this work, we employ the contribution allocation and noise perturbation methods to measure the contributions of visual representations for token predictions. Our attribution analysis shows that visual representations in mid-to-later layers that are highly relevant to the prompt contribute significantly to predictions. Based on these insights, we propose VisEdit, a novel model editor for VLLMs that effectively corrects knowledge by editing intermediate visual representations in regions important to the edit prompt. We evaluated VisEdit using multiple VLLM backbones and public VLLM editing benchmark datasets. The results show the superiority of VisEdit over the strong baselines adapted from existing state-of-the-art editors for LLMs.
Authors: Yunfeng Diao, Baiqi Wu, Ruixuan Zhang, Ajian Liu, Xingxing Wei, Meng Wang, He Wang
Abstract: Skeletal sequences, as well-structured representations of human behaviors, play a vital role in Human Activity Recognition (HAR). The transferability of adversarial skeletal sequences enables attacks in real-world HAR scenarios, such as autonomous driving, intelligent surveillance, and human-computer interactions. However, most existing skeleton-based HAR (S-HAR) attacks are primarily designed for white-box scenarios and exhibit weak adversarial transferability. Therefore, they cannot be considered true transfer-based S-HAR attacks. More importantly, the reason for this failure remains unclear. In this paper, we study this phenomenon through the lens of loss surface, and find that its sharpness contributes to the weak transferability in S-HAR. Inspired by this observation, we assume and empirically validate that smoothening the rugged loss landscape could potentially improve adversarial transferability in S-HAR. To this end, we propose the first \textbf{T}ransfer-based \textbf{A}ttack on \textbf{S}keletal \textbf{A}ction \textbf{R}ecognition, TASAR. TASAR explores the smoothed model posterior without requiring surrogate re-training, which is achieved by a new post-train Dual Bayesian optimization strategy. Furthermore, unlike previous transfer-based attacks that treat each frame independently and overlook temporal coherence within sequences, TASAR incorporates motion dynamics into the Bayesian attack gradient, effectively disrupting the spatial-temporal coherence of S-HARs. To exhaustively evaluate the effectiveness of existing methods and our method, we build the first large-scale robust S-HAR benchmark, comprising 7 S-HAR models, 10 attack methods, 3 S-HAR datasets and 2 defense methods. Extensive results demonstrate the superiority of TASAR. Our benchmark enables easy comparisons for future studies, with the code available in the supplementary material.
Authors: Shijing Wang, Yaping Huang, Jun Xie, Yi Tian, Feng Chen, Zhepeng Wang
Abstract: Achieving accurate and reliable gaze predictions in complex and diverse environments remains challenging. Fortunately, it is straightforward to access diverse gaze datasets in real-world applications. We discover that training these datasets jointly can significantly improve the generalization of gaze estimation, which is overlooked in previous works. However, due to the inherent distribution shift across different datasets, simply mixing multiple dataset decreases the performance in the original domain despite gaining better generalization abilities. To address the problem of ``cross-dataset gaze estimation'', we propose a novel Evidential Inter-intra Fusion EIF framework, for training a cross-dataset model that performs well across all source and unseen domains. Specifically, we build independent single-dataset branches for various datasets where the data space is partitioned into overlapping subspaces within each dataset for local regression, and further create a cross-dataset branch to integrate the generalizable features from single-dataset branches. Furthermore, evidential regressors based on the Normal and Inverse-Gamma (NIG) distribution are designed to additionally provide uncertainty estimation apart from predicting gaze. Building upon this foundation, our proposed framework achieves both intra-evidential fusion among multiple local regressors within each dataset and inter-evidential fusion among multiple branches by Mixture \textbfof Normal Inverse-Gamma (MoNIG distribution. Experiments demonstrate that our method consistently achieves notable improvements in both source domains and unseen domains.
Authors: Jiatao Gu, Yuyang Wang, Yizhe Zhang, Qihang Zhang, Dinghuai Zhang, Navdeep Jaitly, Josh Susskind, Shuangfei Zhai
Abstract: Diffusion models have become the dominant approach for visual generation. They are trained by denoising a Markovian process which gradually adds noise to the input. We argue that the Markovian property limits the model's ability to fully utilize the generation trajectory, leading to inefficiencies during training and inference. In this paper, we propose DART, a transformer-based model that unifies autoregressive (AR) and diffusion within a non-Markovian framework. DART iteratively denoises image patches spatially and spectrally using an AR model that has the same architecture as standard language models. DART does not rely on image quantization, which enables more effective image modeling while maintaining flexibility. Furthermore, DART seamlessly trains with both text and image data in a unified model. Our approach demonstrates competitive performance on class-conditioned and text-to-image generation tasks, offering a scalable, efficient alternative to traditional diffusion models. Through this unified framework, DART sets a new benchmark for scalable, high-quality image synthesis.
Authors: Dabbrata Das, Argho Deb Das, Farhan Sadaf
Abstract: Estimating depth from a single 2D image is a challenging task due to the lack of stereo or multi-view data, which are typically required for depth perception. In state-of-the-art architectures, the main challenge is to efficiently capture complex objects and fine-grained details, which are often difficult to predict. This paper introduces a novel deep learning-based approach using an enhanced encoder-decoder architecture, where the Inception-ResNet-v2 model serves as the encoder. This is the first instance of utilizing Inception-ResNet-v2 as an encoder for monocular depth estimation, demonstrating improved performance over previous models. It incorporates multi-scale feature extraction to enhance depth prediction accuracy across various object sizes and distances. We propose a composite loss function comprising depth loss, gradient edge loss, and Structural Similarity Index Measure (SSIM) loss, with fine-tuned weights to optimize the weighted sum, ensuring a balance across different aspects of depth estimation. Experimental results on the KITTI dataset show that our model achieves a significantly faster inference time of 0.019 seconds, outperforming vision transformers in efficiency while maintaining good accuracy. On the NYU Depth V2 dataset, the model establishes state-of-the-art performance, with an Absolute Relative Error (ARE) of 0.064, a Root Mean Square Error (RMSE) of 0.228, and an accuracy of 89.3% for $\delta$ < 1.25. These metrics demonstrate that our model can accurately and efficiently predict depth even in challenging scenarios, providing a practical solution for real-time applications.
