Authors: Weijie Kong (Refer to the report for detailed contributions), Qi Tian (Refer to the report for detailed contributions), Zijian Zhang (Refer to the report for detailed contributions), Rox Min (Refer to the report for detailed contributions), Zuozhuo Dai (Refer to the report for detailed contributions), Jin Zhou (Refer to the report for detailed contributions), Jiangfeng Xiong (Refer to the report for detailed contributions), Xin Li (Refer to the report for detailed contributions), Bo Wu (Refer to the report for detailed contributions), Jianwei Zhang (Refer to the report for detailed contributions), Kathrina Wu (Refer to the report for detailed contributions), Qin Lin (Refer to the report for detailed contributions), Aladdin Wang (Refer to the report for detailed contributions), Andong Wang (Refer to the report for detailed contributions), Bai Jiawang (Refer to the report for detailed contributions), Changlin Li (Refer to the report for detailed contributions), Duojun Huang (Refer to the report for detailed contributions), Fang Yang (Refer to the report for detailed contributions), Hao Tan (Refer to the report for detailed contributions), Hongmei Wang (Refer to the report for detailed contributions), Jacob Song (Refer to the report for detailed contributions), Jiawang Bai (Refer to the report for detailed contributions), Jianbing Wu (Refer to the report for detailed contributions), Jinbao Xue (Refer to the report for detailed contributions), Joey Wang (Refer to the report for detailed contributions), Junkun Yuan (Refer to the report for detailed contributions), Kai Wang (Refer to the report for detailed contributions), Mengyang Liu (Refer to the report for detailed contributions), Pengyu Li (Refer to the report for detailed contributions), Shuai Li (Refer to the report for detailed contributions), Weiyan Wang (Refer to the report for detailed contributions), Wenqing Yu (Refer to the report for detailed contributions), Xinchi Deng (Refer to the report for detailed contributions), Yanxin Long (Refer to the report for detailed contributions), Yi Chen (Refer to the report for detailed contributions), Yutao Cui (Refer to the report for detailed contributions), Yuanbo Peng (Refer to the report for detailed contributions), Zhentao Yu (Refer to the report for detailed contributions), Zhiyu He (Refer to the report for detailed contributions), Zhiyong Xu (Refer to the report for detailed contributions), Zixiang Zhou (Refer to the report for detailed contributions), Zunnan Xu (Refer to the report for detailed contributions), Yangyu Tao (Refer to the report for detailed contributions), Qinglin Lu (Refer to the report for detailed contributions), Songtao Liu (Refer to the report for detailed contributions), Daquan Zhou (Refer to the report for detailed contributions), Hongfa Wang (Refer to the report for detailed contributions), Yong Yang (Refer to the report for detailed contributions), Di Wang (Refer to the report for detailed contributions), Yuhong Liu (Refer to the report for detailed contributions), Jie Jiang (Refer to the report for detailed contributions), Caesar Zhong (Refer to the report for detailed contributions)
Abstract: Recent advancements in video generation have significantly impacted daily life for both individuals and industries. However, the leading video generation models remain closed-source, resulting in a notable performance gap between industry capabilities and those available to the public. In this report, we introduce HunyuanVideo, an innovative open-source video foundation model that demonstrates performance in video generation comparable to, or even surpassing, that of leading closed-source models. HunyuanVideo encompasses a comprehensive framework that integrates several key elements, including data curation, advanced architectural design, progressive model scaling and training, and an efficient infrastructure tailored for large-scale model training and inference. As a result, we successfully trained a video generative model with over 13 billion parameters, making it the largest among all open-source models. We conducted extensive experiments and implemented a series of targeted designs to ensure high visual quality, motion dynamics, text-video alignment, and advanced filming techniques. According to evaluations by professionals, HunyuanVideo outperforms previous state-of-the-art models, including Runway Gen-3, Luma 1.6, and three top-performing Chinese video generative models. By releasing the code for the foundation model and its applications, we aim to bridge the gap between closed-source and open-source communities. This initiative will empower individuals within the community to experiment with their ideas, fostering a more dynamic and vibrant video generation ecosystem. The code is publicly available at https://github.com/Tencent/HunyuanVideo.
Authors: Jon Guti\'errez-Zaballa, Koldo Basterretxea, Javier Echanobe
Abstract: As the deployment of artifical intelligence (AI) algorithms at edge devices becomes increasingly prevalent, enhancing the robustness and reliability of autonomous AI-based perception and decision systems is becoming as relevant as precision and performance, especially in applications areas considered safety-critical such as autonomous driving and aerospace. This paper delves into the robustness assessment in embedded Deep Neural Networks (DNNs), particularly focusing on the impact of parameter perturbations produced by single event upsets (SEUs) on convolutional neural networks (CNN) for image semantic segmentation. By scrutinizing the layer-by-layer and bit-by-bit sensitivity of various encoder-decoder models to soft errors, this study thoroughly investigates the vulnerability of segmentation DNNs to SEUs and evaluates the consequences of techniques like model pruning and parameter quantization on the robustness of compressed models aimed at embedded implementations. The findings offer valuable insights into the mechanisms underlying SEU-induced failures that allow for evaluating the robustness of DNNs once trained in advance. Moreover, based on the collected data, we propose a set of practical lightweight error mitigation techniques with no memory or computational cost suitable for resource-constrained deployments. The code used to perform the fault injection (FI) campaign is available at https://github.com/jonGuti13/TensorFI2 , while the code to implement proposed techniques is available at https://github.com/jonGuti13/parameterProtection .
URLs: https://github.com/jonGuti13/TensorFI2, https://github.com/jonGuti13/parameterProtection
Authors: Zehuan Huang, Yuan-Chen Guo, Haoran Wang, Ran Yi, Lizhuang Ma, Yan-Pei Cao, Lu Sheng
Abstract: Existing multi-view image generation methods often make invasive modifications to pre-trained text-to-image (T2I) models and require full fine-tuning, leading to (1) high computational costs, especially with large base models and high-resolution images, and (2) degradation in image quality due to optimization difficulties and scarce high-quality 3D data. In this paper, we propose the first adapter-based solution for multi-view image generation, and introduce MV-Adapter, a versatile plug-and-play adapter that enhances T2I models and their derivatives without altering the original network structure or feature space. By updating fewer parameters, MV-Adapter enables efficient training and preserves the prior knowledge embedded in pre-trained models, mitigating overfitting risks. To efficiently model the 3D geometric knowledge within the adapter, we introduce innovative designs that include duplicated self-attention layers and parallel attention architecture, enabling the adapter to inherit the powerful priors of the pre-trained models to model the novel 3D knowledge. Moreover, we present a unified condition encoder that seamlessly integrates camera parameters and geometric information, facilitating applications such as text- and image-based 3D generation and texturing. MV-Adapter achieves multi-view generation at 768 resolution on Stable Diffusion XL (SDXL), and demonstrates adaptability and versatility. It can also be extended to arbitrary view generation, enabling broader applications. We demonstrate that MV-Adapter sets a new quality standard for multi-view image generation, and opens up new possibilities due to its efficiency, adaptability and versatility.
Authors: Davide Bucciarelli, Nicholas Moratelli, Marcella Cornia, Lorenzo Baraldi, Rita Cucchiara
Abstract: The task of image captioning demands an algorithm to generate natural language descriptions of visual inputs. Recent advancements have seen a convergence between image captioning research and the development of Large Language Models (LLMs) and Multimodal LLMs -- like GPT-4V and Gemini -- which extend the capabilities of text-only LLMs to multiple modalities. This paper investigates whether Multimodal LLMs can supplant traditional image captioning networks by evaluating their performance on various image description benchmarks. We explore both the zero-shot capabilities of these models and their adaptability to different semantic domains through fine-tuning methods, including prompt learning, prefix tuning, and low-rank adaptation. Our results demonstrate that while Multimodal LLMs achieve impressive zero-shot performance, fine-tuning for specific domains while maintaining their generalization capabilities intact remains challenging. We discuss the implications of these findings for future research in image captioning and the development of more adaptable Multimodal LLMs.
Authors: Wang Xiyao, Yang Zhengyuan, Li Linjie, Lu Hongjin, Xu Yuancheng, Lin Chung-Ching Lin, Lin Kevin, Huang Furong, Wang Lijuan
Abstract: Despite significant advancements in vision-language models (VLMs), there lacks effective approaches to enhance response quality by scaling inference-time computation. This capability is known to be a core step towards the self-improving models in recent large language model studies. In this paper, we present Vision Value Model (VisVM) that can guide VLM inference-time search to generate responses with better visual comprehension. Specifically, VisVM not only evaluates the generated sentence quality in the current search step, but also anticipates the quality of subsequent sentences that may result from the current step, thus providing a long-term value. In this way, VisVM steers VLMs away from generating sentences prone to hallucinations or insufficient detail, thereby producing higher quality responses. Experimental results demonstrate that VisVM-guided search significantly enhances VLMs' ability to generate descriptive captions with richer visual details and fewer hallucinations, compared with greedy decoding and search methods with other visual reward signals. Furthermore, we find that self-training the model with the VisVM-guided captions improve VLM's performance across a wide range of multimodal benchmarks, indicating the potential for developing self-improving VLMs. Our value model and code are available at https://github.com/si0wang/VisVM.
Authors: Chaoyu Li, Eun Woo Im, Pooyan Fazli
Abstract: Multimodal large language models (MLLMs) have recently shown significant advancements in video understanding, excelling in content reasoning and instruction-following tasks. However, the problem of hallucination, where models generate inaccurate or misleading content, remains underexplored in the video domain. Building on the observation that the visual encoder of MLLMs often struggles to differentiate between video pairs that are visually distinct but semantically similar, we introduce VidHalluc, the largest benchmark designed to examine hallucinations in MLLMs for video understanding tasks. VidHalluc assesses hallucinations across three critical dimensions: (1) action, (2) temporal sequence, and (3) scene transition. VidHalluc consists of 5,002 videos, paired based on semantic similarity and visual differences, focusing on cases where hallucinations are most likely to occur. Through comprehensive testing, our experiments show that most MLLMs are vulnerable to hallucinations across these dimensions. Furthermore, we propose DINO-HEAL, a training-free method that reduces hallucinations by incorporating spatial saliency information from DINOv2 to reweight visual features during inference. Our results demonstrate that DINO-HEAL consistently improves performance on VidHalluc, achieving an average improvement of 3.02% in mitigating hallucinations among all tasks. Both the VidHalluc benchmark and DINO-HEAL code can be accessed via $\href{https://vid-halluc.github.io/}{\text{this link}}$.
Authors: Eun Woo Im, Junsung Shin, Sungyong Baik, Tae Hyun Kim
Abstract: Relying on the representation power of neural networks, most recent works have often neglected several factors involved in haze degradation, such as transmission (the amount of light reaching an observer from a scene over distance) and atmospheric light. These factors are generally unknown, making dehazing problems ill-posed and creating inherent uncertainties. To account for such uncertainties and factors involved in haze degradation, we introduce a variational Bayesian framework for single image dehazing. We propose to take not only a clean image and but also transmission map as latent variables, the posterior distributions of which are parameterized by corresponding neural networks: dehazing and transmission networks, respectively. Based on a physical model for haze degradation, our variational Bayesian framework leads to a new objective function that encourages the cooperation between them, facilitating the joint training of and thereby boosting the performance of each other. In our framework, a dehazing network can estimate a clean image independently of a transmission map estimation during inference, introducing no overhead. Furthermore, our model-agnostic framework can be seamlessly incorporated with other existing dehazing networks, greatly enhancing the performance consistently across datasets and models.
Authors: Yuxuan Jiang, Ho Man Kwan, Tianhao Peng, Ge Gao, Fan Zhang, Xiaoqing Zhu, Joel Sole, David Bull
Abstract: Recent advances in implicit neural representations (INRs) have shown significant promise in modeling visual signals for various low-vision tasks including image super-resolution (ISR). INR-based ISR methods typically learn continuous representations, providing flexibility for generating high-resolution images at any desired scale from their low-resolution counterparts. However, existing INR-based ISR methods utilize multi-layer perceptrons for parameterization in the network; this does not take account of the hierarchical structure existing in local sampling points and hence constrains the representation capability. In this paper, we propose a new \textbf{H}ierarchical encoding based \textbf{I}mplicit \textbf{I}mage \textbf{F}unction for continuous image super-resolution, \textbf{HIIF}, which leverages a novel hierarchical positional encoding that enhances the local implicit representation, enabling it to capture fine details at multiple scales. Our approach also embeds a multi-head linear attention mechanism within the implicit attention network by taking additional non-local information into account. Our experiments show that, when integrated with different backbone encoders, HIIF outperforms the state-of-the-art continuous image super-resolution methods by up to 0.17dB in PSNR. The source code of HIIF will be made publicly available at \url{www.github.com}.
Authors: Justin Theiss, Norman M\"uller, Daeil Kim, Aayush Prakash
Abstract: Recently, text-to-image generation with diffusion models has made significant advancements in both higher fidelity and generalization capabilities compared to previous baselines. However, generating holistic multi-view consistent images from prompts still remains an important and challenging task. To address this challenge, we propose a diffusion process that attends to time-dependent spatial frequencies of features with a novel attention mechanism as well as novel noise initialization technique and cross-attention loss. This Fourier-based attention block focuses on features from non-overlapping regions of the generated scene in order to better align the global appearance. Our noise initialization technique incorporates shared noise and low spatial frequency information derived from pixel coordinates and depth maps to induce noise correlations across views. The cross-attention loss further aligns features sharing the same prompt across the scene. Our technique improves SOTA on several quantitative metrics with qualitatively better results when compared to other state-of-the-art approaches for multi-view consistency.
Authors: Ruibo Ming, Jingwei Wu, Zhewei Huang, Zhuoxuan Ju, Jianming HU, Lihui Peng, Shuchang Zhou
Abstract: Recent advances in auto-regressive large language models (LLMs) have shown their potential in generating high-quality text, inspiring researchers to apply them to image and video generation. This paper explores the application of LLMs to video continuation, a task essential for building world models and predicting future frames. In this paper, we tackle challenges including preventing degeneration in long-term frame generation and enhancing the quality of generated images. We design a scheme named ARCON, which involves training our model to alternately generate semantic tokens and RGB tokens, enabling the LLM to explicitly learn and predict the high-level structural information of the video. We find high consistency in the RGB images and semantic maps generated without special design. Moreover, we employ an optical flow-based texture stitching method to enhance the visual quality of the generated videos. Quantitative and qualitative experiments in autonomous driving scenarios demonstrate our model can consistently generate long videos.
Authors: Quang Nguyen, Truong Vu, Trong-Tung Nguyen, Yuxin Wen, Preston K Robinette, Taylor T Johnson, Tom Goldstein, Anh Tran, Khoi Nguyen
Abstract: Image editing technologies are tools used to transform, adjust, remove, or otherwise alter images. Recent research has significantly improved the capabilities of image editing tools, enabling the creation of photorealistic and semantically informed forged regions that are nearly indistinguishable from authentic imagery, presenting new challenges in digital forensics and media credibility. While current image forensic techniques are adept at localizing forged regions produced by traditional image manipulation methods, current capabilities struggle to localize regions created by diffusion-based techniques. To bridge this gap, we present a novel framework that integrates a multimodal Large Language Model (LLM) for enhanced reasoning capabilities to localize tampered regions in images produced by diffusion model-based editing methods. By leveraging the contextual and semantic strengths of LLMs, our framework achieves promising results on MagicBrush, AutoSplice, and PerfBrush (novel diffusion-based dataset) datasets, outperforming previous approaches in mIoU and F1-score metrics. Notably, our method excels on the PerfBrush dataset, a self-constructed test set featuring previously unseen types of edits. Here, where traditional methods typically falter, achieving markedly low scores, our approach demonstrates promising performance.