Authors: Chengpeng Wang, Li Chen, Lili Wang, Zhaofan Li, Xuebin Lv
Abstract: Facial expression recognition faces challenges where labeled significant features in datasets are mixed with unlabeled redundant ones. In this paper, we introduce Cross Similarity Attention (CSA) to mine richer intrinsic information from image pairs, overcoming a limitation when the Scaled Dot-Product Attention of ViT is directly applied to calculate the similarity between two different images. Based on CSA, we simultaneously minimize intra-class differences and maximize inter-class differences at the fine-grained feature level through interactions among multiple branches. Contrastive residual distillation is utilized to transfer the information learned in the cross module back to the base network. We ingeniously design a four-branch centrally symmetric network, named Quadruplet Cross Similarity (QCS), which alleviates gradient conflicts arising from the cross module and achieves balanced and stable training. It can adaptively extract discriminative features while isolating redundant ones. The cross-attention modules exist during training, and only one base branch is retained during inference, resulting in no increase in inference time. Extensive experiments show that our proposed method achieves state-of-the-art performance on several FER datasets.
Authors: Xianghui Yang, Huiwen Shi, Bowen Zhang, Fan Yang, Jiacheng Wang, Hongxu Zhao, Xinhai Liu, Xinzhou Wang, Qingxiang Lin, Jiaao Yu, Lifu Wang, Jing Xu, Zebin He, Zhuo Chen, Sicong Liu, Junta Wu, Yihang Lian, Shaoxiong Yang, Yuhong Liu, Yong Yang, Di Wang, Jie Jiang, Chunchao Guo
Abstract: While 3D generative models have greatly improved artists' workflows, the existing diffusion models for 3D generation suffer from slow generation and poor generalization. To address this issue, we propose a two-stage approach named Hunyuan3D 1.0 including a lite version and a standard version, that both support text- and image-conditioned generation. In the first stage, we employ a multi-view diffusion model that efficiently generates multi-view RGB in approximately 4 seconds. These multi-view images capture rich details of the 3D asset from different viewpoints, relaxing the tasks from single-view to multi-view reconstruction. In the second stage, we introduce a feed-forward reconstruction model that rapidly and faithfully reconstructs the 3D asset given the generated multi-view images in approximately 7 seconds. The reconstruction network learns to handle noises and in-consistency introduced by the multi-view diffusion and leverages the available information from the condition image to efficiently recover the 3D structure. Our framework involves the text-to-image model, i.e., Hunyuan-DiT, making it a unified framework to support both text- and image-conditioned 3D generation. Our standard version has 3x more parameters than our lite and other existing model. Our Hunyuan3D 1.0 achieves an impressive balance between speed and quality, significantly reducing generation time while maintaining the quality and diversity of the produced assets.
Authors: Mehmet Can Yavuz, Yang Yang
Abstract: In biomedical imaging analysis, the dichotomy between 2D and 3D data presents a significant challenge. While 3D volumes offer superior real-world applicability, they are less available for each modality and not easy to train in large scale, whereas 2D samples are abundant but less comprehensive. This paper introduces Cross-D Conv operation, a novel approach that bridges the dimensional gap by learning the phase shifting in the Fourier domain. Our method enables seamless weight transfer between 2D and 3D convolution operations, effectively facilitating cross-dimensional learning. The proposed architecture leverages the abundance of 2D training data to enhance 3D model performance, offering a practical solution to the multimodal data scarcity challenge in 3D medical model pretraining. Experimental validation on the RadImagenet (2D) and multimodal volumetric sets demonstrates that our approach achieves comparable or superior performance in feature quality assessment. The enhanced convolution operation presents new opportunities for developing efficient classification and segmentation models in medical imaging. This work represents an advancement in cross-dimensional and multimodal medical image analysis, offering a robust framework for utilizing 2D priors in 3D model pretraining while maintaining computational efficiency of 2D training.
Authors: Amirabbas Afzali, Borna Khodabandeh, Ali Rasekh, Mahyar JafariNodeh, Sepehr kazemi, Simon Gottschalk
Abstract: Contrastive learning models have demonstrated impressive abilities to capture semantic similarities by aligning representations in the embedding space. However, their performance can be limited by the quality of the training data and its inherent biases. While Reinforcement Learning from Human Feedback (RLHF) and Direct Preference Optimization (DPO) have been applied to generative models to align them with human preferences, their use in contrastive learning has yet to be explored. This paper introduces a novel method for training contrastive learning models using Preference Optimization (PO) to break down complex concepts. Our method systematically aligns model behavior with desired preferences, enhancing performance on the targeted task. In particular, we focus on enhancing model robustness against typographic attacks, commonly seen in contrastive models like CLIP. We further apply our method to disentangle gender understanding and mitigate gender biases, offering a more nuanced control over these sensitive attributes. Our experiments demonstrate that models trained using PO outperform standard contrastive learning techniques while retaining their ability to handle adversarial challenges and maintain accuracy on other downstream tasks. This makes our method well-suited for tasks requiring fairness, robustness, and alignment with specific preferences. We evaluate our method on several vision-language tasks, tackling challenges such as typographic attacks. Additionally, we explore the model's ability to disentangle gender concepts and mitigate gender bias, showcasing the versatility of our approach.