Authors: Sanjoeng Wong, Yan Yan
Abstract: Automatic detection of prohibited items in X-ray images plays a crucial role in public security. However, existing methods rely heavily on labor-intensive box annotations. To address this, we investigate X-ray prohibited item detection under labor-efficient point supervision and develop an intra-inter objectness learning network (I$^2$OL-Net). I$^2$OL-Net consists of two key modules: an intra-modality objectness learning (intra-OL) module and an inter-modality objectness learning (inter-OL) module. The intra-OL module designs a local focus Gaussian masking block and a global random Gaussian masking block to collaboratively learn the objectness in X-ray images. Meanwhile, the inter-OL module introduces the wavelet decomposition-based adversarial learning block and the objectness block, effectively reducing the modality discrepancy and transferring the objectness knowledge learned from natural images with box annotations to X-ray images. Based on the above, I$^2$OL-Net greatly alleviates the problem of part domination caused by severe intra-class variations in X-ray images. Experimental results on four X-ray datasets show that I$^2$OL-Net can achieve superior performance with a significant reduction of annotation cost, thus enhancing its accessibility and practicality.
Authors: Guangben Lu, Yuzhen Du, Zhimin Sun, Ran Yi, Yifan Qi, Yizhe Tang, Tianyi Wang, Lizhuang Ma, Fangyuan Zou
Abstract: Foreground-conditioned inpainting aims to seamlessly fill the background region of an image by utilizing the provided foreground subject and a text description. While existing T2I-based image inpainting methods can be applied to this task, they suffer from issues of subject shape expansion, distortion, or impaired ability to align with the text description, resulting in inconsistencies between the visual elements and the text description. To address these challenges, we propose Pinco, a plug-and-play foreground-conditioned inpainting adapter that generates high-quality backgrounds with good text alignment while effectively preserving the shape of the foreground subject. Firstly, we design a Self-Consistent Adapter that integrates the foreground subject features into the layout-related self-attention layer, which helps to alleviate conflicts between the text and subject features by ensuring that the model can effectively consider the foreground subject's characteristics while processing the overall image layout. Secondly, we design a Decoupled Image Feature Extraction method that employs distinct architectures to extract semantic and shape features separately, significantly improving subject feature extraction and ensuring high-quality preservation of the subject's shape. Thirdly, to ensure precise utilization of the extracted features and to focus attention on the subject region, we introduce a Shared Positional Embedding Anchor, greatly improving the model's understanding of subject features and boosting training efficiency. Extensive experiments demonstrate that our method achieves superior performance and efficiency in foreground-conditioned inpainting.
Authors: Yuzhen Du, Teng Hu, Jiangning Zhang, Ran Yi Chengming Xu, Xiaobin Hu, Kai Wu, Donghao Luo, Yabiao Wang, Lizhuang Ma
Abstract: Image Restoration aims to restore degraded images, with deep learning, especially CNNs and Transformers, enhancing performance. However, there's a lack of a unified training benchmark for IR. We identified a bias in image complexity between training and testing datasets, affecting restoration quality. To address this, we created ReSyn, a large-scale IR dataset with balanced complexity, including real and synthetic images. We also established a unified training standard for IR models. Our RWKV-IR model integrates linear complexity RWKV into transformers for global and local receptive fields. It replaces Q-Shift with Depth-wise Convolution for local dependencies and combines Bi-directional attention for global-local awareness. The Cross-Bi-WKV module balances horizontal and vertical attention. Experiments show RWKV-IR's effectiveness in image restoration.
Authors: Zuo Zuo, Jiahao Dong, Yao Wu, Yanyun Qu, Zongze Wu
Abstract: Industrial anomaly classification (AC) is an indispensable task in industrial manufacturing, which guarantees quality and safety of various product. To address the scarcity of data in industrial scenarios, lots of few-shot anomaly detection methods emerge recently. In this paper, we propose an effective few-shot anomaly classification (FSAC) framework with one-stage training, dubbed CLIP-FSAC++. Specifically, we introduce a cross-modality interaction module named Anomaly Descriptor following image and text encoders, which enhances the correlation of visual and text embeddings and adapts the representations of CLIP from pre-trained data to target data. In anomaly descriptor, image-to-text cross-attention module is used to obtain image-specific text embeddings and text-to-image cross-attention module is used to obtain text-specific visual embeddings. Then these modality-specific embeddings are used to enhance original representations of CLIP for better matching ability. Comprehensive experiment results are provided for evaluating our method in few-normal shot anomaly classification on VisA and MVTEC-AD for 1, 2, 4 and 8-shot settings. The source codes are at https://github.com/Jay-zzcoder/clip-fsac-pp
Authors: Yuan Xue, Qi Zhang, Chuanmin Jia, Shiqi Wang
Abstract: Image Compression for Machines (ICM) aims to compress images for machine vision tasks rather than human viewing. Current works predominantly concentrate on high-level tasks like object detection and semantic segmentation. However, the quality of original images is usually not guaranteed in the real world, leading to even worse perceptual quality or downstream task performance after compression. Low-level (LL) machine vision models, like image restoration models, can help improve such quality, and thereby their compression requirements should also be considered. In this paper, we propose a pioneered ICM framework for LL machine vision tasks, namely LL-ICM. By jointly optimizing compression and LL tasks, the proposed LL-ICM not only enriches its encoding ability in generalizing to versatile LL tasks but also optimizes the processing ability of down-stream LL task models, achieving mutual adaptation for image codecs and LL task models. Furthermore, we integrate large-scale vision-language models into the LL-ICM framework to generate more universal and distortion-robust feature embeddings for LL vision tasks. Therefore, one LL-ICM codec can generalize to multiple tasks. We establish a solid benchmark to evaluate LL-ICM, which includes extensive objective experiments by using both full and no-reference image quality assessments. Experimental results show that LL-ICM can achieve 22.65% BD-rate reductions over the state-of-the-art methods.
Authors: Jingyu Lin, Jiaqi Gu, Lubin Fan, Bojian Wu, Yujing Lou, Renjie Chen, Ligang Liu, Jieping Ye
Abstract: Generating high-quality novel view renderings of 3D Gaussian Splatting (3DGS) in scenes featuring transient objects is challenging. We propose a novel hybrid representation, termed as HybridGS, using 2D Gaussians for transient objects per image and maintaining traditional 3D Gaussians for the whole static scenes. Note that, the 3DGS itself is better suited for modeling static scenes that assume multi-view consistency, but the transient objects appear occasionally and do not adhere to the assumption, thus we model them as planar objects from a single view, represented with 2D Gaussians. Our novel representation decomposes the scene from the perspective of fundamental viewpoint consistency, making it more reasonable. Additionally, we present a novel multi-view regulated supervision method for 3DGS that leverages information from co-visible regions, further enhancing the distinctions between the transients and statics. Then, we propose a straightforward yet effective multi-stage training strategy to ensure robust training and high-quality view synthesis across various settings. Experiments on benchmark datasets show our state-of-the-art performance of novel view synthesis in both indoor and outdoor scenes, even in the presence of distracting elements.
Authors: Jayaprakash Sundararaj, Akhil Vyas, Benjamin Gonzalez-Maldonado
Abstract: Converting mathematical expressions into LaTeX is challenging. In this paper, we explore using newer transformer based architectures for addressing the problem of converting handwritten/digital mathematical expression images into equivalent LaTeX code. We use the current state of the art CNN encoder and RNN decoder as a baseline for our experiments. We also investigate improvements to CNN-RNN architecture by replacing the CNN encoder with the ResNet50 model. Our experiments show that transformer architectures achieve a higher overall accuracy and BLEU scores along with lower Levenschtein scores compared to the baseline CNN/RNN architecture with room to achieve even better results with appropriate fine-tuning of model parameters.
Authors: Hui Zhang, Dexiang Hong, Tingwei Gao, Yitong Wang, Jie Shao, Xinglong Wu, Zuxuan Wu, Yu-Gang Jiang
Abstract: Diffusion models have been recognized for their ability to generate images that are not only visually appealing but also of high artistic quality. As a result, Layout-to-Image (L2I) generation has been proposed to leverage region-specific positions and descriptions to enable more precise and controllable generation. However, previous methods primarily focus on UNet-based models (e.g., SD1.5 and SDXL), and limited effort has explored Multimodal Diffusion Transformers (MM-DiTs), which have demonstrated powerful image generation capabilities. Enabling MM-DiT for layout-to-image generation seems straightforward but is challenging due to the complexity of how layout is introduced, integrated, and balanced among multiple modalities. To this end, we explore various network variants to efficiently incorporate layout guidance into MM-DiT, and ultimately present SiamLayout. To Inherit the advantages of MM-DiT, we use a separate set of network weights to process the layout, treating it as equally important as the image and text modalities. Meanwhile, to alleviate the competition among modalities, we decouple the image-layout interaction into a siamese branch alongside the image-text one and fuse them in the later stage. Moreover, we contribute a large-scale layout dataset, named LayoutSAM, which includes 2.7 million image-text pairs and 10.7 million entities. Each entity is annotated with a bounding box and a detailed description. We further construct the LayoutSAM-Eval benchmark as a comprehensive tool for evaluating the L2I generation quality. Finally, we introduce the Layout Designer, which taps into the potential of large language models in layout planning, transforming them into experts in layout generation and optimization. Our code, model, and dataset will be available at https://creatilayout.github.io.
Authors: Chu Myaet Thwal, Ye Lin Tun, Minh N. H. Nguyen, Eui-Nam Huh, Choong Seon Hong
Abstract: Beyond the success of Contrastive Language-Image Pre-training (CLIP), recent trends mark a shift toward exploring the applicability of lightweight vision-language models for resource-constrained scenarios. These models often deliver suboptimal performance when relying solely on a single image-text contrastive learning objective, spotlighting the need for more effective training mechanisms that guarantee robust cross-modal feature alignment. In this work, we propose CLIP-PING: Contrastive Language-Image Pre-training with Proximus Intrinsic Neighbors Guidance, a simple and efficient training paradigm designed to boost the performance of lightweight vision-language models with minimal computational overhead and lower data demands. CLIP-PING bootstraps unimodal features extracted from arbitrary pre-trained encoders to obtain intrinsic guidance of proximus neighbor samples, i.e., nearest-neighbor (NN) and cross nearest-neighbor (XNN). We find that extra contrastive supervision from these neighbors substantially boosts cross-modal alignment, enabling lightweight models to learn more generic features with rich semantic diversity. Extensive experiments reveal that CLIP-PING notably surpasses its peers in zero-shot generalization and cross-modal retrieval tasks. Specifically, a 5.5% gain on zero-shot ImageNet1K with 10.7% (I2T) and 5.7% (T2I) on Flickr30K, compared to the original CLIP when using ViT-XS image encoder trained on 3 million (image, text) pairs. Moreover, CLIP-PING showcases strong transferability under the linear evaluation protocol across several downstream tasks.
Authors: Jiangweizhi Peng, Zhiwei Tang, Gaowen Liu, Charles Fleming, Mingyi Hong
Abstract: Text-to-Image (T2I) diffusion models are widely recognized for their ability to generate high-quality and diverse images based on text prompts. However, despite recent advances, these models are still prone to generating unsafe images containing sensitive or inappropriate content, which can be harmful to users. Current efforts to prevent inappropriate image generation for diffusion models are easy to bypass and vulnerable to adversarial attacks. How to ensure that T2I models align with specific safety goals remains a significant challenge. In this work, we propose a novel, training-free approach, called Prompt-Noise Optimization (PNO), to mitigate unsafe image generation. Our method introduces a novel optimization framework that leverages both the continuous prompt embedding and the injected noise trajectory in the sampling process to generate safe images. Extensive numerical results demonstrate that our framework achieves state-of-the-art performance in suppressing toxic image generations and demonstrates robustness to adversarial attacks, without needing to tune the model parameters. Furthermore, compared with existing methods, PNO uses comparable generation time while offering the best tradeoff between the conflicting goals of safe generation and prompt-image alignment.
Authors: Sudha Krishnamurthy, Vimal Bhat, Abhinav Jain
Abstract: The proliferation of several streaming services in recent years has now made it possible for a diverse audience across the world to view the same media content, such as movies or TV shows. While translation and dubbing services are being added to make content accessible to the local audience, the support for making content accessible to people with different abilities, such as the Deaf and Hard of Hearing (DHH) community, is still lagging. Our goal is to make media content more accessible to the DHH community by generating sign language videos with synthetic signers that are realistic and expressive. Using the same signer for a given media content that is viewed globally may have limited appeal. Hence, our approach combines parametric modeling and generative modeling to generate realistic-looking synthetic signers and customize their appearance based on user preferences. We first retarget human sign language poses to 3D sign language avatars by optimizing a parametric model. The high-fidelity poses from the rendered avatars are then used to condition the poses of synthetic signers generated using a diffusion-based generative model. The appearance of the synthetic signer is controlled by an image prompt supplied through a visual adapter. Our results show that the sign language videos generated using our approach have better temporal consistency and realism than signing videos generated by a diffusion model conditioned only on text prompts. We also support multimodal prompts to allow users to further customize the appearance of the signer to accommodate diversity (e.g. skin tone, gender). Our approach is also useful for signer anonymization.
Authors: Michelle Guo, Mia Tang, Hannah Cha, Ruohan Zhang, C. Karen Liu, Jiajun Wu
Abstract: For designing a wide range of everyday objects, the design process should be aware of both the human body and the underlying semantics of the design specification. However, these two objectives present significant challenges to the current AI-based designing tools. In this work, we present a method to synthesize body-aware 3D objects from a base mesh given an input body geometry and either text or image as guidance. The generated objects can be simulated on virtual characters, or fabricated for real-world use. We propose to use a mesh deformation procedure that optimizes for both semantic alignment as well as contact and penetration losses. Using our method, users can generate both virtual or real-world objects from text, image, or sketch, without the need for manual artist intervention. We present both qualitative and quantitative results on various object categories, demonstrating the effectiveness of our approach.
Authors: Donghoon Ahn, Jiwon Kang, Sanghyun Lee, Jaewon Min, Minjae Kim, Wooseok Jang, Hyoungwon Cho, Sayak Paul, SeonHwa Kim, Eunju Cha, Kyong Hwan Jin, Seungryong Kim
Abstract: Diffusion models excel in generating high-quality images. However, current diffusion models struggle to produce reliable images without guidance methods, such as classifier-free guidance (CFG). Are guidance methods truly necessary? Observing that noise obtained via diffusion inversion can reconstruct high-quality images without guidance, we focus on the initial noise of the denoising pipeline. By mapping Gaussian noise to `guidance-free noise', we uncover that small low-magnitude low-frequency components significantly enhance the denoising process, removing the need for guidance and thus improving both inference throughput and memory. Expanding on this, we propose \ours, a novel method that replaces guidance methods with a single refinement of the initial noise. This refined noise enables high-quality image generation without guidance, within the same diffusion pipeline. Our noise-refining model leverages efficient noise-space learning, achieving rapid convergence and strong performance with just 50K text-image pairs. We validate its effectiveness across diverse metrics and analyze how refined noise can eliminate the need for guidance. See our project page: https://cvlab-kaist.github.io/NoiseRefine/.
Authors: Zhu Han, Ce Zhang, Lianru Gao, Zhiqiang Zeng, Michael K. Ng, Bing Zhang, Jocelyn Chanussot
Abstract: Cross-scene image classification aims to transfer prior knowledge of ground materials to annotate regions with different distributions and reduce hand-crafted cost in the field of remote sensing. However, existing approaches focus on single-source domain generalization to unseen target domains, and are easily confused by large real-world domain shifts due to the limited training information and insufficient diversity modeling capacity. To address this gap, we propose a novel multi-source collaborative domain generalization framework (MS-CDG) based on homogeneity and heterogeneity characteristics of multi-source remote sensing data, which considers data-aware adversarial augmentation and model-aware multi-level diversification simultaneously to enhance cross-scene generalization performance. The data-aware adversarial augmentation adopts an adversary neural network with semantic guide to generate MS samples by adaptively learning realistic channel and distribution changes across domains. In views of cross-domain and intra-domain modeling, the model-aware diversification transforms the shared spatial-channel features of MS data into the class-wise prototype and kernel mixture module, to address domain discrepancies and cluster different classes effectively. Finally, the joint classification of original and augmented MS samples is employed by introducing a distribution consistency alignment to increase model diversity and ensure better domain-invariant representation learning. Extensive experiments on three public MS remote sensing datasets demonstrate the superior performance of the proposed method when benchmarked with the state-of-the-art methods.
Authors: Shuang Li, Yibing Wang, Jian Gao, Chulhong Kim, Seongwook Choi, Yu Zhang, Qian Chen, Yao Yao, Changhui Li
Abstract: Large-scale dynamic three-dimensional (3D) photoacoustic imaging (PAI) is significantly important in clinical applications. In practical implementations, large-scale 3D real-time PAI systems typically utilize sparse two-dimensional (2D) sensor arrays with certain angular deficiencies, necessitating advanced iterative reconstruction (IR) algorithms to achieve quantitative PAI and reduce reconstruction artifacts. However, for existing IR algorithms, multi-frame 3D reconstruction leads to extremely high memory consumption and prolonged computation time, with limited consideration of the spatial-temporal continuity between data frames. Here, we propose a novel method, named the 4D sliding Gaussian ball adaptive growth (4D SlingBAG) algorithm, based on the current point cloud-based IR algorithm sliding Gaussian ball adaptive growth (SlingBAG), which has minimal memory consumption among IR methods. Our 4D SlingBAG method applies spatial-temporal coupled deformation functions to each Gaussian sphere in point cloud, thus explicitly learning the deformations features of the dynamic 3D PA scene. This allows for the efficient representation of various physiological processes (such as pulsation) or external pressures (e.g., blood perfusion experiments) contributing to changes in vessel morphology and blood flow during dynamic 3D PAI, enabling highly efficient IR for dynamic 3D PAI. Simulation experiments demonstrate that 4D SlingBAG achieves high-quality dynamic 3D PA reconstruction. Compared to performing reconstructions by using SlingBAG algorithm individually for each frame, our method significantly reduces computational time and keeps a extremely low memory consumption. The project for 4D SlingBAG can be found in the following GitHub repository: \href{https://github.com/JaegerCQ/4D-SlingBAG}{https://github.com/JaegerCQ/4D-SlingBAG}.
URLs: https://github.com/JaegerCQ/4D-SlingBAG, https://github.com/JaegerCQ/4D-SlingBAG
Authors: Yizhou Jin, Jiahui Zhu, Guodong Wang, Shiwei Li, Jinjin Zhang, Qingjie Liu, Xinyue Liu, Yunhong Wang
Abstract: Incremental anomaly detection sequentially recognizes abnormal regions in novel categories for dynamic industrial scenarios. This remains highly challenging due to knowledge overwriting and feature conflicts, leading to catastrophic forgetting. In this work, we propose ONER, an end-to-end ONline Experience Replay method, which efficiently mitigates catastrophic forgetting while adapting to new tasks with minimal cost. Specifically, our framework utilizes two types of experiences from past tasks: decomposed prompts and semantic prototypes, addressing both model parameter updates and feature optimization. The decomposed prompts consist of learnable components that assemble to produce attention-conditioned prompts. These prompts reuse previously learned knowledge, enabling model to learn novel tasks effectively. The semantic prototypes operate at both pixel and image levels, performing regularization in the latent feature space to prevent forgetting across various tasks. Extensive experiments demonstrate that our method achieves state-of-the-art performance in incremental anomaly detection with significantly reduced forgetting, as well as efficiently adapting to new categories with minimal costs. These results confirm the efficiency and stability of ONER, making it a powerful solution for real-world applications.
Authors: Zhizhen Chen, Subrat Kishore Dutta, Zhengyu Zhao, Chenhao Lin, Chao Shen, Xiao Zhang
Abstract: Targeted poisoning attacks aim to compromise the model's prediction on specific target samples. In a common clean-label setting, they are achieved by slightly perturbing a subset of training samples given access to those specific targets. Despite continuous efforts, it remains unexplored whether such attacks can generalize to unknown variations of those targets. In this paper, we take the first step to systematically study this generalization problem. Observing that the widely adopted, cosine similarity-based attack exhibits limited generalizability, we propose a well-generalizable attack that leverages both the direction and magnitude of model gradients. In particular, we explore diverse target variations, such as an object with varied viewpoints and an animal species with distinct appearances. Extensive experiments across various generalization scenarios demonstrate that our method consistently achieves the best attack effectiveness. For example, our method outperforms the cosine similarity-based attack by 20.95% in attack success rate with similar overall accuracy, averaged over four models on two image benchmark datasets. The code is available at https://github.com/jiaangk/generalizable_tcpa
Authors: Xuesong Li, Jinguang Tong, Jie Hong, Vivien Rolland, Lars Petersson
Abstract: Dynamic scene reconstruction from monocular video is critical for real-world applications. This paper tackles the dual challenges of dynamic novel-view synthesis and 3D geometry reconstruction by introducing a hybrid framework: Deformable Gaussian Splatting and Dynamic Neural Surfaces (DGNS), in which both modules can leverage each other for both tasks. During training, depth maps generated by the deformable Gaussian splatting module guide the ray sampling for faster processing and provide depth supervision within the dynamic neural surface module to improve geometry reconstruction. Simultaneously, the dynamic neural surface directs the distribution of Gaussian primitives around the surface, enhancing rendering quality. To further refine depth supervision, we introduce a depth-filtering process on depth maps derived from Gaussian rasterization. Extensive experiments on public datasets demonstrate that DGNS achieves state-of-the-art performance in both novel-view synthesis and 3D reconstruction.
Authors: Chamuditha Jayanga Galappaththige, Jason Lai, Lloyd Windrim, Donald Dansereau, Niko Suenderhauf, Dimity Miller
Abstract: Autonomous agents often require accurate methods for detecting and localizing changes in their environment, particularly when observations are captured from unconstrained and inconsistent viewpoints. We propose a novel label-free, pose-agnostic change detection method that integrates information from multiple viewpoints to construct a change-aware 3D Gaussian Splatting (3DGS) representation of the scene. With as few as 5 images of the post-change scene, our approach can learn additional change channels in a 3DGS and produce change masks that outperform single-view techniques. Our change-aware 3D scene representation additionally enables the generation of accurate change masks for unseen viewpoints. Experimental results demonstrate state-of-the-art performance in complex multi-object scenes, achieving a 1.7$\times$ and 1.6$\times$ improvement in Mean Intersection Over Union and F1 score respectively over other baselines. We also contribute a new real-world dataset to benchmark change detection in diverse challenging scenes in the presence of lighting variations.
Authors: Yanming Zhu, Xuefei Yin, Alan Wee-Chung Liew, Hui Tian
Abstract: With the rapid advancement of artificial intelligence and deep learning, medical image analysis has become a critical tool in modern healthcare, significantly improving diagnostic accuracy and efficiency. However, AI-based methods also raise serious privacy concerns, as medical images often contain highly sensitive patient information. This review offers a comprehensive overview of privacy-preserving techniques in medical image analysis, including encryption, differential privacy, homomorphic encryption, federated learning, and generative adversarial networks. We explore the application of these techniques across various medical image analysis tasks, such as diagnosis, pathology, and telemedicine. Notably, we organizes the review based on specific challenges and their corresponding solutions in different medical image analysis applications, so that technical applications are directly aligned with practical issues, addressing gaps in the current research landscape. Additionally, we discuss emerging trends, such as zero-knowledge proofs and secure multi-party computation, offering insights for future research. This review serves as a valuable resource for researchers and practitioners and can help advance privacy-preserving in medical image analysis.
Authors: Ming-Chang Chiu, Shicheng Wen, Pin-Yu Chen, Xuezhe Ma
Abstract: In vision-language models (VLMs), the ability to perceive and interpret color and physical environment is crucial for achieving contextually accurate understanding and interaction. However, despite advances in multimodal modeling, there remains a significant lack of specialized datasets that rigorously evaluate a model's capacity to discern subtle color variations and spatial context -- critical elements for situational comprehension and reliable deployment across real-world applications. Toward that goal, we curate MegaCOIN, a high-quality, human-labeled dataset based on \emph{real} images with various contextual attributes. MegaCOIN consists of two parts: MegaCOIN-Instruct, which serves as a supervised fine-tuning (SFT) dataset for VLMs; and MegaCOIN-Bench, an annotated test set that can be used as a stand-alone QA dataset. MegaCOIN~provides three annotated features for 220,000 real images: foreground color, background color, and description of an object's physical environment, constituting 660k human annotations. In addition, MegaCOIN can be applied to benchmark domain generalization (DG) algorithms. We explore benchmarking DG methods in the linear probing setup for VLM and show some new insights. Last but not least, we show that VLMs, including GPT-4o, have subpar color recognition capabilities, and fine-tuning with MegaCOIN can result in improved performance on visual evaluation tasks. In certain cases, MegaCOIN fine-tuned small-scale opensource models such as LLaVA and Bunny can outperform closed-source GPT-4o. We hope the utilities of MegaCOIN can shed light on the directions VLMs can improve and provide a more complex platform for domain generalization algorithms.
Authors: Mithun Parab, Pranay Lendave, Jiyoung Kim, Thi Quynh Dan Nguyen, Palash Ingle
Abstract: In image-assisted minimally invasive surgeries (MIS), understanding surgical scenes is vital for real-time feedback to surgeons, skill evaluation, and improving outcomes through collaborative human-robot procedures. Within this context, the challenge lies in accurately detecting, segmenting, and estimating the depth of surgical scenes depicted in high-resolution images, while simultaneously reconstructing the scene in 3D and providing segmentation of surgical instruments along with detection labels for each instrument. To address this challenge, a novel Multi-Task Learning (MTL) network is proposed for performing these tasks concurrently. A key aspect of this approach involves overcoming the optimization hurdles associated with handling multiple tasks concurrently by integrating a Adversarial Weight Update into the MTL framework, the proposed MTL model achieves 3D reconstruction through the integration of segmentation, depth estimation, and object detection, thereby enhancing the understanding of surgical scenes, which marks a significant advancement compared to existing studies that lack 3D capabilities. Comprehensive experiments on the EndoVis2018 benchmark dataset underscore the adeptness of the model in efficiently addressing all three tasks, demonstrating the efficacy of the proposed techniques.
Authors: Yifan Lu, Xuanchi Ren, Jiawei Yang, Tianchang Shen, Zhangjie Wu, Jun Gao, Yue Wang, Siheng Chen, Mike Chen, Sanja Fidler, Jiahui Huang
Abstract: We present InfiniCube, a scalable method for generating unbounded dynamic 3D driving scenes with high fidelity and controllability. Previous methods for scene generation either suffer from limited scales or lack geometric and appearance consistency along generated sequences. In contrast, we leverage the recent advancements in scalable 3D representation and video models to achieve large dynamic scene generation that allows flexible controls through HD maps, vehicle bounding boxes, and text descriptions. First, we construct a map-conditioned sparse-voxel-based 3D generative model to unleash its power for unbounded voxel world generation. Then, we re-purpose a video model and ground it on the voxel world through a set of carefully designed pixel-aligned guidance buffers, synthesizing a consistent appearance. Finally, we propose a fast feed-forward approach that employs both voxel and pixel branches to lift the dynamic videos to dynamic 3D Gaussians with controllable objects. Our method can generate controllable and realistic 3D driving scenes, and extensive experiments validate the effectiveness and superiority of our model.
Authors: Kiyohiro Nakayama, Jan Ackermann, Timur Levent Kesdogan, Yang Zheng, Maria Korosteleva, Olga Sorkine-Hornung, Leonidas J. Guibas, Guandao Yang, Gordon Wetzstein
Abstract: Apparel is essential to human life, offering protection, mirroring cultural identities, and showcasing personal style. Yet, the creation of garments remains a time-consuming process, largely due to the manual work involved in designing them. To simplify this process, we introduce AIpparel, a large multimodal model for generating and editing sewing patterns. Our model fine-tunes state-of-the-art large multimodal models (LMMs) on a custom-curated large-scale dataset of over 120,000 unique garments, each with multimodal annotations including text, images, and sewing patterns. Additionally, we propose a novel tokenization scheme that concisely encodes these complex sewing patterns so that LLMs can learn to predict them efficiently. \methodname achieves state-of-the-art performance in single-modal tasks, including text-to-garment and image-to-garment prediction, and enables novel multimodal garment generation applications such as interactive garment editing. The project website is at georgenakayama.github.io/AIpparel/.
Authors: Tianyu Chen, Zhendong Wang, Mingyuan Zhou
Abstract: Diffusion models have recently demonstrated notable success in solving inverse problems. However, current diffusion model-based solutions typically require a large number of function evaluations (NFEs) to generate high-quality images conditioned on measurements, as they incorporate only limited information at each step. To accelerate the diffusion-based inverse problem-solving process, we introduce \textbf{M}easurements \textbf{O}ptimization (MO), a more efficient plug-and-play module for integrating measurement information at each step of the inverse problem-solving process. This method is comprehensively evaluated across eight diverse linear and nonlinear tasks on the FFHQ and ImageNet datasets. By using MO, we establish state-of-the-art (SOTA) performance across multiple tasks, with key advantages: (1) it operates with no more than 100 NFEs, with phase retrieval on ImageNet being the sole exception; (2) it achieves SOTA or near-SOTA results even at low NFE counts; and (3) it can be seamlessly integrated into existing diffusion model-based solutions for inverse problems, such as DPS \cite{chung2022diffusion} and Red-diff \cite{mardani2023variational}. For example, DPS-MO attains a peak signal-to-noise ratio (PSNR) of 28.71 dB on the FFHQ 256 dataset for high dynamic range imaging, setting a new SOTA benchmark with only 100 NFEs, whereas current methods require between 1000 and 4000 NFEs for comparable performance.
Authors: Yibin Liu, Jianyu Zhang, Li Zhang, Shijian Li, Gang Pan
Abstract: Text-to-image (T2I) generation aims at producing realistic images corresponding to text descriptions. Generative Adversarial Network (GAN) has proven to be successful in this task. Typical T2I GANs are 2 phase methods that first pretrain an inter-modal representation from aligned image-text pairs and then use GAN to train image generator on that basis. However, such representation ignores the inner-modal semantic correspondence, e.g. the images with same label. The semantic label in priory describes the inherent distribution pattern with underlying cross-image relationships, which is supplement to the text description for understanding the full characteristics of image. In this paper, we propose a framework leveraging both inter- and inner-modal correspondence by label guided supervised contrastive learning. We extend the T2I GANs to two parameter-sharing contrast branches in both pretraining and generation phases. This integration effectively clusters the semantically similar image-text pair representations, thereby fostering the generation of higher-quality images. We demonstrate our framework on four novel T2I GANs by both single-object dataset CUB and multi-object dataset COCO, achieving significant improvements in the Inception Score (IS) and Frechet Inception Distance (FID) metrics of imagegeneration evaluation. Notably, on more complex multi-object COCO, our framework improves FID by 30.1%, 27.3%, 16.2% and 17.1% for AttnGAN, DM-GAN, SSA-GAN and GALIP, respectively. We also validate our superiority by comparing with other label guided T2I GANs. The results affirm the effectiveness and competitiveness of our approach in advancing the state-of-the-art GAN for T2I generation
Authors: Hao Zhu, Yan Zhu, Jiayu Xiao, Tianxiang Xiao, Yike Ma, Yucheng Zhang, Feng Dai
Abstract: Automated crop mapping through Satellite Image Time Series (SITS) has emerged as a crucial avenue for agricultural monitoring and management. However, due to the low resolution and unclear parcel boundaries, annotating pixel-level masks is exceptionally complex and time-consuming in SITS. This paper embraces the weakly supervised paradigm (i.e., only image-level categories available) to liberate the crop mapping task from the exhaustive annotation burden. The unique characteristics of SITS give rise to several challenges in weakly supervised learning: (1) noise perturbation from spatially neighboring regions, and (2) erroneous semantic bias from anomalous temporal periods. To address the above difficulties, we propose a novel method, termed exploring space-time perceptive clues (Exact). First, we introduce a set of spatial clues to explicitly capture the representative patterns of different crops from the most class-relative regions. Besides, we leverage the temporal-to-class interaction of the model to emphasize the contributions of pivotal clips, thereby enhancing the model perception for crop regions. Build upon the space-time perceptive clues, we derive the clue-based CAMs to effectively supervise the SITS segmentation network. Our method demonstrates impressive performance on various SITS benchmarks. Remarkably, the segmentation network trained on Exact-generated masks achieves 95% of its fully supervised performance, showing the bright promise of weakly supervised paradigm in crop mapping scenario. Our code will be publicly available.