Authors: Hanning Yuan, Zhihui Zhang, Qi Guo, Lianhua Chi, Sijie Ruan, Jinhui Pang, Xiaoshuai Hao
Abstract: Multi-view contrastive clustering (MVCC) has gained significant attention for generating consistent clustering structures from multiple views through contrastive learning. However, most existing MVCC methods create cross-views by combining any two views, leading to a high volume of unreliable pairs. Furthermore, these approaches often overlook discrepancies in multi-view representations, resulting in representation degeneration. To address these challenges, we introduce a novel model called Dual-Weighted Contrastive Learning (DWCL) for Multi-View Clustering. Specifically, to reduce the impact of unreliable cross-views, we introduce an innovative Best-Other (B-O) contrastive mechanism that enhances the representation of individual views at a low computational cost. Furthermore, we develop a dual weighting strategy that combines a view quality weight, reflecting the quality of each view, with a view discrepancy weight. This approach effectively mitigates representation degeneration by downplaying cross-views that are both low in quality and high in discrepancy. We theoretically validate the efficiency of the B-O contrastive mechanism and the effectiveness of the dual weighting strategy. Extensive experiments demonstrate that DWCL outperforms previous methods across eight multi-view datasets, showcasing superior performance and robustness in MVCC. Specifically, our method achieves absolute accuracy improvements of 5.4\% and 5.6\% compared to state-of-the-art methods on the Caltech6V7 and MSRCv1 datasets, respectively.
Authors: Ziyang Chen, Prem Seetharaman, Bryan Russell, Oriol Nieto, David Bourgin, Andrew Owens, Justin Salamon
Abstract: Generating sound effects for videos often requires creating artistic sound effects that diverge significantly from real-life sources and flexible control in the sound design. To address this problem, we introduce MultiFoley, a model designed for video-guided sound generation that supports multimodal conditioning through text, audio, and video. Given a silent video and a text prompt, MultiFoley allows users to create clean sounds (e.g., skateboard wheels spinning without wind noise) or more whimsical sounds (e.g., making a lion's roar sound like a cat's meow). MultiFoley also allows users to choose reference audio from sound effects (SFX) libraries or partial videos for conditioning. A key novelty of our model lies in its joint training on both internet video datasets with low-quality audio and professional SFX recordings, enabling high-quality, full-bandwidth (48kHz) audio generation. Through automated evaluations and human studies, we demonstrate that MultiFoley successfully generates synchronized high-quality sounds across varied conditional inputs and outperforms existing methods. Please see our project page for video results: https://ificl.github.io/MultiFoley/
Authors: Piotr Teterwak, Kuniaki Saito, Theodoros Tsiligkaridis, Bryan A. Plummer, Kate Saenko
Abstract: Multi-Source Domain Generalization (DG) is the task of training on multiple source domains and achieving high classification performance on unseen target domains. Recent methods combine robust features from web-scale pretrained backbones with new features learned from source data, and this has dramatically improved benchmark results. However, it remains unclear if DG finetuning methods are becoming better over time, or if improved benchmark performance is simply an artifact of stronger pre-training. Prior studies have shown that perceptual similarity to pre-training data correlates with zero-shot performance, but we find the effect limited in the DG setting. Instead, we posit that having perceptually similar data in pretraining is not enough; and that it is how well these data were learned that determines performance. This leads us to introduce the Alignment Hypothesis, which states that the final DG performance will be high if and only if alignment of image and class label text embeddings is high. Our experiments confirm the Alignment Hypothesis is true, and we use it as an analysis tool of existing DG methods evaluated on DomainBed datasets by splitting evaluation data into In-pretraining (IP) and Out-of-pretraining (OOP). We show that all evaluated DG methods struggle on DomainBed-OOP, while recent methods excel on DomainBed-IP. Put together, our findings highlight the need for DG methods which can generalize beyond pretraining alignment.
Authors: Lianqing Zheng, Long Yang, Qunshu Lin, Wenjin Ai, Minghao Liu, Shouyi Lu, Jianan Liu, Hongze Ren, Jingyue Mo, Xiaokai Bai, Jie Bai, Zhixiong Ma, Xichan Zhu
Abstract: The rapid advancement of deep learning has intensified the need for comprehensive data for use by autonomous driving algorithms. High-quality datasets are crucial for the development of effective data-driven autonomous driving solutions. Next-generation autonomous driving datasets must be multimodal, incorporating data from advanced sensors that feature extensive data coverage, detailed annotations, and diverse scene representation. To address this need, we present OmniHD-Scenes, a large-scale multimodal dataset that provides comprehensive omnidirectional high-definition data. The OmniHD-Scenes dataset combines data from 128-beam LiDAR, six cameras, and six 4D imaging radar systems to achieve full environmental perception. The dataset comprises 1501 clips, each approximately 30-s long, totaling more than 450K synchronized frames and more than 5.85 million synchronized sensor data points. We also propose a novel 4D annotation pipeline. To date, we have annotated 200 clips with more than 514K precise 3D bounding boxes. These clips also include semantic segmentation annotations for static scene elements. Additionally, we introduce a novel automated pipeline for generation of the dense occupancy ground truth, which effectively leverages information from non-key frames. Alongside the proposed dataset, we establish comprehensive evaluation metrics, baseline models, and benchmarks for 3D detection and semantic occupancy prediction. These benchmarks utilize surround-view cameras and 4D imaging radar to explore cost-effective sensor solutions for autonomous driving applications. Extensive experiments demonstrate the effectiveness of our low-cost sensor configuration and its robustness under adverse conditions. Data will be released at https://www.2077ai.com/OmniHD-Scenes.
Authors: Ruibin Li, Tao Yang, Song Guo, Lei Zhang
Abstract: Despite the significant advancements, existing object removal methods struggle with incomplete removal, incorrect content synthesis and blurry synthesized regions, resulting in low success rates. Such issues are mainly caused by the lack of high-quality paired training data, as well as the self-supervised training paradigm adopted in these methods, which forces the model to in-paint the masked regions, leading to ambiguity between synthesizing the masked objects and restoring the background. To address these issues, we propose a semi-supervised learning strategy with human-in-the-loop to create high-quality paired training data, aiming to train a Robust Object Remover (RORem). We first collect 60K training pairs from open-source datasets to train an initial object removal model for generating removal samples, and then utilize human feedback to select a set of high-quality object removal pairs, with which we train a discriminator to automate the following training data generation process. By iterating this process for several rounds, we finally obtain a substantial object removal dataset with over 200K pairs. Fine-tuning the pre-trained stable diffusion model with this dataset, we obtain our RORem, which demonstrates state-of-the-art object removal performance in terms of both reliability and image quality. Particularly, RORem improves the object removal success rate over previous methods by more than 18\%. The dataset, source code and trained model are available at https://github.com/leeruibin/RORem.