Authors: Zuo Zuo, Jiahao Dong, Yue Gao, Zongze Wu
Abstract: In the manufacturing industry, defect detection is an essential but challenging task aiming to detect defects generated in the process of production. Though traditional YOLO models presents a good performance in defect detection, they still have limitations in capturing high-order feature interrelationships, which hurdles defect detection in the complex scenarios and across the scales. To this end, we introduce hypergraph computation into YOLO framework, dubbed HyperDefect-YOLO (HD-YOLO), to improve representative ability and semantic exploitation. HD-YOLO consists of Defect Aware Module (DAM) and Mixed Graph Network (MGNet) in the backbone, which specialize for perception and extraction of defect features. To effectively aggregate multi-scale features, we propose HyperGraph Aggregation Network (HGANet) which combines hypergraph and attention mechanism to aggregate multi-scale features. Cross-Scale Fusion (CSF) is proposed to adaptively fuse and handle features instead of simple concatenation and convolution. Finally, we propose Semantic Aware Module (SAM) in the neck to enhance semantic exploitation for accurately localizing defects with different sizes in the disturbed background. HD-YOLO undergoes rigorous evaluation on public HRIPCB and NEU-DET datasets with significant improvements compared to state-of-the-art methods. We also evaluate HD-YOLO on self-built MINILED dataset collected in real industrial scenarios to demonstrate the effectiveness of the proposed method. The source codes are at https://github.com/Jay-zzcoder/HD-YOLO.
Authors: Jon Guti\'errez-Zaballa, Koldo Basterretxea, Javier Echanobe, M. Victoria Mart\'inez, In\'es del Campo
Abstract: Advanced Driver Assistance Systems (ADAS) are designed with the main purpose of increasing the safety and comfort of vehicle occupants. Most of current computer vision-based ADAS perform detection and tracking tasks quite successfully under regular conditions, but are not completely reliable, particularly under adverse weather and changing lighting conditions, neither in complex situations with many overlapping objects. In this work we explore the use of hyperspectral imaging (HSI) in ADAS on the assumption that the distinct near infrared (NIR) spectral reflectances of different materials can help to better separate the objects in a driving scene. In particular, this paper describes some experimental results of the application of fully convolutional networks (FCN) to the image segmentation of HSI for ADAS applications. More specifically, our aim is to investigate to what extent the spatial features codified by convolutional filters can be helpful to improve the performance of HSI segmentation systems. With that aim, we use the HSI-Drive v1.1 dataset, which provides a set of labelled images recorded in real driving conditions with a small-size snapshot NIR-HSI camera. Finally, we analyze the implementability of such a HSI segmentation system by prototyping the developed FCN model together with the necessary hyperspectral cube preprocessing stage and characterizing its performance on an MPSoC.
Authors: Lars Schmarje, Kaspar Sakman, Reinhard Koch, Dan Zhang
Abstract: Autonomous driving (AD) operates in open-world scenarios, where encountering unknown objects is inevitable. However, standard object detectors trained on a limited number of base classes tend to ignore any unknown objects, posing potential risks on the road. To address this, it is important to learn a generic rather than a class specific objectness from objects seen during training. We therefore introduce an occupancy prediction together with bounding box regression. It learns to score the objectness by calculating the ratio of the predicted area occupied by actual objects. To enhance its generalizability, we increase the object diversity by exploiting data from other domains via Mosaic and Mixup augmentation. The objects outside the AD training classes are classified as a newly added out-of-distribution (OOD) class. Our solution UNCOVER, for UNknown Class Object detection for autonomous VEhicles in Real-time, excels at achieving both real-time detection and high recall of unknown objects on challenging AD benchmarks. To further attain very low false positive rates, particularly for close objects, we introduce a post-hoc filtering step that utilizes geometric cues extracted from the depth map, typically available within the AD system.
Authors: Ozer Can Devecioglu, Serkan Kiranyaz, Turker Ince, Moncef Gabbouj
Abstract: The exploration of underwater environments is essential for applications such as biological research, archaeology, and infrastructure maintenanceHowever, underwater imaging is challenging due to the waters unique properties, including scattering, absorption, color distortion, and reduced visibility. To address such visual degradations, a variety of approaches have been proposed covering from basic signal processing methods to deep learning models; however, none of them has proven to be consistently successful. In this paper, we propose a novel machine learning model, Co-Operational Regressor Networks (CoRe-Nets), designed to achieve the best possible underwater image restoration. A CoRe-Net consists of two co-operating networks: the Apprentice Regressor (AR), responsible for image transformation, and the Master Regressor (MR), which evaluates the Peak Signal-to-Noise Ratio (PSNR) of the images generated by the AR and feeds it back to AR. CoRe-Nets are built on Self-Organized Operational Neural Networks (Self-ONNs), which offer a superior learning capability by modulating nonlinearity in kernel transformations. The effectiveness of the proposed model is demonstrated on the benchmark Large Scale Underwater Image (LSUI) dataset. Leveraging the joint learning capabilities of the two cooperating networks, the proposed model achieves the state-of-art restoration performance with significantly reduced computational complexity and often presents such results that can even surpass the visual quality of the ground truth with a 2-pass application. Our results and the optimized PyTorch implementation of the proposed approach are now publicly shared on GitHub.
Authors: Sejong Yang, Seoung Wug Oh, Yang Zhou, Seon Joo Kim
Abstract: We introduce a novel approach for high-resolution talking head generation from a single image and audio input. Prior methods using explicit face models, like 3D morphable models (3DMM) and facial landmarks, often fall short in generating high-fidelity videos due to their lack of appearance-aware motion representation. While generative approaches such as video diffusion models achieve high video quality, their slow processing speeds limit practical application. Our proposed model, Implicit Face Motion Diffusion Model (IF-MDM), employs implicit motion to encode human faces into appearance-aware compressed facial latents, enhancing video generation. Although implicit motion lacks the spatial disentanglement of explicit models, which complicates alignment with subtle lip movements, we introduce motion statistics to help capture fine-grained motion information. Additionally, our model provides motion controllability to optimize the trade-off between motion intensity and visual quality during inference. IF-MDM supports real-time generation of 512x512 resolution videos at up to 45 frames per second (fps). Extensive evaluations demonstrate its superior performance over existing diffusion and explicit face models. The code will be released publicly, available alongside supplementary materials. The video results can be found on https://bit.ly/ifmdm_supplementary.
Authors: Kangan Qian, Xinyu Jiao, Yining Shi, Yunlong Wang, Ziang Luo, Zheng Fu, Kun Jiang, Diange Yang
Abstract: Reliable perception of spatial and motion information is crucial for safe autonomous navigation. Traditional approaches typically fall into two categories: object-centric and class-agnostic methods. While object-centric methods often struggle with missed detections, leading to inaccuracies in motion prediction, many class-agnostic methods focus heavily on encoder design, often overlooking important priors like rigidity and temporal consistency, leading to suboptimal performance, particularly with sparse LiDAR data at distant region. To address these issues, we propose $\textbf{PriorMotion}$, a generative framework that extracts rasterized and vectorized scene representations to model spatio-temporal priors. Our model comprises a BEV encoder, an Raster-Vector prior Encoder, and a Spatio-Temporal prior Generator, improving both spatial and temporal consistency in motion prediction. Additionally, we introduce a standardized evaluation protocol for class-agnostic motion prediction. Experiments on the nuScenes dataset show that PriorMotion achieves state-of-the-art performance, with further validation on advanced FMCW LiDAR confirming its robustness.
Authors: Th\'eo Sourget, Michelle Hestbek-M{\o}ller, Amelia Jim\'enez-S\'anchez, Jack Junchi Xu, Veronika Cheplygina
Abstract: The development of larger models for medical image analysis has led to increased performance. However, it also affected our ability to explain and validate model decisions. Models can use non-relevant parts of images, also called spurious correlations or shortcuts, to obtain high performance on benchmark datasets but fail in real-world scenarios. In this work, we challenge the capacity of convolutional neural networks (CNN) to classify chest X-rays and eye fundus images while masking out clinically relevant parts of the image. We show that all models trained on the PadChest dataset, irrespective of the masking strategy, are able to obtain an Area Under the Curve (AUC) above random. Moreover, the models trained on full images obtain good performance on images without the region of interest (ROI), even superior to the one obtained on images only containing the ROI. We also reveal a possible spurious correlation in the Chaksu dataset while the performances are more aligned with the expectation of an unbiased model. We go beyond the performance analysis with the usage of the explainability method SHAP and the analysis of embeddings. We asked a radiology resident to interpret chest X-rays under different masking to complement our findings with clinical knowledge. Our code is available at https://github.com/TheoSourget/MMC_Masking and https://github.com/TheoSourget/MMC_Masking_EyeFundus
URLs: https://github.com/TheoSourget/MMC_Masking, https://github.com/TheoSourget/MMC_Masking_EyeFundus
Authors: Yongming Zhu, Longhao Zhang, Zhengkun Rong, Tianshu Hu, Shuang Liang, Zhipeng Ge
Abstract: Imagine having a conversation with a socially intelligent agent. It can attentively listen to your words and offer visual and linguistic feedback promptly. This seamless interaction allows for multiple rounds of conversation to flow smoothly and naturally. In pursuit of actualizing it, we propose INFP, a novel audio-driven head generation framework for dyadic interaction. Unlike previous head generation works that only focus on single-sided communication, or require manual role assignment and explicit role switching, our model drives the agent portrait dynamically alternates between speaking and listening state, guided by the input dyadic audio. Specifically, INFP comprises a Motion-Based Head Imitation stage and an Audio-Guided Motion Generation stage. The first stage learns to project facial communicative behaviors from real-life conversation videos into a low-dimensional motion latent space, and use the motion latent codes to animate a static image. The second stage learns the mapping from the input dyadic audio to motion latent codes through denoising, leading to the audio-driven head generation in interactive scenarios. To facilitate this line of research, we introduce DyConv, a large scale dataset of rich dyadic conversations collected from the Internet. Extensive experiments and visualizations demonstrate superior performance and effectiveness of our method. Project Page: https://grisoon.github.io/INFP/.
Authors: Yiping Li, Romy van Jaarsveld, Ronald de Jong, Jasper Bongers, Gino Kuiper, Richard van Hillegersberg, Jelle Ruurda, Marcel Breeuwer, Yasmina Al Khalil
Abstract: Robotic-assisted minimally invasive esophagectomy (RAMIE) is a recognized treatment for esophageal cancer, offering better patient outcomes compared to open surgery and traditional minimally invasive surgery. RAMIE is highly complex, spanning multiple anatomical areas and involving repetitive phases and non-sequential phase transitions. Our goal is to leverage deep learning for surgical phase recognition in RAMIE to provide intraoperative support to surgeons. To achieve this, we have developed a new surgical phase recognition dataset comprising 27 videos. Using this dataset, we conducted a comparative analysis of state-of-the-art surgical phase recognition models. To more effectively capture the temporal dynamics of this complex procedure, we developed a novel deep learning model featuring an encoder-decoder structure with causal hierarchical attention, which demonstrates superior performance compared to existing models.
Authors: Yefei He, Feng Chen, Yuanyu He, Shaoxuan He, Hong Zhou, Kaipeng Zhang, Bohan Zhuang
Abstract: In this paper, we propose ZipAR, a training-free, plug-and-play parallel decoding framework for accelerating auto-regressive (AR) visual generation. The motivation stems from the observation that images exhibit local structures, and spatially distant regions tend to have minimal interdependence. Given a partially decoded set of visual tokens, in addition to the original next-token prediction scheme in the row dimension, the tokens corresponding to spatially adjacent regions in the column dimension can be decoded in parallel, enabling the ``next-set prediction'' paradigm. By decoding multiple tokens simultaneously in a single forward pass, the number of forward passes required to generate an image is significantly reduced, resulting in a substantial improvement in generation efficiency. Experiments demonstrate that ZipAR can reduce the number of model forward passes by up to 91% on the Emu3-Gen model without requiring any additional retraining.
Authors: A. Enes Doruk, Erhan Oztop, Hasan F. Ates
Abstract: Unsupervised Domain Adaptation (UDA) aims to utilize labeled data from a source domain to solve tasks in an unlabeled target domain, often hindered by significant domain gaps. Traditional CNN-based methods struggle to fully capture complex domain relationships, motivating the shift to vision transformers like the Swin Transformer, which excel in modeling both local and global dependencies. In this work, we propose a novel UDA approach leveraging the Swin Transformer with three key modules. A Graph Domain Discriminator enhances domain alignment by capturing inter-pixel correlations through graph convolutions and entropy-based attention differentiation. An Adaptive Double Attention module combines Windows and Shifted Windows attention with dynamic reweighting to align long-range and local features effectively. Finally, a Cross-Feature Transform modifies Swin Transformer blocks to improve generalization across domains. Extensive benchmarks confirm the state-of-the-art performance of our versatile method, which requires no task-specific alignment modules, establishing its adaptability to diverse applications.
Authors: Seokju Yun, Seunghye Chae, Dongheon Lee, Youngmin Ro
Abstract: Domain generalization (DG) aims to adapt a model using one or multiple source domains to ensure robust performance in unseen target domains. Recently, Parameter-Efficient Fine-Tuning (PEFT) of foundation models has shown promising results in the context of DG problem. Nevertheless, existing PEFT methods still struggle to strike a balance between preserving generalizable components of the pre-trained model and learning task-specific features. To gain insights into the distribution of generalizable components, we begin by analyzing the pre-trained weights through the lens of singular value decomposition. Building on these insights, we introduce Singular Value Decomposed Low-Rank Adaptation (SoRA), an approach that selectively tunes minor singular components while keeping the residual parts frozen. SoRA effectively retains the generalization ability of the pre-trained model while efficiently acquiring task-specific skills. Furthermore, we freeze domain-generalizable blocks and employ an annealing weight decay strategy, thereby achieving an optimal balance in the delicate trade-off between generalizability and discriminability. SoRA attains state-of-the-art results on multiple benchmarks that span both domain generalized semantic segmentation to domain generalized object detection. In addition, our methods introduce no additional inference overhead or regularization loss, maintain compatibility with any backbone or head, and are designed to be versatile, allowing easy integration into a wide range of tasks.
Authors: Hirunima Jayasekara, Khoi Pham, Nirat Saini, Abhinav Shrivastava
Abstract: Open-World Compositional Zero-Shot Learning (OW-CZSL) addresses the challenge of recognizing novel compositions of known primitives and entities. Even though prior works utilize language knowledge for recognition, such approaches exhibit limited interactions between language-image modalities. Our approach primarily focuses on enhancing the inter-modality interactions through fostering richer interactions between image and textual data. Additionally, we introduce a novel module aimed at alleviating the computational burden associated with exhaustive exploration of all possible compositions during the inference stage. While previous methods exclusively learn compositions jointly or independently, we introduce an advanced hybrid procedure that leverages both learning mechanisms to generate final predictions. Our proposed model, achieves state-of-the-art in OW-CZSL in three datasets, while surpassing Large Vision Language Models (LLVM) in two datasets.
Authors: Nefeli Andreou, Varsha Vivek, Ying Wang, Alex Vorobiov, Tiffany Deng, Raja Bala, Larry Davis, Betty Mohler Tesch
Abstract: Accurately generating images of human bodies from text remains a challenging problem for state of the art text-to-image models. Commonly observed body-related artifacts include extra or missing limbs, unrealistic poses, blurred body parts, etc. Currently, evaluation of such artifacts relies heavily on time-consuming human judgments, limiting the ability to benchmark models at scale. We address this by proposing BodyMetric, a learnable metric that predicts body realism in images. BodyMetric is trained on realism labels and multi-modal signals including 3D body representations inferred from the input image, and textual descriptions. In order to facilitate this approach, we design an annotation pipeline to collect expert ratings on human body realism leading to a new dataset for this task, namely, BodyRealism. Ablation studies support our architectural choices for BodyMetric and the importance of leveraging a 3D human body prior in capturing body-related artifacts in 2D images. In comparison to concurrent metrics which evaluate general user preference in images, BodyMetric specifically reflects body-related artifacts. We demonstrate the utility of BodyMetric through applications that were previously infeasible at scale. In particular, we use BodyMetric to benchmark the generation ability of text-to-image models to produce realistic human bodies. We also demonstrate the effectiveness of BodyMetric in ranking generated images based on the predicted realism scores.