Authors: Miao Rang, Zhenni Bi, Chuanjian Liu, Yehui Tang, Kai Han, Yunhe Wang
Abstract: Multimodal vision language models (VLMs) have made significant progress with the support of continuously increasing model sizes and data volumes. Running VLMs on edge devices has become a challenge for their widespread application. There are several efficient VLM efforts, but they often sacrifice linguistic capabilities to enhance multimodal abilities, or require extensive training. To address this quandary,we introduce the innovative framework of Efficient Vision Language Models with Elastic Visual Experts (Eve). By strategically incorporating adaptable visual expertise at multiple stages of training, Eve strikes a balance between preserving linguistic abilities and augmenting multimodal capabilities. This balanced approach results in a versatile model with only 1.8B parameters that delivers significant improvements in both multimodal and linguistic tasks. Notably, in configurations below 3B parameters, Eve distinctly outperforms in language benchmarks and achieves state-of-the-art results 68.87% in VLM Benchmarks. Additionally, its multimodal accuracy outstrips that of the larger 7B LLaVA-1.5 model. Our code is available at https://github.com/rangmiao/Eve.
Authors: Ryan Burgert, Yuancheng Xu, Wenqi Xian, Oliver Pilarski, Pascal Clausen, Mingming He, Li Ma, Yitong Deng, Lingxiao Li, Mohsen Mousavi, Michael Ryoo, Paul Debevec, Ning Yu
Abstract: Generative modeling aims to transform random noise into structured outputs. In this work, we enhance video diffusion models by allowing motion control via structured latent noise sampling. This is achieved by just a change in data: we pre-process training videos to yield structured noise. Consequently, our method is agnostic to diffusion model design, requiring no changes to model architectures or training pipelines. Specifically, we propose a novel noise warping algorithm, fast enough to run in real time, that replaces random temporal Gaussianity with correlated warped noise derived from optical flow fields, while preserving the spatial Gaussianity. The efficiency of our algorithm enables us to fine-tune modern video diffusion base models using warped noise with minimal overhead, and provide a one-stop solution for a wide range of user-friendly motion control: local object motion control, global camera movement control, and motion transfer. The harmonization between temporal coherence and spatial Gaussianity in our warped noise leads to effective motion control while maintaining per-frame pixel quality. Extensive experiments and user studies demonstrate the advantages of our method, making it a robust and scalable approach for controlling motion in video diffusion models. Video results are available on our webpage: https://eyeline-research.github.io/Go-with-the-Flow. Source code and model checkpoints are available on GitHub: https://github.com/Eyeline-Research/Go-with-the-Flow.
URLs: https://eyeline-research.github.io/Go-with-the-Flow., https://github.com/Eyeline-Research/Go-with-the-Flow.
Authors: Qingyun Li, Yushi Chen, Xinya Shu, Dong Chen, Xin He, Yi Yu, Xue Yang
Abstract: The multimodal language models (MLMs) based on generative pre-trained Transformer are considered powerful candidates for unifying various domains and tasks. MLMs developed for remote sensing (RS) have demonstrated outstanding performance in multiple tasks, such as visual question answering and visual grounding. In addition to visual grounding that detects specific objects corresponded to given instruction, aerial detection, which detects all objects of multiple categories, is also a valuable and challenging task for RS foundation models. However, aerial detection has not been explored by existing RS MLMs because the autoregressive prediction mechanism of MLMs differs significantly from the detection outputs. In this paper, we present a simple baseline for applying MLMs to aerial detection for the first time, named LMMRotate. Specifically, we first introduce a normalization method to transform detection outputs into textual outputs to be compatible with the MLM framework. Then, we propose a evaluation method, which ensures a fair comparison between MLMs and conventional object detection models. We construct the baseline by fine-tuning open-source general-purpose MLMs and achieve impressive detection performance comparable to conventional detector. We hope that this baseline will serve as a reference for future MLM development, enabling more comprehensive capabilities for understanding RS images. Code is available at https://github.com/Li-Qingyun/mllm-mmrotate.
Authors: Pengcheng Dong, Wenbo Wan, Huaxiang Zhang, Jiande Sun
Abstract: In recent years, action recognition has received much attention and wide application due to its important role in video understanding. Most of the researches on action recognition methods focused on improving the performance via various deep learning methods rather than the classification of skeleton points. The topological modeling between skeleton points and body parts was seldom considered. Although some studies have used a data-driven approach to classify the topology of the skeleton point, the nature of the skeleton point in terms of kinematics has not been taken into consideration. Therefore, in this paper, we draw on the theory of kinematics to adapt the topological relations of the skeleton point and propose a topological relation classification based on body parts and distance from core of body. To synthesize these topological relations for action recognition, we propose a novel Hypergraph Fusion Graph Convolutional Network (HFGCN). In particular, the proposed model is able to focus on the human skeleton points and the different body parts simultaneously, and thus construct the topology, which improves the recognition accuracy obviously. We use a hypergraph to represent the categorical relationships of these skeleton points and incorporate the hypergraph into a graph convolution network to model the higher-order relationships among the skeleton points and enhance the feature representation of the network. In addition, our proposed hypergraph attention module and hypergraph graph convolution module optimize topology modeling in temporal and channel dimensions, respectively, to further enhance the feature representation of the network. We conducted extensive experiments on three widely used datasets.The results validate that our proposed method can achieve the best performance when compared with the state-of-the-art skeleton-based methods.
Authors: Liam Chalcroft, Jenny Crinion, Cathy J. Price, John Ashburner
Abstract: Self-supervised deep learning has accelerated 2D natural image analysis but remains difficult to translate into 3D MRI, where data are scarce and pre-trained 2D backbones cannot capture volumetric context. We present a sequence-invariant self-supervised framework leveraging quantitative MRI (qMRI). By simulating multiple MRI contrasts from a single 3D qMRI scan and enforcing consistent representations across these contrasts, we learn anatomy-centric rather than sequence-specific features. This yields a robust 3D encoder that performs strongly across varied tasks and protocols. Experiments on healthy brain segmentation (IXI), stroke lesion segmentation (ARC), and MRI denoising show significant gains over baseline SSL approaches, especially in low-data settings (up to +8.3% Dice, +4.2 dB PSNR). Our model also generalises effectively to unseen sites, demonstrating potential for more scalable and clinically reliable volumetric analysis. All code and trained models are publicly available.
Authors: Jos\'e Rodr\'iguez-Ortega (Nemotec, Madrid, Spain, Department of Computer Science and Artificial Intelligence, University of Granada, Granada, Spain), Siham Tabik (Department of Computer Science and Artificial Intelligence, University of Granada, Granada, Spain)
Abstract: Identifying anatomical landmarks in 3D dental models is crucial for orthodontic treatment. Manually placing these key points is complex, time-consuming, and requires expert knowledge. While some machine learning methods have been proposed for automatic tooth landmark detection in 3D Intraoral Scans (IOS), research remains limited, with no fully end-to-end approaches that avoid teeth segmentation. We propose CHaRNet (Conditioned Heatmap Regression Network), the first end-to-end deep learning method for tooth landmark detection in 3D IOS. Unlike traditional two-stage methods that segment teeth before detecting landmarks, CHaRNet directly detects landmarks on the input point cloud. It consists of four key modules: (1) a point cloud encoder, (2) a point cloud decoder with a heatmap regression head, (3) a teeth presence classification head, and (4) the innovative Conditioned Heatmap Regression (CHaR) module. The CHaR module refines landmark regression by leveraging teeth presence classification, enabling dynamic adaptation to cases with missing teeth and improving accuracy in complex dental models. We evaluate CHaRNet using five point cloud learning algorithms to validate the effectiveness of the CHaR module and test it on a clinical dataset of 1,214 annotated 3D dental models. Both the dataset and code will be publicly released to address the lack of open datasets in orthodontics, promote benchmarking, and inspire new research. CHaRNet achieves a Mean Euclidean Distance Error (MEDE) of 1.28 mm and a Mean Success Ratio (MSR) of 82.40%, demonstrating robust performance. Notably, it excels in handling irregular dental geometries, such as models with missing teeth. This end-to-end approach streamlines orthodontic workflows, improves 3D IOS analysis precision, and facilitates efficient computer-assisted treatment planning.
Authors: Boqiang Zhang, Kehan Li, Zesen Cheng, Zhiqiang Hu, Yuqian Yuan, Guanzheng Chen, Sicong Leng, Yuming Jiang, Hang Zhang, Xin Li, Peng Jin, Wenqi Zhang, Fan Wang, Lidong Bing, Deli Zhao
Abstract: In this paper, we propose VideoLLaMA3, a more advanced multimodal foundation model for image and video understanding. The core design philosophy of VideoLLaMA3 is vision-centric. The meaning of "vision-centric" is two-fold: the vision-centric training paradigm and vision-centric framework design. The key insight of our vision-centric training paradigm is that high-quality image-text data is crucial for both image and video understanding. Instead of preparing massive video-text datasets, we focus on constructing large-scale and high-quality image-text datasets. VideoLLaMA3 has four training stages: 1) Vision Encoder Adaptation, which enables vision encoder to accept images of variable resolutions as input; 2) Vision-Language Alignment, which jointly tunes the vision encoder, projector, and LLM with large-scale image-text data covering multiple types (including scene images, documents, charts) as well as text-only data. 3) Multi-task Fine-tuning, which incorporates image-text SFT data for downstream tasks and video-text data to establish a foundation for video understanding. 4) Video-centric Fine-tuning, which further improves the model's capability in video understanding. As for the framework design, to better capture fine-grained details in images, the pretrained vision encoder is adapted to encode images of varying sizes into vision tokens with corresponding numbers, rather than a fixed number of tokens. For video inputs, we reduce the number of vision tokens according to their similarity so that the representation of videos will be more precise and compact. Benefit from vision-centric designs, VideoLLaMA3 achieves compelling performances in both image and video understanding benchmarks.
Authors: Matthew Gwilliam, Han Cai, Di Wu, Abhinav Shrivastava, Zhiyu Cheng
Abstract: We propose Inner Loop Feedback (ILF), a novel approach to accelerate diffusion models' inference. ILF trains a lightweight module to predict future features in the denoising process by leveraging the outputs from a chosen diffusion backbone block at a given time step. This approach exploits two key intuitions; (1) the outputs of a given block at adjacent time steps are similar, and (2) performing partial computations for a step imposes a lower burden on the model than skipping the step entirely. Our method is highly flexible, since we find that the feedback module itself can simply be a block from the diffusion backbone, with all settings copied. Its influence on the diffusion forward can be tempered with a learnable scaling factor from zero initialization. We train this module using distillation losses; however, unlike some prior work where a full diffusion backbone serves as the student, our model freezes the backbone, training only the feedback module. While many efforts to optimize diffusion models focus on achieving acceptable image quality in extremely few steps (1-4 steps), our emphasis is on matching best case results (typically achieved in 20 steps) while significantly reducing runtime. ILF achieves this balance effectively, demonstrating strong performance for both class-to-image generation with diffusion transformer (DiT) and text-to-image generation with DiT-based PixArt-alpha and PixArt-sigma. The quality of ILF's 1.7x-1.8x speedups are confirmed by FID, CLIP score, CLIP Image Quality Assessment, ImageReward, and qualitative comparisons. Project information is available at https://mgwillia.github.io/ilf.