Authors: Bingchen Li, Xin Li, Yiting Lu, Zhibo Chen
Abstract: We present the first loss agent, dubbed LossAgent, for low-level image processing tasks, e.g., image super-resolution and restoration, intending to achieve any customized optimization objectives of low-level image processing in different practical applications. Notably, not all optimization objectives, such as complex hand-crafted perceptual metrics, text description, and intricate human feedback, can be instantiated with existing low-level losses, e.g., MSE loss. which presents a crucial challenge in optimizing image processing networks in an end-to-end manner. To eliminate this, our LossAgent introduces the powerful large language model (LLM) as the loss agent, where the rich textual understanding of prior knowledge empowers the loss agent with the potential to understand complex optimization objectives, trajectory, and state feedback from external environments in the optimization process of the low-level image processing networks. In particular, we establish the loss repository by incorporating existing loss functions that support the end-to-end optimization for low-level image processing. Then, we design the optimization-oriented prompt engineering for the loss agent to actively and intelligently decide the compositional weights for each loss in the repository at each optimization interaction, thereby achieving the required optimization trajectory for any customized optimization objectives. Extensive experiments on three typical low-level image processing tasks and multiple optimization objectives have shown the effectiveness and applicability of our proposed LossAgent. Code and pre-trained models will be available at https://github.com/lbc12345/LossAgent.
Authors: Hamid Gadirov, Qi Wu, David Bauer, Kwan-Liu Ma, Jos Roerdink, Steffen Frey
Abstract: We present HyperFLINT (Hypernetwork-based FLow estimation and temporal INTerpolation), a novel deep learning-based approach for estimating flow fields, temporally interpolating scalar fields, and facilitating parameter space exploration in spatio-temporal scientific ensemble data. This work addresses the critical need to explicitly incorporate ensemble parameters into the learning process, as traditional methods often neglect these, limiting their ability to adapt to diverse simulation settings and provide meaningful insights into the data dynamics. HyperFLINT introduces a hypernetwork to account for simulation parameters, enabling it to generate accurate interpolations and flow fields for each timestep by dynamically adapting to varying conditions, thereby outperforming existing parameter-agnostic approaches. The architecture features modular neural blocks with convolutional and deconvolutional layers, supported by a hypernetwork that generates weights for the main network, allowing the model to better capture intricate simulation dynamics. A series of experiments demonstrates HyperFLINT's significantly improved performance in flow field estimation and temporal interpolation, as well as its potential in enabling parameter space exploration, offering valuable insights into complex scientific ensembles.
Authors: Meenakshi Gupta, Mingyuan Lei, Tat-Jen Cham, Hwee Kuan Lee
Abstract: This paper introduces a novel framework named D-LORD (Double Latent Optimization for Representation Disentanglement), which is designed for motion stylization (motion style transfer and motion retargeting). The primary objective of this framework is to separate the class and content information from a given motion sequence using a data-driven latent optimization approach. Here, class refers to person-specific style, such as a particular emotion or an individual's identity, while content relates to the style-agnostic aspect of an action, such as walking or jumping, as universally understood concepts. The key advantage of D-LORD is its ability to perform style transfer without needing paired motion data. Instead, it utilizes class and content labels during the latent optimization process. By disentangling the representation, the framework enables the transformation of one motion sequences style to another's style using Adaptive Instance Normalization. The proposed D-LORD framework is designed with a focus on generalization, allowing it to handle different class and content labels for various applications. Additionally, it can generate diverse motion sequences when specific class and content labels are provided. The framework's efficacy is demonstrated through experimentation on three datasets: the CMU XIA dataset for motion style transfer, the MHAD dataset, and the RRIS Ability dataset for motion retargeting. Notably, this paper presents the first generalized framework for motion style transfer and motion retargeting, showcasing its potential contributions in this area.
Authors: Haoning Wu, Ziheng Zhao, Ya Zhang, Weidi Xie, Yanfeng Wang
Abstract: Medical image segmentation has recently demonstrated impressive progress with deep neural networks, yet the heterogeneous modalities and scarcity of mask annotations limit the development of segmentation models on unannotated modalities. This paper investigates a new paradigm for leveraging generative models in medical applications: controllably synthesizing data for unannotated modalities, without requiring registered data pairs. Specifically, we make the following contributions in this paper: (i) we collect and curate a large-scale radiology image-text dataset, MedGen-1M, comprising modality labels, attributes, region, and organ information, along with a subset of organ mask annotations, to support research in controllable medical image generation; (ii) we propose a diffusion-based data engine, termed MRGen, which enables generation conditioned on text prompts and masks, synthesizing MR images for diverse modalities lacking mask annotations, to train segmentation models on unannotated modalities; (iii) we conduct extensive experiments across various modalities, illustrating that our data engine can effectively synthesize training samples and extend MRI segmentation towards unannotated modalities.
Authors: Serhii Svystun, Oleksandr Melnychenko, Pavlo Radiuk, Oleg Savenko, Anatoliy Sachenko, Andrii Lysyi
Abstract: The inspection of wind turbine blades (WTBs) is crucial for ensuring their structural integrity and operational efficiency. Traditional inspection methods can be dangerous and inefficient, prompting the use of unmanned aerial vehicles (UAVs) that access hard-to-reach areas and capture high-resolution imagery. In this study, we address the challenge of enhancing defect detection on WTBs by integrating thermal and RGB images obtained from UAVs. We propose a multispectral image composition method that combines thermal and RGB imagery through spatial coordinate transformation, key point detection, binary descriptor creation, and weighted image overlay. Using a benchmark dataset of WTB images annotated for defects, we evaluated several state-of-the-art object detection models. Our results show that composite images significantly improve defect detection efficiency. Specifically, the YOLOv8 model's accuracy increased from 91% to 95%, precision from 89% to 94%, recall from 85% to 92%, and F1-score from 87% to 93%. The number of false positives decreased from 6 to 3, and missed defects reduced from 5 to 2. These findings demonstrate that integrating thermal and RGB imagery enhances defect detection on WTBs, contributing to improved maintenance and reliability.
Authors: Erik Brorsson, Lennart Svensson, Kristofer Bengtsson, Knut {\AA}kesson
Abstract: We address multi-view pedestrian detection in a setting where labeled data is collected using a multi-camera setup different from the one used for testing. While recent multi-view pedestrian detectors perform well on the camera rig used for training, their performance declines when applied to a different setup. To facilitate seamless deployment across varied camera rigs, we propose an unsupervised domain adaptation (UDA) method that adapts the model to new rigs without requiring additional labeled data. Specifically, we leverage the mean teacher self-training framework with a novel pseudo-labeling technique tailored to multi-view pedestrian detection. This method achieves state-of-the-art performance on multiple benchmarks, including MultiviewX$\rightarrow$Wildtrack. Unlike previous methods, our approach eliminates the need for external labeled monocular datasets, thereby reducing reliance on labeled data. Extensive evaluations demonstrate the effectiveness of our method and validate key design choices. By enabling robust adaptation across camera setups, our work enhances the practicality of multi-view pedestrian detectors and establishes a strong UDA baseline for future research.
Authors: Thomas Walker, Salvatore Esposito, Daniel Rebain, Amir Vaxman, Arno Onken, Changjian Li, Oisin Mac Aodha
Abstract: Reconstructing complex structures from planar cross-sections is a challenging problem, with wide-reaching applications in medical imaging, manufacturing, and topography. Out-of-the-box point cloud reconstruction methods can often fail due to the data sparsity between slicing planes, while current bespoke methods struggle to reconstruct thin geometric structures and preserve topological continuity. This is important for medical applications where thin vessel structures are present in CT and MRI scans. This paper introduces \method, a novel approach for extracting a 3D signed distance field from 2D signed distances generated from planar contours. Our approach makes the training of neural SDFs contour-aware by using losses designed for the case where geometry is known within 2D slices. Our results demonstrate a significant improvement over existing methods, effectively reconstructing thin structures and producing accurate 3D models without the interpolation artifacts or over-smoothing of prior approaches.
Authors: Biquard Maud, Marie Chabert, Florence Genin, Christophe Latry, Thomas Oberlin
Abstract: Satellite optical images, upon their on-ground receipt, offer a distorted view of the observed scene. Their restoration, classically including denoising, deblurring, and sometimes super-resolution, is required before their exploitation. Moreover, quantifying the uncertainty related to this restoration could be valuable by lowering the risk of hallucination and avoiding propagating these biases in downstream applications. Deep learning methods are now state-of-the-art for satellite image restoration. However, they require to train a specific network for each sensor and they do not provide the associated uncertainties. This paper proposes a generic method involving a single network to restore images from several sensors and a scalable way to derive the uncertainties. We focus on deep regularization (DR) methods, which learn a deep prior on target images before plugging it into a model-based optimization scheme. First, we introduce VBLE-xz, which solves the inverse problem in the latent space of a variational compressive autoencoder, estimating the uncertainty jointly in the latent and in the image spaces. It enables scalable posterior sampling with relevant and calibrated uncertainties. Second, we propose the denoiser-based method SatDPIR, adapted from DPIR, which efficiently computes accurate point estimates. We conduct a comprehensive set of experiments on very high resolution simulated and real Pleiades images, asserting both the performance and robustness of the proposed methods. VBLE-xz and SatDPIR achieve state-of-the-art results compared to direct inversion methods. In particular, VBLE-xz is a scalable method to get realistic posterior samples and accurate uncertainties, while SatDPIR represents a compelling alternative to direct inversion methods when uncertainty quantification is not required.
Authors: Doyoung Park, Naresh Reddy Yarram, Sunjin Kim, Minkyu Kim, Seongho Cho, Taehee Lee
Abstract: Document comparison typically relies on optical character recognition (OCR) as its core technology. However, OCR requires the selection of appropriate language models for each document and the performance of multilingual or hybrid models remains limited. To overcome these challenges, we propose text change detection (TCD) using an image comparison model tailored for multilingual documents. Unlike OCR-based approaches, our method employs word-level text image-to-image comparison to detect changes. Our model generates bidirectional change segmentation maps between the source and target documents. To enhance performance without requiring explicit text alignment or scaling preprocessing, we employ correlations among multi-scale attention features. We also construct a benchmark dataset comprising actual printed and scanned word pairs in various languages to evaluate our model. We validate our approach using our benchmark dataset and public benchmarks Distorted Document Images and the LRDE Document Binarization Dataset. We compare our model against state-of-the-art semantic segmentation and change detection models, as well as to conventional OCR-based models.
Authors: Xinghui Li, Qichao Sun, Pengze Zhang, Fulong Ye, Zhichao Liao, Wanquan Feng, Songtao Zhao, Qian He
Abstract: Recent advances in garment-centric image generation from text and image prompts based on diffusion models are impressive. However, existing methods lack support for various combinations of attire, and struggle to preserve the garment details while maintaining faithfulness to the text prompts, limiting their performance across diverse scenarios. In this paper, we focus on a new task, i.e., Multi-Garment Virtual Dressing, and we propose a novel AnyDressing method for customizing characters conditioned on any combination of garments and any personalized text prompts. AnyDressing comprises two primary networks named GarmentsNet and DressingNet, which are respectively dedicated to extracting detailed clothing features and generating customized images. Specifically, we propose an efficient and scalable module called Garment-Specific Feature Extractor in GarmentsNet to individually encode garment textures in parallel. This design prevents garment confusion while ensuring network efficiency. Meanwhile, we design an adaptive Dressing-Attention mechanism and a novel Instance-Level Garment Localization Learning strategy in DressingNet to accurately inject multi-garment features into their corresponding regions. This approach efficiently integrates multi-garment texture cues into generated images and further enhances text-image consistency. Additionally, we introduce a Garment-Enhanced Texture Learning strategy to improve the fine-grained texture details of garments. Thanks to our well-craft design, AnyDressing can serve as a plug-in module to easily integrate with any community control extensions for diffusion models, improving the diversity and controllability of synthesized images. Extensive experiments show that AnyDressing achieves state-of-the-art results.
Authors: Haitian Zhang, Xiangyuan Wang, Chang Xu, Xinya Wang, Fang Xu, Huai Yu, Lei Yu, Wen Yang
Abstract: Fusing Events and RGB images for object detection leverages the robustness of Event cameras in adverse environments and the rich semantic information provided by RGB cameras. However, two critical mismatches: low-latency Events \textit{vs.}~high-latency RGB frames; temporally sparse labels in training \textit{vs.}~continuous flow in inference, significantly hinder the high-frequency fusion-based object detection. To address these challenges, we propose the \textbf{F}requency-\textbf{A}daptive Low-Latency \textbf{O}bject \textbf{D}etector (FAOD). FAOD aligns low-frequency RGB frames with high-frequency Events through an Align Module, which reinforces cross-modal style and spatial proximity to address the Event-RGB Mismatch. We further propose a training strategy, Time Shift, which enforces the module to align the prediction from temporally shifted Event-RGB pairs and their original representation, that is, consistent with Event-aligned annotations. This strategy enables the network to use high-frequency Event data as the primary reference while treating low-frequency RGB images as supplementary information, retaining the low-latency nature of the Event stream toward high-frequency detection. Furthermore, we observe that these corrected Event-RGB pairs demonstrate better generalization from low training frequency to higher inference frequencies compared to using Event data alone. Extensive experiments on the PKU-DAVIS-SOD and DSEC-Detection datasets demonstrate that our FAOD achieves SOTA performance. Specifically, in the PKU-DAVIS-SOD Dataset, FAOD achieves 9.8 points improvement in terms of the mAP in fully paired Event-RGB data with only a quarter of the parameters compared to SODFormer, and even maintains robust performance (only a 3 points drop in mAP) under 80$\times$ Event-RGB frequency mismatch.
Authors: Yayuan Li, Zhi Cao, Jason J. Corso
Abstract: Despite the recent strides in video generation, state-of-the-art methods still struggle with elements of visual detail. One particularly challenging case is the class of egocentric instructional videos in which the intricate motion of the hand coupled with a mostly stable and non-distracting environment is necessary to convey the appropriate visual action instruction. To address these challenges, we introduce a new method for instructional video generation. Our diffusion-based method incorporates two distinct innovations. First, we propose an automatic method to generate the expected region of motion, guided by both the visual context and the action text. Second, we introduce a critical hand structure loss to guide the diffusion model to focus on smooth and consistent hand poses. We evaluate our method on augmented instructional datasets based on EpicKitchens and Ego4D, demonstrating significant improvements over state-of-the-art methods in terms of instructional clarity, especially of the hand motion in the target region, across diverse environments and actions.Video results can be found on the project webpage: https://excitedbutter.github.io/Instructional-Video-Generation/
URLs: https://excitedbutter.github.io/Instructional-Video-Generation/
Authors: Shuang Xu, Zixiang Zhao, Haowen Bai, Chang Yu, Jiangjun Peng, Xiangyong Cao, Deyu Meng
Abstract: Hyperspectral images (HSIs) are frequently noisy and of low resolution due to the constraints of imaging devices. Recently launched satellites can concurrently acquire HSIs and panchromatic (PAN) images, enabling the restoration of HSIs to generate clean and high-resolution imagery through fusing PAN images for denoising and super-resolution. However, previous studies treated these two tasks as independent processes, resulting in accumulated errors. This paper introduces \textbf{H}yperspectral \textbf{I}mage Joint \textbf{Pand}enoising \textbf{a}nd Pan\textbf{s}harpening (Hipandas), a novel learning paradigm that reconstructs HRHS images from noisy low-resolution HSIs (LRHS) and high-resolution PAN images. The proposed zero-shot Hipandas framework consists of a guided denoising network, a guided super-resolution network, and a PAN reconstruction network, utilizing an HSI low-rank prior and a newly introduced detail-oriented low-rank prior. The interconnection of these networks complicates the training process, necessitating a two-stage training strategy to ensure effective training. Experimental results on both simulated and real-world datasets indicate that the proposed method surpasses state-of-the-art algorithms, yielding more accurate and visually pleasing HRHS images.
Authors: Valerio Marsocci, Yuru Jia, Georges Le Bellier, David Kerekes, Liang Zeng, Sebastian Hafner, Sebastian Gerard, Eric Brune, Ritu Yadav, Ali Shibli, Heng Fang, Yifang Ban, Maarten Vergauwen, Nicolas Audebert, Andrea Nascetti
Abstract: Geospatial Foundation Models (GFMs) have emerged as powerful tools for extracting representations from Earth observation data, but their evaluation remains inconsistent and narrow. Existing works often evaluate on suboptimal downstream datasets and tasks, that are often too easy or too narrow, limiting the usefulness of the evaluations to assess the real-world applicability of GFMs. Additionally, there is a distinct lack of diversity in current evaluation protocols, which fail to account for the multiplicity of image resolutions, sensor types, and temporalities, which further complicates the assessment of GFM performance. In particular, most existing benchmarks are geographically biased towards North America and Europe, questioning the global applicability of GFMs. To overcome these challenges, we introduce PANGAEA, a standardized evaluation protocol that covers a diverse set of datasets, tasks, resolutions, sensor modalities, and temporalities. It establishes a robust and widely applicable benchmark for GFMs. We evaluate the most popular GFMs openly available on this benchmark and analyze their performance across several domains. In particular, we compare these models to supervised baselines (e.g. UNet and vanilla ViT), and assess their effectiveness when faced with limited labeled data. Our findings highlight the limitations of GFMs, under different scenarios, showing that they do not consistently outperform supervised models. PANGAEA is designed to be highly extensible, allowing for the seamless inclusion of new datasets, models, and tasks in future research. By releasing the evaluation code and benchmark, we aim to enable other researchers to replicate our experiments and build upon our work, fostering a more principled evaluation protocol for large pre-trained geospatial models. The code is available at https://github.com/VMarsocci/pangaea-bench.
Authors: Eliseo Fuentes-Mart\'inez, Antonio R\'ios-Vila, Juan C. Martinez-Sevilla, David Rizo, Jorge Calvo-Zaragoza
Abstract: The digitization of vocal music scores presents unique challenges that go beyond traditional Optical Music Recognition (OMR) and Optical Character Recognition (OCR), as it necessitates preserving the critical alignment between music notation and lyrics. This alignment is essential for proper interpretation and processing in practical applications. This paper introduces and formalizes, for the first time, the Aligned Music Notation and Lyrics Transcription (AMNLT) challenge, which addresses the complete transcription of vocal scores by jointly considering music symbols, lyrics, and their synchronization. We analyze different approaches to address this challenge, ranging from traditional divide-and-conquer methods that handle music and lyrics separately, to novel end-to-end solutions including direct transcription, unfolding mechanisms, and language modeling. To evaluate these methods, we introduce four datasets of Gregorian chants, comprising both real and synthetic sources, along with custom metrics specifically designed to assess both transcription and alignment accuracy. Our experimental results demonstrate that end-to-end approaches generally outperform heuristic methods in the alignment challenge, with language models showing particular promise in scenarios where sufficient training data is available. This work establishes the first comprehensive framework for AMNLT, providing both theoretical foundations and practical solutions for preserving and digitizing vocal music heritage.
Authors: Chenyang Zhu, Bin Xiao, Lin Shi, Shoukun Xu, Xu Zheng
Abstract: The recent Segment Anything Model (SAM) represents a significant breakthrough in scaling segmentation models, delivering strong performance across various downstream applications in the RGB modality. However, directly applying SAM to emerging visual modalities, such as depth and event data results in suboptimal performance in multi-modal segmentation tasks. In this paper, we make the first attempt to adapt SAM for multi-modal semantic segmentation by proposing a Mixture of Low-Rank Adaptation Experts (MoE-LoRA) tailored for different input visual modalities. By training only the MoE-LoRA layers while keeping SAM's weights frozen, SAM's strong generalization and segmentation capabilities can be preserved for downstream tasks. Specifically, to address cross-modal inconsistencies, we propose a novel MoE routing strategy that adaptively generates weighted features across modalities, enhancing multi-modal feature integration. Additionally, we incorporate multi-scale feature extraction and fusion by adapting SAM's segmentation head and introducing an auxiliary segmentation head to combine multi-scale features for improved segmentation performance effectively. Extensive experiments were conducted on three multi-modal benchmarks: DELIVER, MUSES, and MCubeS. The results consistently demonstrate that the proposed method significantly outperforms state-of-the-art approaches across diverse scenarios. Notably, under the particularly challenging condition of missing modalities, our approach exhibits a substantial performance gain, achieving an improvement of 32.15% compared to existing methods.
Authors: Shihua Huang, Zhichao Lu, Xiaodong Cun, Yongjun Yu, Xiao Zhou, Xi Shen
Abstract: We introduce DEIM, an innovative and efficient training framework designed to accelerate convergence in real-time object detection with Transformer-based architectures (DETR). To mitigate the sparse supervision inherent in one-to-one (O2O) matching in DETR models, DEIM employs a Dense O2O matching strategy. This approach increases the number of positive samples per image by incorporating additional targets, using standard data augmentation techniques. While Dense O2O matching speeds up convergence, it also introduces numerous low-quality matches that could affect performance. To address this, we propose the Matchability-Aware Loss (MAL), a novel loss function that optimizes matches across various quality levels, enhancing the effectiveness of Dense O2O. Extensive experiments on the COCO dataset validate the efficacy of DEIM. When integrated with RT-DETR and D-FINE, it consistently boosts performance while reducing training time by 50%. Notably, paired with RT-DETRv2, DEIM achieves 53.2% AP in a single day of training on an NVIDIA 4090 GPU. Additionally, DEIM-trained real-time models outperform leading real-time object detectors, with DEIM-D-FINE-L and DEIM-D-FINE-X achieving 54.7% and 56.5% AP at 124 and 78 FPS on an NVIDIA T4 GPU, respectively, without the need for additional data. We believe DEIM sets a new baseline for advancements in real-time object detection. Our code and pre-trained models are available at https://github.com/ShihuaHuang95/DEIM.
Authors: Jiahao Zhang, Ryota Yoshihashi, Shunsuke Kitada, Atsuki Osanai, Yuta Nakashima
Abstract: Large language models (LLMs) have proven effective for layout generation due to their ability to produce structure-description languages, such as HTML or JSON, even without access to visual information. Recently, LLM providers have evolved these models into large vision-language models (LVLM), which shows prominent multi-modal understanding capabilities. Then, how can we leverage this multi-modal power for layout generation? To answer this, we propose Visual-Aware Self-Correction LAyout GeneRation (VASCAR) for LVLM-based content-aware layout generation. In our method, LVLMs iteratively refine their outputs with reference to rendered layout images, which are visualized as colored bounding boxes on poster backgrounds. In experiments, we demonstrate that our method combined with the Gemini. Without any additional training, VASCAR achieves state-of-the-art (SOTA) layout generation quality outperforming both existing layout-specific generative models and other LLM-based methods.
Authors: Yixin Zhang, Nicholas Konz, Kevin Kramer, Maciej A. Mazurowski
Abstract: Segment Anything Model (SAM) has shown impressive performance in interactive and zero-shot segmentation across diverse domains, suggesting that they have learned a general concept of "objects" from their large-scale training. However, we observed that SAM struggles with certain types of objects, particularly those featuring dense, tree-like structures and low textural contrast from their surroundings. These failure modes are critical for understanding its limitations in real-world use. In order to systematically examine this issue, we propose metrics to quantify two key object characteristics: tree-likeness and textural separability. Through extensive controlled synthetic experiments and testing on real datasets, we demonstrate that SAM's performance is noticeably correlated with these factors. We link these behaviors under the concept of "textural confusion", where SAM misinterprets local structure as global texture, leading to over-segmentation, or struggles to differentiate objects from similarly textured backgrounds. These findings offer the first quantitative framework to model SAM's challenges, providing valuable insights into its limitations and guiding future improvements for vision foundation models.
Authors: Rao Fu, Dingxi Zhang, Alex Jiang, Wanjia Fu, Austin Funk, Daniel Ritchie, Srinath Sridhar
Abstract: Understanding bimanual human hand activities is a critical problem in AI and robotics. We cannot build large models of bimanual activities because existing datasets lack the scale, coverage of diverse hand activities, and detailed annotations. We introduce GigaHands, a massive annotated dataset capturing 34 hours of bimanual hand activities from 56 subjects and 417 objects, totaling 14k motion clips derived from 183 million frames paired with 84k text annotations. Our markerless capture setup and data acquisition protocol enable fully automatic 3D hand and object estimation while minimizing the effort required for text annotation. The scale and diversity of GigaHands enable broad applications, including text-driven action synthesis, hand motion captioning, and dynamic radiance field reconstruction.
Authors: Bernd Prach, Christoph H. Lampert
Abstract: Despite extensive research since the community learned about adversarial examples 10 years ago, we still do not know how to train high-accuracy classifiers that are guaranteed to be robust to small perturbations of their inputs. Previous works often argued that this might be because no classifier exists that is robust and accurate at the same time. However, in computer vision this assumption does not match reality where humans are usually accurate and robust on most tasks of interest. We offer an alternative explanation and show that in certain settings robust generalization is only possible with unrealistically large amounts of data. More precisely we find a setting where a robust classifier exists, it is easy to learn an accurate classifier, yet it requires an exponential amount of data to learn a robust classifier. Based on this theoretical result, we explore how well robust classifiers generalize on datasets such as CIFAR-10. We come to the conclusion that on this datasets, the limitation of current robust models also lies in the generalization, and that they require a lot of data to do well on the test set. We also show that the problem is not in the expressiveness or generalization capabilities of current architectures, and that there are low magnitude features in the data which are useful for non-robust generalization but are not available for robust classifiers.
Authors: Marco Garosi, Riccardo Tedoldi, Davide Boscaini, Massimiliano Mancini, Nicu Sebe, Fabio Poiesi
Abstract: Supervised 3D part segmentation models are tailored for a fixed set of objects and parts, limiting their transferability to open-set, real-world scenarios. Recent works have explored vision-language models (VLMs) as a promising alternative, using multi-view rendering and textual prompting to identify object parts. However, naively applying VLMs in this context introduces several drawbacks, such as the need for meticulous prompt engineering, and fails to leverage the 3D geometric structure of objects. To address these limitations, we propose COPS, a COmprehensive model for Parts Segmentation that blends the semantics extracted from visual concepts and 3D geometry to effectively identify object parts. COPS renders a point cloud from multiple viewpoints, extracts 2D features, projects them back to 3D, and uses a novel geometric-aware feature aggregation procedure to ensure spatial and semantic consistency. Finally, it clusters points into parts and labels them. We demonstrate that COPS is efficient, scalable, and achieves zero-shot state-of-the-art performance across five datasets, covering synthetic and real-world data, texture-less and coloured objects, as well as rigid and non-rigid shapes. The code is available at https://3d-cops.github.io.
Authors: Il\'an Carretero, Pablo Meseguer, Roc\'io del Amor, Valery Naranjo
Abstract: Domain shift in the field of histopathological imaging is a common phenomenon due to the intra- and inter-hospital variability of staining and digitization protocols. The implementation of robust models, capable of creating generalized domains, represents a need to be solved. In this work, a new domain adaptation method to deal with the variability between histopathological images from multiple centers is presented. In particular, our method adds a training constraint to the supervised contrastive learning approach to achieve domain adaptation and improve inter-class separability. Experiments performed on domain adaptation and classification of whole-slide images of six skin cancer subtypes from two centers demonstrate the method's usefulness. The results reflect superior performance compared to not using domain adaptation after feature extraction or staining normalization.
Authors: Alan Li, Angela P. Schoellig
Abstract: 6D Object pose estimation is a fundamental component in robotics enabling efficient interaction with the environment. It is particularly challenging in bin-picking applications, where objects may be textureless and in difficult poses, and occlusion between objects of the same type may cause confusion even in well-trained models. We propose a novel method of hard example synthesis that is model-agnostic, using existing simulators and the modeling of pose error in both the camera-to-object viewsphere and occlusion space. Through evaluation of the model performance with respect to the distribution of object poses and occlusions, we discover regions of high error and generate realistic training samples to specifically target these regions. With our training approach, we demonstrate an improvement in correct detection rate of up to 20% across several ROBI-dataset objects using state-of-the-art pose estimation models.
Authors: Jinbin Bai, Wei Chow, Ling Yang, Xiangtai Li, Juncheng Li, Hanwang Zhang, Shuicheng Yan
Abstract: We present HumanEdit, a high-quality, human-rewarded dataset specifically designed for instruction-guided image editing, enabling precise and diverse image manipulations through open-form language instructions. Previous large-scale editing datasets often incorporate minimal human feedback, leading to challenges in aligning datasets with human preferences. HumanEdit bridges this gap by employing human annotators to construct data pairs and administrators to provide feedback. With meticulously curation, HumanEdit comprises 5,751 images and requires more than 2,500 hours of human effort across four stages, ensuring both accuracy and reliability for a wide range of image editing tasks. The dataset includes six distinct types of editing instructions: Action, Add, Counting, Relation, Remove, and Replace, encompassing a broad spectrum of real-world scenarios. All images in the dataset are accompanied by masks, and for a subset of the data, we ensure that the instructions are sufficiently detailed to support mask-free editing. Furthermore, HumanEdit offers comprehensive diversity and high-resolution $1024 \times 1024$ content sourced from various domains, setting a new versatile benchmark for instructional image editing datasets. With the aim of advancing future research and establishing evaluation benchmarks in the field of image editing, we release HumanEdit at \url{https://huggingface.co/datasets/BryanW/HumanEdit}.
Authors: Bingbing Hu, Yanyan Li, Rui Xie, Bo Xu, Haoye Dong, Junfeng Yao, Gim Hee Lee
Abstract: Capturing the temporal evolution of Gaussian properties such as position, rotation, and scale is a challenging task due to the vast number of time-varying parameters and the limited photometric data available, which generally results in convergence issues, making it difficult to find an optimal solution. While feeding all inputs into an end-to-end neural network can effectively model complex temporal dynamics, this approach lacks explicit supervision and struggles to generate high-quality transformation fields. On the other hand, using time-conditioned polynomial functions to model Gaussian trajectories and orientations provides a more explicit and interpretable solution, but requires significant handcrafted effort and lacks generalizability across diverse scenes. To overcome these limitations, this paper introduces a novel approach based on a learnable infinite Taylor Formula to model the temporal evolution of Gaussians. This method offers both the flexibility of an implicit network-based approach and the interpretability of explicit polynomial functions, allowing for more robust and generalizable modeling of Gaussian dynamics across various dynamic scenes. Extensive experiments on dynamic novel view rendering tasks are conducted on public datasets, demonstrating that the proposed method achieves state-of-the-art performance in this domain. More information is available on our project page(https://ellisonking.github.io/TaylorGaussian).
Authors: Zhenglin Huang, Jinwei Hu, Xiangtai Li, Yiwei He, Xingyu Zhao, Bei Peng, Baoyuan Wu, Xiaowei Huang, Guangliang Cheng
Abstract: The rapid advancement of generative models in creating highly realistic images poses substantial risks for misinformation dissemination. For instance, a synthetic image, when shared on social media, can mislead extensive audiences and erode trust in digital content, resulting in severe repercussions. Despite some progress, academia has not yet created a large and diversified deepfake detection dataset for social media, nor has it devised an effective solution to address this issue. In this paper, we introduce the Social media Image Detection dataSet (SID-Set), which offers three key advantages: (1) extensive volume, featuring 300K AI-generated/tampered and authentic images with comprehensive annotations, (2) broad diversity, encompassing fully synthetic and tampered images across various classes, and (3) elevated realism, with images that are predominantly indistinguishable from genuine ones through mere visual inspection. Furthermore, leveraging the exceptional capabilities of large multimodal models, we propose a new image deepfake detection, localization, and explanation framework, named SIDA (Social media Image Detection, localization, and explanation Assistant). SIDA not only discerns the authenticity of images, but also delineates tampered regions through mask prediction and provides textual explanations of the model's judgment criteria. Compared with state-of-the-art deepfake detection models on SID-Set and other benchmarks, extensive experiments demonstrate that SIDA achieves superior performance among diversified settings. The code, model, and dataset will be released.