Authors: Finn Behrendt, Debayan Bhattacharya, Robin Mieling, Lennart Maack, Julia Kr\"uger, Roland Opfer, Alexander Schlaefer
Abstract: The application of supervised models to clinical screening tasks is challenging due to the need for annotated data for each considered pathology. Unsupervised Anomaly Detection (UAD) is an alternative approach that aims to identify any anomaly as an outlier from a healthy training distribution. A prevalent strategy for UAD in brain MRI involves using generative models to learn the reconstruction of healthy brain anatomy for a given input image. As these models should fail to reconstruct unhealthy structures, the reconstruction errors indicate anomalies. However, a significant challenge is to balance the accurate reconstruction of healthy anatomy and the undesired replication of abnormal structures. While diffusion models have shown promising results with detailed and accurate reconstructions, they face challenges in preserving intensity characteristics, resulting in false positives. We propose conditioning the denoising process of diffusion models with additional information derived from a latent representation of the input image. We demonstrate that this conditioning allows for accurate and local adaptation to the general input intensity distribution while avoiding the replication of unhealthy structures. We compare the novel approach to different state-of-the-art methods and for different data sets. Our results show substantial improvements in the segmentation performance, with the Dice score improved by 11.9%, 20.0%, and 44.6%, for the BraTS, ATLAS and MSLUB data sets, respectively, while maintaining competitive performance on the WMH data set. Furthermore, our results indicate effective domain adaptation across different MRI acquisitions and simulated contrasts, an important attribute for general anomaly detection methods. The code for our work is available at https://github.com/FinnBehrendt/Conditioned-Diffusion-Models-UAD
URLs: https://github.com/FinnBehrendt/Conditioned-Diffusion-Models-UAD
Authors: Eunjee Choi, Jong-Kook Kim
Abstract: Detecting fake news has received a lot of attention. Many previous methods concatenate independently encoded unimodal data, ignoring the benefits of integrated multimodal information. Also, the absence of specialized feature extraction for text and images further limits these methods. This paper introduces an end-to-end model called TT-BLIP that applies the bootstrapping language-image pretraining for unified vision-language understanding and generation (BLIP) for three types of information: BERT and BLIPTxt for text, ResNet and BLIPImg for images, and bidirectional BLIP encoders for multimodal information. The Multimodal Tri-Transformer fuses tri-modal features using three types of multi-head attention mechanisms, ensuring integrated modalities for enhanced representations and improved multimodal data analysis. The experiments are performed using two fake news datasets, Weibo and Gossipcop. The results indicate TT-BLIP outperforms the state-of-the-art models.
Authors: Heejun Lee, Geon Park, Youngwan Lee, Jaduk Suh, Jina Kim, Wonyoung Jeong, Bumsik Kim, Hyemin Lee, Myeongjae Jeon, Sung Ju Hwang
Abstract: In modern large language models (LLMs), increasing the context length is crucial for improving comprehension and coherence in long-context, multi-modal, and retrieval-augmented language generation. While many recent transformer models attempt to extend their context length over a million tokens, they remain impractical due to the quadratic time and space complexities. Although recent works on linear and sparse attention mechanisms can achieve this goal, their real-world applicability is often limited by the need to re-train from scratch and significantly worse performance. In response, we propose a novel approach, Hierarchically Pruned Attention (HiP), which reduces the time complexity of the attention mechanism to $O(T \log T)$ and the space complexity to $O(T)$, where $T$ is the sequence length. We notice a pattern in the attention scores of pretrained LLMs where tokens close together tend to have similar scores, which we call ``attention locality''. Based on this observation, we utilize a novel tree-search-like algorithm that estimates the top-$k$ key tokens for a given query on the fly, which is mathematically guaranteed to have better performance than random attention pruning. In addition to improving the time complexity of the attention mechanism, we further optimize GPU memory usage by implementing KV cache offloading, which stores only $O(\log T)$ tokens on the GPU while maintaining similar decoding throughput. Experiments on benchmarks show that HiP, with its training-free nature, significantly reduces both prefill and decoding latencies, as well as memory usage, while maintaining high-quality generation with minimal degradation. HiP enables pretrained LLMs to scale up to millions of tokens on commodity GPUs, potentially unlocking long-context LLM applications previously deemed infeasible.
Authors: Mengdan Zhu, Raasikh Kanjiani, Jiahui Lu, Andrew Choi, Qirui Ye, Liang Zhao
Abstract: Deep generative models like VAEs and diffusion models have advanced various generation tasks by leveraging latent variables to learn data distributions and generate high-quality samples. Despite the field of explainable AI making strides in interpreting machine learning models, understanding latent variables in generative models remains challenging. This paper introduces \textit{LatentExplainer}, a framework for automatically generating semantically meaningful explanations of latent variables in deep generative models. \textit{LatentExplainer} tackles three main challenges: inferring the meaning of latent variables, aligning explanations with inductive biases, and handling varying degrees of explainability. Our approach perturbs latent variables, interpreting changes in generated data, and uses multi-modal large language models (MLLMs) to produce human-understandable explanations. We evaluate our proposed method on several real-world and synthetic datasets, and the results demonstrate superior performance in generating high-quality explanations for latent variables. The results highlight the effectiveness of incorporating inductive biases and uncertainty quantification, significantly enhancing model interpretability.
Authors: Shoujin Huang, Guanxiong Luo, Yunlin Zhao, Yilong Liu, Yuwan Wang, Kexin Yang, Jingzhe Liu, Hua Guo, Min Wang, Lingyan Zhang, Mengye Lyu
Abstract: Simultaneous multislice (SMS) imaging is a powerful technique for accelerating magnetic resonance imaging (MRI) acquisitions. However, SMS reconstruction remains challenging due to complex signal interactions between and within the excited slices. In this study, we introduce ROGER, a robust SMS MRI reconstruction method based on deep generative priors. Utilizing denoising diffusion probabilistic models (DDPM), ROGER begins with Gaussian noise and gradually recovers individual slices through reverse diffusion iterations while enforcing data consistency from measured k-space data within the readout concatenation framework. The posterior sampling procedure is designed such that the DDPM training can be performed on single-slice images without requiring modifications for SMS tasks. Additionally, our method incorporates a low-frequency enhancement (LFE) module to address the practical issue that SMS-accelerated fast spin echo (FSE) and echo planar imaging (EPI) sequences cannot easily embed fully-sampled autocalibration signals. Extensive experiments on both retrospectively and prospectively accelerated datasets demonstrate that ROGER consistently outperforms existing methods, enhancing both anatomical and functional imaging with strong out-of-distribution generalization. The source code and sample data for ROGER are available at https://github.com/Solor-pikachu/ROGER.