Authors: Ziwei Huang, Wanggui He, Quanyu Long, Yandi Wang, Haoyuan Li, Zhelun Yu, Fangxun Shu, Long Chen, Hao Jiang, Leilei Gan
Abstract: Evaluating the quality of synthesized images remains a significant challenge in the development of text-to-image (T2I) generation. Most existing studies in this area primarily focus on evaluating text-image alignment, image quality, and object composition capabilities, with comparatively fewer studies addressing the evaluation of the factuality of T2I models, particularly when the concepts involved are knowledge-intensive. To mitigate this gap, we present T2I-FactualBench in this work - the largest benchmark to date in terms of the number of concepts and prompts specifically designed to evaluate the factuality of knowledge-intensive concept generation. T2I-FactualBench consists of a three-tiered knowledge-intensive text-to-image generation framework, ranging from the basic memorization of individual knowledge concepts to the more complex composition of multiple knowledge concepts. We further introduce a multi-round visual question answering (VQA) based evaluation framework to assess the factuality of three-tiered knowledge-intensive text-to-image generation tasks. Experiments on T2I-FactualBench indicate that current state-of-the-art (SOTA) T2I models still leave significant room for improvement.
Authors: Trong-Tung Nguyen, Quang Nguyen, Khoi Nguyen, Anh Tran, Cuong Pham
Abstract: Recent advances in text-guided image editing enable users to perform image edits through simple text inputs, leveraging the extensive priors of multi-step diffusion-based text-to-image models. However, these methods often fall short of the speed demands required for real-world and on-device applications due to the costly multi-step inversion and sampling process involved. In response to this, we introduce SwiftEdit, a simple yet highly efficient editing tool that achieve instant text-guided image editing (in 0.23s). The advancement of SwiftEdit lies in its two novel contributions: a one-step inversion framework that enables one-step image reconstruction via inversion and a mask-guided editing technique with our proposed attention rescaling mechanism to perform localized image editing. Extensive experiments are provided to demonstrate the effectiveness and efficiency of SwiftEdit. In particular, SwiftEdit enables instant text-guided image editing, which is extremely faster than previous multi-step methods (at least 50 times faster) while maintain a competitive performance in editing results. Our project page is at: https://swift-edit.github.io/
Authors: Yizhou Wang, Kuan-Chuan Peng, Yun Fu
Abstract: 3D anomaly detection and localization is of great significance for industrial inspection. Prior 3D anomaly detection and localization methods focus on the setting that the testing data share the same category as the training data which is normal. However, in real-world applications, the normal training data for the target 3D objects can be unavailable due to issues like data privacy or export control regulation. To tackle these challenges, we identify a new task -- zero-shot 3D anomaly detection and localization, where the training and testing classes do not overlap. To this end, we design 3DzAL, a novel patch-level contrastive learning framework based on pseudo anomalies generated using the inductive bias from task-irrelevant 3D xyz data to learn more representative feature representations. Furthermore, we train a normalcy classifier network to classify the normal patches and pseudo anomalies and utilize the classification result jointly with feature distance to design anomaly scores. Instead of directly using the patch point clouds, we introduce adversarial perturbations to the input patch xyz data before feeding into the 3D normalcy classifier for the classification-based anomaly score. We show that 3DzAL outperforms the state-of-the-art anomaly detection and localization performance.
Authors: S\'ebastien Pi\'erard, Ana\"is Halin, Anthony Cioppa, Adrien Deli\`ege, Marc Van Droogenbroeck
Abstract: In the computer vision and machine learning communities, as well as in many other research domains, rigorous evaluation of any new method, including classifiers, is essential. One key component of the evaluation process is the ability to compare and rank methods. However, ranking classifiers and accurately comparing their performances, especially when taking application-specific preferences into account, remains challenging. For instance, commonly used evaluation tools like Receiver Operating Characteristic (ROC) and Precision/Recall (PR) spaces display performances based on two scores. Hence, they are inherently limited in their ability to compare classifiers across a broader range of scores and lack the capability to establish a clear ranking among classifiers. In this paper, we present a novel versatile tool, named the Tile, that organizes an infinity of ranking scores in a single 2D map for two-class classifiers, including common evaluation scores such as the accuracy, the true positive rate, the positive predictive value, Jaccard's coefficient, and all F-beta scores. Furthermore, we study the properties of the underlying ranking scores, such as the influence of the priors or the correspondences with the ROC space, and depict how to characterize any other score by comparing them to the Tile. Overall, we demonstrate that the Tile is a powerful tool that effectively captures all the rankings in a single visualization and allows interpreting them.
Authors: Bo Ji, Angela Yao
Abstract: Standard single-image super-resolution (SR) upsamples and restores entire images. Yet several real-world applications require higher resolutions only in specific regions, such as license plates or faces, making the super-resolution of the entire image, along with the associated memory and computational cost, unnecessary. We propose a novel task, called LocalSR, to restore only local regions of the low-resolution image. For this problem setting, we propose a context-based local super-resolution (CLSR) to super-resolve only specified regions of interest (ROI) while leveraging the entire image as context. Our method uses three parallel processing modules: a base module for super-resolving the ROI, a global context module for gathering helpful features from across the image, and a proximity integration module for concentrating on areas surrounding the ROI, progressively propagating features from distant pixels to the target region. Experimental results indicate that our approach, with its reduced low complexity, outperforms variants that focus exclusively on the ROI.
Authors: Bo Tong, Bokai Lai, Yiyi Zhou, Gen Luo, Yunhang Shen, Ke Li, Xiaoshuai Sun, Rongrong Ji
Abstract: Despite a big leap forward in capability, multimodal large language models (MLLMs) tend to behave like a sloth in practical use, i.e., slow response and large latency. Recent efforts are devoted to building tiny MLLMs for better efficiency, but the plethora of visual tokens still used limit their actual speedup. In this paper, we propose a powerful and fast tiny MLLM called FlashSloth. Different from previous efforts, FlashSloth focuses on improving the descriptive power of visual tokens in the process of compressing their redundant semantics. In particular, FlashSloth introduces embedded visual compression designs to capture both visually salient and instruction-related image information, so as to achieving superior multimodal performance with fewer visual tokens. Extensive experiments are conducted to validate the proposed FlashSloth, and a bunch of tiny but strong MLLMs are also comprehensively compared, e.g., InternVL2, MiniCPM-V2 and Qwen2-VL. The experimental results show that compared with these advanced tiny MLLMs, our FlashSloth can greatly reduce the number of visual tokens, training memory and computation complexity while retaining high performance on various VL tasks.
Authors: Junfeng Wu, Yi Jiang, Chuofan Ma, Yuliang Liu, Hengshuang Zhao, Zehuan Yuan, Song Bai, Xiang Bai
Abstract: We present Liquid, an auto-regressive generation paradigm that seamlessly integrates visual comprehension and generation by tokenizing images into discrete codes and learning these code embeddings alongside text tokens within a shared feature space for both vision and language. Unlike previous multimodal large language model (MLLM), Liquid achieves this integration using a single large language model (LLM), eliminating the need for external pretrained visual embeddings such as CLIP. For the first time, Liquid uncovers a scaling law that performance drop unavoidably brought by the unified training of visual and language tasks diminishes as the model size increases. Furthermore, the unified token space enables visual generation and comprehension tasks to mutually enhance each other, effectively removing the typical interference seen in earlier models. We show that existing LLMs can serve as strong foundations for Liquid, saving 100x in training costs while outperforming Chameleon in multimodal capabilities and maintaining language performance comparable to mainstream LLMs like LLAMA2. Liquid also outperforms models like SD v2.1 and SD-XL (FID of 5.47 on MJHQ-30K), excelling in both vision-language and text-only tasks. This work demonstrates that LLMs such as LLAMA3.2 and GEMMA2 are powerful multimodal generators, offering a scalable solution for enhancing both vision-language understanding and generation. The code and models will be released.
Authors: Saheli Hazra, Sudip Das, Rohit Choudhary, Arindam Das, Ganesh Sistu, Ciaran Eising, Ujjwal Bhattacharya
Abstract: Applying pseudo labeling techniques has been found to be advantageous in semi-supervised 3D object detection (SSOD) in Bird's-Eye-View (BEV) for autonomous driving, particularly where labeled data is limited. In the literature, Exponential Moving Average (EMA) has been used for adjustments of the weights of teacher network by the student network. However, the same induces catastrophic forgetting in the teacher network. In this work, we address this issue by introducing a novel concept of Reflective Teacher where the student is trained by both labeled and pseudo labeled data while its knowledge is progressively passed to the teacher through a regularizer to ensure retention of previous knowledge. Additionally, we propose Geometry Aware BEV Fusion (GA-BEVFusion) for efficient alignment of multi-modal BEV features, thus reducing the disparity between the modalities - camera and LiDAR. This helps to map the precise geometric information embedded among LiDAR points reliably with the spatial priors for extraction of semantic information from camera images. Our experiments on the nuScenes and Waymo datasets demonstrate: 1) improved performance over state-of-the-art methods in both fully supervised and semi-supervised settings; 2) Reflective Teacher achieves equivalent performance with only 25% and 22% of labeled data for nuScenes and Waymo datasets respectively, in contrast to other fully supervised methods that utilize the full labeled dataset.
Authors: Zhouyingcheng Liao, Mingyuan Zhang, Wenjia Wang, Lei Yang, Taku Komura
Abstract: While motion generation has made substantial progress, its practical application remains constrained by dataset diversity and scale, limiting its ability to handle out-of-distribution scenarios. To address this, we propose a simple and effective baseline, RMD, which enhances the generalization of motion generation through retrieval-augmented techniques. Unlike previous retrieval-based methods, RMD requires no additional training and offers three key advantages: (1) the external retrieval database can be flexibly replaced; (2) body parts from the motion database can be reused, with an LLM facilitating splitting and recombination; and (3) a pre-trained motion diffusion model serves as a prior to improve the quality of motions obtained through retrieval and direct combination. Without any training, RMD achieves state-of-the-art performance, with notable advantages on out-of-distribution data.
Authors: Dayoung Gong, Suha Kwak, Minsu Cho
Abstract: Temporal action segmentation and long-term action anticipation are two popular vision tasks for the temporal analysis of actions in videos. Despite apparent relevance and potential complementarity, these two problems have been investigated as separate and distinct tasks. In this work, we tackle these two problems, action segmentation and action anticipation, jointly using a unified diffusion model dubbed ActFusion. The key idea to unification is to train the model to effectively handle both visible and invisible parts of the sequence in an integrated manner; the visible part is for temporal segmentation, and the invisible part is for future anticipation. To this end, we introduce a new anticipative masking strategy during training in which a late part of the video frames is masked as invisible, and learnable tokens replace these frames to learn to predict the invisible future. Experimental results demonstrate the bi-directional benefits between action segmentation and anticipation. ActFusion achieves the state-of-the-art performance across the standard benchmarks of 50 Salads, Breakfast, and GTEA, outperforming task-specific models in both of the two tasks with a single unified model through joint learning.
Authors: Ana\"is Halin, S\'ebastien Pi\'erard, Anthony Cioppa, Marc Van Droogenbroeck
Abstract: Properly understanding the performances of classifiers is essential in various scenarios. However, the literature often relies only on one or two standard scores to compare classifiers, which fails to capture the nuances of application-specific requirements, potentially leading to suboptimal classifier selection. Recently, a paper on the foundations of the theory of performance-based ranking introduced a tool, called the Tile, that organizes an infinity of ranking scores into a 2D map. Thanks to the Tile, it is now possible to evaluate and compare classifiers efficiently, displaying all possible application-specific preferences instead of having to rely on a pair of scores. In this paper, we provide a first hitchhiker's guide for understanding the performances of two-class classifiers by presenting four scenarios, each showcasing a different user profile: a theoretical analyst, a method designer, a benchmarker, and an application developer. Particularly, we show that we can provide different interpretative flavors that are adapted to the user's needs by mapping different values on the Tile. As an illustration, we leverage the newly introduced Tile tool and the different flavors to rank and analyze the performances of 74 state-of-the-art semantic segmentation models in two-class classification through the eyes of the four user profiles. Through these user profiles, we demonstrate that the Tile effectively captures the behavior of classifiers in a single visualization, while accommodating an infinite number of ranking scores.
Authors: Yassine Ouali, Adrian Bulat, Alexandros Xenos, Anestis Zaganidis, Ioannis Maniadis Metaxas, Georgios Tzimiropoulos, Brais Martinez
Abstract: Contrastively-trained Vision-Language Models (VLMs) like CLIP have become the de facto approach for discriminative vision-language representation learning. However, these models have limited language understanding, often exhibiting a "bag of words" behavior. At the same time, Large Vision-Language Models (LVLMs), which combine vision encoders with LLMs, have been shown capable of detailed vision-language reasoning, yet their autoregressive nature renders them less suitable for discriminative tasks. In this work, we propose to combine "the best of both worlds": a new training approach for discriminative fine-tuning of LVLMs that results in strong discriminative and compositional capabilities. Essentially, our approach converts a generative LVLM into a discriminative one, unlocking its capability for powerful image-text discrimination combined with enhanced language understanding. Our contributions include: (1) A carefully designed training/optimization framework that utilizes image-text pairs of variable length and granularity for training the model with both contrastive and next-token prediction losses. This is accompanied by ablation studies that justify the necessity of our framework's components. (2) A parameter-efficient adaptation method using a combination of soft prompting and LoRA adapters. (3) Significant improvements over state-of-the-art CLIP-like models of similar size, including standard image-text retrieval benchmarks and notable gains in compositionality.
Authors: Yuqi Wu, Wenzhao Zheng, Sicheng Zuo, Yuanhui Huang, Jie Zhou, Jiwen Lu
Abstract: 3D occupancy prediction provides a comprehensive description of the surrounding scenes and has become an essential task for 3D perception. Most existing methods focus on offline perception from one or a few views and cannot be applied to embodied agents which demands to gradually perceive the scene through progressive embodied exploration. In this paper, we formulate an embodied 3D occupancy prediction task to target this practical scenario and propose a Gaussian-based EmbodiedOcc framework to accomplish it. We initialize the global scene with uniform 3D semantic Gaussians and progressively update local regions observed by the embodied agent. For each update, we extract semantic and structural features from the observed image and efficiently incorporate them via deformable cross-attention to refine the regional Gaussians. Finally, we employ Gaussian-to-voxel splatting to obtain the global 3D occupancy from the updated 3D Gaussians. Our EmbodiedOcc assumes an unknown (i.e., uniformly distributed) environment and maintains an explicit global memory of it with 3D Gaussians. It gradually gains knowledge through local refinement of regional Gaussians, which is consistent with how humans understand new scenes through embodied exploration. We reorganize an EmbodiedOcc-ScanNet benchmark based on local annotations to facilitate the evaluation of the embodied 3D occupancy prediction task. Experiments demonstrate that our EmbodiedOcc outperforms existing local prediction methods and accomplishes the embodied occupancy prediction with high accuracy and strong expandability. Our code is available at: https://github.com/YkiWu/EmbodiedOcc.