Authors: Haojun Shi, Suyu Ye, Xinyu Fang, Chuanyang Jin, Leyla Isik, Yen-Ling Kuo, Tianmin Shu
Abstract: Understanding people's social interactions in complex real-world scenarios often relies on intricate mental reasoning. To truly understand how and why people interact with one another, we must infer the underlying mental states that give rise to the social interactions, i.e., Theory of Mind reasoning in multi-agent interactions. Additionally, social interactions are often multi-modal -- we can watch people's actions, hear their conversations, and/or read about their past behaviors. For AI systems to successfully and safely interact with people in real-world environments, they also need to understand people's mental states as well as their inferences about each other's mental states based on multi-modal information about their interactions. For this, we introduce MuMA-ToM, a Multi-modal Multi-Agent Theory of Mind benchmark. MuMA-ToM is the first multi-modal Theory of Mind benchmark that evaluates mental reasoning in embodied multi-agent interactions. In MuMA-ToM, we provide video and text descriptions of people's multi-modal behavior in realistic household environments. Based on the context, we then ask questions about people's goals, beliefs, and beliefs about others' goals. We validated MuMA-ToM in a human experiment and provided a human baseline. We also proposed a novel multi-modal, multi-agent ToM model, LIMP (Language model-based Inverse Multi-agent Planning). Our experimental results show that LIMP significantly outperforms state-of-the-art methods, including large multi-modal models (e.g., GPT-4o, Gemini-1.5 Pro) and a recent multi-modal ToM model, BIP-ALM.
Authors: Arjun Roy, Kaushik Roy
Abstract: The convergence of fully homomorphic encryption (FHE) and machine learning offers unprecedented opportunities for private inference of sensitive data. FHE enables computation directly on encrypted data, safeguarding the entire machine learning pipeline, including data and model confidentiality. However, existing FHE-based implementations for deep neural networks face significant challenges in computational cost, latency, and scalability, limiting their practical deployment. This paper introduces DCT-CryptoNets, a novel approach that operates directly in the frequency-domain to reduce the burden of computationally expensive non-linear activations and homomorphic bootstrap operations during private inference. It does so by utilizing the discrete cosine transform (DCT), commonly employed in JPEG encoding, which has inherent compatibility with remote computing services where images are generally stored and transmitted in this encoded format. DCT-CryptoNets demonstrates a substantial latency reductions of up to 5.3$\times$ compared to prior work on benchmark image classification tasks. Notably, it demonstrates inference on the ImageNet dataset within 2.5 hours (down from 12.5 hours on equivalent 96-thread compute resources). Furthermore, by learning perceptually salient low-frequency information DCT-CryptoNets improves the reliability of encrypted predictions compared to RGB-based networks by reducing error accumulating homomorphic bootstrap operations. DCT-CryptoNets also demonstrates superior scalability to RGB-based networks by further reducing computational cost as image size increases. This study demonstrates a promising avenue for achieving efficient and practical private inference of deep learning models on high resolution images seen in real-world applications.
Authors: Sangyun Lee, Yilun Xu, Tomas Geffner, Giulia Fanti, Karsten Kreis, Arash Vahdat, Weili Nie
Abstract: Consistency models have recently been introduced to accelerate sampling from diffusion models by directly predicting the solution (i.e., data) of the probability flow ODE (PF ODE) from initial noise. However, the training of consistency models requires learning to map all intermediate points along PF ODE trajectories to their corresponding endpoints. This task is much more challenging than the ultimate objective of one-step generation, which only concerns the PF ODE's noise-to-data mapping. We empirically find that this training paradigm limits the one-step generation performance of consistency models. To address this issue, we generalize consistency training to the truncated time range, which allows the model to ignore denoising tasks at earlier time steps and focus its capacity on generation. We propose a new parameterization of the consistency function and a two-stage training procedure that prevents the truncated-time training from collapsing to a trivial solution. Experiments on CIFAR-10 and ImageNet $64\times64$ datasets show that our method achieves better one-step and two-step FIDs than the state-of-the-art consistency models such as iCT-deep, using more than 2$\times$ smaller networks. Project page: https://truncated-cm.github.io/
Authors: Indranil Halder
Abstract: We develop an analytically tractable single-step diffusion model based on a linear denoiser and present explicit formula for the Kullback-Leibler divergence between generated and sampling distribution, taken to be isotropic Gaussian, showing the effect of finite diffusion time and noise scale. Our study further reveals that the monotonic fall phase of Kullback-Leibler divergence begins when the training dataset size reaches the dimension of the data points. Along the way, we provide a mathematically precise definition of memorization to non-memorization transition when only finite number of data points are available. It is shown that the simplified model also features this transition during the monotonic fall phase of the aforementioned Kullback-Leibler divergence. For large-scale practical diffusion models, we explain why higher number of diffusion steps enhance production quality based on the theoretical arguments presented before. In addition, we show that higher diffusion steps does not necessarily help in reducing memorization. These two facts combined suggests existence of an optimal number of diffusion steps for finite number of training samples.
Authors: Run Luo, Ting-En Lin, Haonan Zhang, Yuchuan Wu, Xiong Liu, Min Yang, Yongbin Li, Longze Chen, Jiaming Li, Lei Zhang, Yangyi Chen, Hamid Alinejad-Rokny, Fei Huang
Abstract: Recent advancements in omnimodal learning have been achieved in understanding and generation across images, text, and speech, though mainly within proprietary models. Limited omnimodal datasets and the inherent challenges associated with real-time emotional speech generation have hindered open-source progress. To address these issues, we propose openomni, a two-stage training method combining omnimodal alignment and speech generation to develop a state-of-the-art omnimodal large language model. In the alignment phase, a pre-trained speech model is further trained on text-image tasks to generalize from vision to speech in a (near) zero-shot manner, outperforming models trained on tri-modal datasets. In the speech generation phase, a lightweight decoder facilitates real-time emotional speech through training on speech tasks and preference learning. Experiments demonstrate that openomni consistently improves across omnimodal, vision-language, and speech-language evaluations, enabling natural, emotion-rich dialogues and real-time emotional speech generation.
Authors: Ping Guo, Cheng Gong, Xi Lin, Fei Liu, Zhichao Lu, Qingfu Zhang, Zhenkun Wang
Abstract: Crafting adversarial examples is crucial for evaluating and enhancing the robustness of Deep Neural Networks (DNNs), presenting a challenge equivalent to maximizing a non-differentiable 0-1 loss function. However, existing single objective methods, namely adversarial attacks focus on a surrogate loss function, do not fully harness the benefits of engaging multiple loss functions, as a result of insufficient understanding of their synergistic and conflicting nature. To overcome these limitations, we propose the Multi-Objective Set-based Attack (MOS Attack), a novel adversarial attack framework leveraging multiple loss functions and automatically uncovering their interrelations. The MOS Attack adopts a set-based multi-objective optimization strategy, enabling the incorporation of numerous loss functions without additional parameters. It also automatically mines synergistic patterns among various losses, facilitating the generation of potent adversarial attacks with fewer objectives. Extensive experiments have shown that our MOS Attack outperforms single-objective attacks. Furthermore, by harnessing the identified synergistic patterns, MOS Attack continues to show superior results with a reduced number of loss functions.
Authors: Shao-Hao Lu, Ren Wang, Ching-Chun Huang, Wei-Chen Chiu
Abstract: Recently, diffusion-based blind super-resolution (SR) methods have shown great ability to generate high-resolution images with abundant high-frequency detail, but the detail is often achieved at the expense of fidelity. Meanwhile, another line of research focusing on rectifying the reverse process of diffusion models (i.e., diffusion guidance), has demonstrated the power to generate high-fidelity results for non-blind SR. However, these methods rely on known degradation kernels, making them difficult to apply to blind SR. To address these issues, we present DADiff in this paper. DADiff incorporates degradation-aware models into the diffusion guidance framework, eliminating the need to know degradation kernels. Additionally, we propose two novel techniques: input perturbation and guidance scalar, to further improve our performance. Extensive experimental results show that our proposed method has superior performance over state-of-the-art methods on blind SR benchmarks.
Authors: Guido Nannini, Julian Suk, Patryk Rygiel, Simone Saitta, Luca Mariani, Riccardo Maragna, Andrea Baggiano, Gianluca Pontone, Jelmer M. Wolterink, Alberto Redaelli
Abstract: Coronary artery disease, caused by the narrowing of coronary vessels due to atherosclerosis, is the leading cause of death worldwide. The diagnostic gold standard, fractional flow reserve (FFR), measures the trans-stenotic pressure ratio during maximal vasodilation but is invasive and costly. This has driven the development of virtual FFR (vFFR) using computational fluid dynamics (CFD) to simulate coronary flow. Geometric deep learning algorithms have shown promise for learning features on meshes, including cardiovascular research applications. This study empirically analyzes various backends for predicting vFFR fields in coronary arteries as CFD surrogates, comparing six backends for learning hemodynamics on meshes using CFD solutions as ground truth. The study has two parts: i) Using 1,500 synthetic left coronary artery bifurcations, models were trained to predict pressure-related fields for vFFR reconstruction, comparing different learning variables. ii) Using 427 patient-specific CFD simulations, experiments were repeated focusing on the best-performing learning variable from the synthetic dataset. Most backends performed well on the synthetic dataset, especially when predicting pressure drop over the manifold. Transformer-based backends outperformed others when predicting pressure and vFFR fields and were the only models achieving strong performance on patient-specific data, excelling in both average per-point error and vFFR accuracy in stenotic lesions. These results suggest geometric deep learning backends can effectively replace CFD for simple geometries, while transformer-based networks are superior for complex, heterogeneous datasets. Pressure drop was identified as the optimal network output for learning pressure-related fields.
Authors: Carlos Augusto Pinheiro de Sousa, Heiko Hamann, Oliver Deussen
Abstract: SLAM is a foundational technique with broad applications in robotics and AR/VR. SLAM simulations evaluate new concepts, but testing on resource-constrained devices, such as VR HMDs, faces challenges: high computational cost and restricted sensor data access. This work proposes a sparse framework using mesh geometry projections as features, which improves efficiency and circumvents direct sensor data access, advancing SLAM research as we demonstrate in VR and through numerical evaluation.
Authors: Yuecheng Liu, Dafeng Chi, Shiguang Wu, Zhanguang Zhang, Yaochen Hu, Lingfeng Zhang, Yingxue Zhang, Shuang Wu, Tongtong Cao, Guowei Huang, Helong Huang, Guangjian Tian, Weichao Qiu, Xingyue Quan, Jianye Hao, Yuzheng Zhuang
Abstract: Spatial reasoning is an essential problem in embodied AI research. Efforts to enhance spatial reasoning abilities through supplementary spatial data and fine-tuning have proven limited and ineffective when addressing complex embodied tasks, largely due to their dependence on language-based outputs. While some approaches have introduced a point-based action space to mitigate this issue, they fall short in managing more intricate tasks within complex environments. This deficiency arises from their failure to fully exploit the inherent thinking and reasoning capabilities that are fundamental strengths of Vision-Language Models (VLMs). To address these limitations, we propose a novel approach named SpatialCoT, specifically designed to bolster the spatial reasoning capabilities of VLMs. Our approach comprises two stages: spatial coordinate bi-directional alignment, which aligns vision-language inputs with spatial coordinates, and chain-of-thought spatial grounding, which harnesses the reasoning capabilities of language models for advanced spatial reasoning. We evaluate SpatialCoT on challenging navigation and manipulation tasks, both in simulation and real-world settings. Experimental results demonstrate that our method significantly outperforms previous state-of-the-art approaches in both tasks.