Authors: Rong Li, Shijie Li, Lingdong Kong, Xulei Yang, Junwei Liang
Abstract: 3D Visual Grounding (3DVG) aims to locate objects in 3D scenes based on textual descriptions, which is essential for applications like augmented reality and robotics. Traditional 3DVG approaches rely on annotated 3D datasets and predefined object categories, limiting scalability and adaptability. To overcome these limitations, we introduce SeeGround, a zero-shot 3DVG framework leveraging 2D Vision-Language Models (VLMs) trained on large-scale 2D data. We propose to represent 3D scenes as a hybrid of query-aligned rendered images and spatially enriched text descriptions, bridging the gap between 3D data and 2D-VLMs input formats. We propose two modules: the Perspective Adaptation Module, which dynamically selects viewpoints for query-relevant image rendering, and the Fusion Alignment Module, which integrates 2D images with 3D spatial descriptions to enhance object localization. Extensive experiments on ScanRefer and Nr3D demonstrate that our approach outperforms existing zero-shot methods by large margins. Notably, we exceed weakly supervised methods and rival some fully supervised ones, outperforming previous SOTA by 7.7% on ScanRefer and 7.1% on Nr3D, showcasing its effectiveness.
Authors: Yuanhui Huang, Amonnut Thammatadatrakoon, Wenzhao Zheng, Yunpeng Zhang, Dalong Du, Jiwen Lu
Abstract: 3D semantic occupancy prediction is an important task for robust vision-centric autonomous driving, which predicts fine-grained geometry and semantics of the surrounding scene. Most existing methods leverage dense grid-based scene representations, overlooking the spatial sparsity of the driving scenes. Although 3D semantic Gaussian serves as an object-centric sparse alternative, most of the Gaussians still describe the empty region with low efficiency. To address this, we propose a probabilistic Gaussian superposition model which interprets each Gaussian as a probability distribution of its neighborhood being occupied and conforms to probabilistic multiplication to derive the overall geometry. Furthermore, we adopt the exact Gaussian mixture model for semantics calculation to avoid unnecessary overlapping of Gaussians. To effectively initialize Gaussians in non-empty region, we design a distribution-based initialization module which learns the pixel-aligned occupancy distribution instead of the depth of surfaces. We conduct extensive experiments on nuScenes and KITTI-360 datasets and our GaussianFormer-2 achieves state-of-the-art performance with high efficiency. Code: https://github.com/huang-yh/GaussianFormer.
Authors: Jiuhai Chen, Jianwei Yang, Haiping Wu, Dianqi Li, Jianfeng Gao, Tianyi Zhou, Bin Xiao
Abstract: We present Florence-VL, a new family of multimodal large language models (MLLMs) with enriched visual representations produced by Florence-2, a generative vision foundation model. Unlike the widely used CLIP-style vision transformer trained by contrastive learning, Florence-2 can capture different levels and aspects of visual features, which are more versatile to be adapted to diverse downstream tasks. We propose a novel feature-fusion architecture and an innovative training recipe that effectively integrates Florence-2's visual features into pretrained LLMs, such as Phi 3.5 and LLama 3. In particular, we propose "depth-breath fusion (DBFusion)" to fuse the visual features extracted from different depths and under multiple prompts. Our model training is composed of end-to-end pretraining of the whole model followed by finetuning of the projection layer and the LLM, on a carefully designed recipe of diverse open-source datasets that include high-quality image captions and instruction-tuning pairs. Our quantitative analysis and visualization of Florence-VL's visual features show its advantages over popular vision encoders on vision-language alignment, where the enriched depth and breath play important roles. Florence-VL achieves significant improvements over existing state-of-the-art MLLMs across various multi-modal and vision-centric benchmarks covering general VQA, perception, hallucination, OCR, Chart, knowledge-intensive understanding, etc. To facilitate future research, our models and the complete training recipe are open-sourced. https://github.com/JiuhaiChen/Florence-VL
Authors: Shaunak Halbe, Junjiao Tian, K J Joseph, James Seale Smith, Katherine Stevo, Vineeth N Balasubramanian, Zsolt Kira
Abstract: Vision-language models (VLMs) like CLIP have been cherished for their ability to perform zero-shot visual recognition on open-vocabulary concepts. This is achieved by selecting the object category whose textual representation bears the highest similarity with the query image. While successful in some domains, this method struggles with identifying fine-grained entities as well as generalizing to unseen concepts that are not captured by the training distribution. Recent works attempt to mitigate these challenges by integrating category descriptions at test time, albeit yielding modest improvements. We attribute these limited gains to a fundamental misalignment between image and description representations, which is rooted in the pretraining structure of CLIP. In this paper, we propose GRAIN, a new pretraining strategy aimed at aligning representations at both fine and coarse levels simultaneously. Our approach learns to jointly ground textual descriptions in image regions along with aligning overarching captions with global image representations. To drive this pre-training, we leverage frozen Multimodal Large Language Models (MLLMs) to derive large-scale synthetic annotations. We demonstrate the enhanced zero-shot performance of our model compared to current state-of-the art methods across 11 diverse image classification datasets. Additionally, we introduce Products-2023, a newly curated, manually labeled dataset featuring novel concepts, and showcase our model's ability to recognize these concepts by benchmarking on it. Significant improvements achieved by our model on other downstream tasks like retrieval further highlight the superior quality of representations learned by our approach. Code available at https://github.com/shaunak27/grain-clip .
Authors: Jian Han, Jinlai Liu, Yi Jiang, Bin Yan, Yuqi Zhang, Zehuan Yuan, Bingyue Peng, Xiaobing Liu
Abstract: We present Infinity, a Bitwise Visual AutoRegressive Modeling capable of generating high-resolution, photorealistic images following language instruction. Infinity redefines visual autoregressive model under a bitwise token prediction framework with an infinite-vocabulary tokenizer & classifier and bitwise self-correction mechanism, remarkably improving the generation capacity and details. By theoretically scaling the tokenizer vocabulary size to infinity and concurrently scaling the transformer size, our method significantly unleashes powerful scaling capabilities compared to vanilla VAR. Infinity sets a new record for autoregressive text-to-image models, outperforming top-tier diffusion models like SD3-Medium and SDXL. Notably, Infinity surpasses SD3-Medium by improving the GenEval benchmark score from 0.62 to 0.73 and the ImageReward benchmark score from 0.87 to 0.96, achieving a win rate of 66%. Without extra optimization, Infinity generates a high-quality 1024x1024 image in 0.8 seconds, making it 2.6x faster than SD3-Medium and establishing it as the fastest text-to-image model. Models and codes will be released to promote further exploration of Infinity for visual generation and unified tokenizer modeling.
Authors: Yuying Ge, Yizhuo Li, Yixiao Ge, Ying Shan
Abstract: In recent years, there has been a significant surge of interest in unifying image comprehension and generation within Large Language Models (LLMs). This growing interest has prompted us to explore extending this unification to videos. The core challenge lies in developing a versatile video tokenizer that captures both the spatial characteristics and temporal dynamics of videos to obtain representations for LLMs, and the representations can be further decoded into realistic video clips to enable video generation. In this work, we introduce Divot, a Diffusion-Powered Video Tokenizer, which leverages the diffusion process for self-supervised video representation learning. We posit that if a video diffusion model can effectively de-noise video clips by taking the features of a video tokenizer as the condition, then the tokenizer has successfully captured robust spatial and temporal information. Additionally, the video diffusion model inherently functions as a de-tokenizer, decoding videos from their representations. Building upon the Divot tokenizer, we present Divot-Vicuna through video-to-text autoregression and text-to-video generation by modeling the distributions of continuous-valued Divot features with a Gaussian Mixture Model. Experimental results demonstrate that our diffusion-based video tokenizer, when integrated with a pre-trained LLM, achieves competitive performance across various video comprehension and generation benchmarks. The instruction tuned Divot-Vicuna also excels in video storytelling, generating interleaved narratives and corresponding videos.
Authors: Shota Sasaki, Jane Wu, Ko Nishino
Abstract: This paper introduces a novel clothed human model that can be learned from multiview RGB videos, with a particular emphasis on recovering physically accurate body and cloth movements. Our method, Position Based Dynamic Gaussians (PBDyG), realizes ``movement-dependent'' cloth deformation via physical simulation, rather than merely relying on ``pose-dependent'' rigid transformations. We model the clothed human holistically but with two distinct physical entities in contact: clothing modeled as 3D Gaussians, which are attached to a skinned SMPL body that follows the movement of the person in the input videos. The articulation of the SMPL body also drives physically-based simulation of the clothes' Gaussians to transform the avatar to novel poses. In order to run position based dynamics simulation, physical properties including mass and material stiffness are estimated from the RGB videos through Dynamic 3D Gaussian Splatting. Experiments demonstrate that our method not only accurately reproduces appearance but also enables the reconstruction of avatars wearing highly deformable garments, such as skirts or coats, which have been challenging to reconstruct using existing methods.
Authors: Bin Yan, Martin Sundermeyer, David Joseph Tan, Huchuan Lu, Federico Tombari
Abstract: In this paper, we address the challenge of performing open-vocabulary video instance segmentation (OV-VIS) in real-time. We analyze the computational bottlenecks of state-of-the-art foundation models that performs OV-VIS, and propose a new method, TROY-VIS, that significantly improves processing speed while maintaining high accuracy. We introduce three key techniques: (1) Decoupled Attention Feature Enhancer to speed up information interaction between different modalities and scales; (2) Flash Embedding Memory for obtaining fast text embeddings of object categories; and, (3) Kernel Interpolation for exploiting the temporal continuity in videos. Our experiments demonstrate that TROY-VIS achieves the best trade-off between accuracy and speed on two large-scale OV-VIS benchmarks, BURST and LV-VIS, running 20x faster than GLEE-Lite (25 FPS v.s. 1.25 FPS) with comparable or even better accuracy. These results demonstrate TROY-VIS's potential for real-time applications in dynamic environments such as mobile robotics and augmented reality. Code and model will be released at https://github.com/google-research/troyvis.
Authors: Kaiyi Huang, Yukun Huang, Xuefei Ning, Zinan Lin, Yu Wang, Xihui Liu
Abstract: Text-to-video generation models have shown significant progress in the recent years. However, they still struggle with generating complex dynamic scenes based on compositional text prompts, such as attribute binding for multiple objects, temporal dynamics associated with different objects, and interactions between objects. Our key motivation is that complex tasks can be decomposed into simpler ones, each handled by a role-specialized MLLM agent. Multiple agents can collaborate together to achieve collective intelligence for complex goals. We propose GenMAC, an iterative, multi-agent framework that enables compositional text-to-video generation. The collaborative workflow includes three stages: Design, Generation, and Redesign, with an iterative loop between the Generation and Redesign stages to progressively verify and refine the generated videos. The Redesign stage is the most challenging stage that aims to verify the generated videos, suggest corrections, and redesign the text prompts, frame-wise layouts, and guidance scales for the next iteration of generation. To avoid hallucination of a single MLLM agent, we decompose this stage to four sequentially-executed MLLM-based agents: verification agent, suggestion agent, correction agent, and output structuring agent. Furthermore, to tackle diverse scenarios of compositional text-to-video generation, we design a self-routing mechanism to adaptively select the proper correction agent from a collection of correction agents each specialized for one scenario. Extensive experiments demonstrate the effectiveness of GenMAC, achieving state-of-the art performance in compositional text-to-video generation.
Authors: Emma Finn, T. Anderson Keller, Emmanouil Theodosis, Demba E. Ba
Abstract: Despite nearly a decade of literature on style transfer, there is no undisputed definition of artistic style. State-of-the-art models produce impressive results but are difficult to interpret since, without a coherent definition of style, the problem of style transfer is inherently ill-posed. Early work framed style-transfer as an optimization problem but treated style as a measure only of texture. This led to artifacts in the outputs of early models where content features from the style image sometimes bled into the output image. Conversely, more recent work with diffusion models offers compelling empirical results but provides little theoretical grounding. To address these issues, we propose an alternative definition of artistic style. We suggest that style should be thought of as a set of global symmetries that dictate the arrangement of local textures. We validate this perspective empirically by learning the symmetries of a large dataset of paintings and showing that symmetries are predictive of the artistic movement to which each painting belongs. Finally, we show that by considering both local and global features, using both Lie generators and traditional measures of texture, we can quantitatively capture the stylistic similarity between artists better than with either set of features alone. This approach not only aligns well with art historians' consensus but also offers a robust framework for distinguishing nuanced stylistic differences, allowing for a more interpretable, theoretically grounded approach to style transfer.
Authors: Yizhuo Li, Yuying Ge, Yixiao Ge, Ping Luo, Ying Shan
Abstract: Videos are inherently temporal sequences by their very nature. In this work, we explore the potential of modeling videos in a chronological and scalable manner with autoregressive (AR) language models, inspired by their success in natural language processing. We introduce DiCoDe, a novel approach that leverages Diffusion-Compressed Deep Tokens to generate videos with a language model in an autoregressive manner. Unlike existing methods that employ low-level representations with limited compression rates, DiCoDe utilizes deep tokens with a considerable compression rate (a 1000x reduction in token count). This significant compression is made possible by a tokenizer trained through leveraging the prior knowledge of video diffusion models. Deep tokens enable DiCoDe to employ vanilla AR language models for video generation, akin to translating one visual "language" into another. By treating videos as temporal sequences, DiCoDe fully harnesses the capabilities of language models for autoregressive generation. DiCoDe is scalable using readily available AR architectures, and is capable of generating videos ranging from a few seconds to one minute using only 4 A100 GPUs for training. We evaluate DiCoDe both quantitatively and qualitatively, demonstrating that it performs comparably to existing methods in terms of quality while ensuring efficient training. To showcase its scalability, we release a series of DiCoDe configurations with varying parameter sizes and observe a consistent improvement in performance as the model size increases from 100M to 3B. We believe that DiCoDe's exploration in academia represents a promising initial step toward scalable video modeling with AR language models, paving the way for the development of larger and more powerful video generation models.
Authors: Longtao Zheng, Yifan Zhang, Hanzhong Guo, Jiachun Pan, Zhenxiong Tan, Jiahao Lu, Chuanxin Tang, Bo An, Shuicheng Yan
Abstract: Recent advances in video diffusion models have unlocked new potential for realistic audio-driven talking video generation. However, achieving seamless audio-lip synchronization, maintaining long-term identity consistency, and producing natural, audio-aligned expressions in generated talking videos remain significant challenges. To address these challenges, we propose Memory-guided EMOtion-aware diffusion (MEMO), an end-to-end audio-driven portrait animation approach to generate identity-consistent and expressive talking videos. Our approach is built around two key modules: (1) a memory-guided temporal module, which enhances long-term identity consistency and motion smoothness by developing memory states to store information from a longer past context to guide temporal modeling via linear attention; and (2) an emotion-aware audio module, which replaces traditional cross attention with multi-modal attention to enhance audio-video interaction, while detecting emotions from audio to refine facial expressions via emotion adaptive layer norm. Extensive quantitative and qualitative results demonstrate that MEMO generates more realistic talking videos across diverse image and audio types, outperforming state-of-the-art methods in overall quality, audio-lip synchronization, identity consistency, and expression-emotion alignment.
Authors: Jun Zhang, Desen Meng, Ji Qi, Zhenpeng Huang, Tao Wu, Limin Wang
Abstract: Despite the remarkable performance of multimodal large language models (MLLMs) across diverse tasks, the substantial training and inference costs impede their advancement. The majority of computation stems from the overwhelming volume of vision tokens processed by the transformer decoder. In this paper, we propose to build efficient MLLMs by leveraging the Mixture-of-Depths (MoD) mechanism, where each transformer decoder layer selects essential vision tokens to process while skipping redundant ones. However, integrating MoD into MLLMs is non-trivial. To address the challenges of training and inference stability as well as limited training data, we adapt the MoD module with two novel designs: tanh-gated weight normalization (TanhNorm) and symmetric token reweighting (STRing). Moreover, we observe that vision tokens exhibit higher redundancy in deeper layer and thus design a progressive ratio decay (PRD) strategy, which gradually reduces the token retention ratio layer by layer, employing a shifted cosine schedule. This crucial design fully unleashes the potential of MoD, significantly boosting the efficiency and performance of our models. To validate the effectiveness of our approach, we conduct extensive experiments with two baseline models across 14 benchmarks. Our model, p-MoD, matches or even surpasses the performance of the baseline models, with only 55.6% TFLOPs and 53.8% KV cache storage during inference, and 77.7% GPU hours during training.
Authors: Mohammed Suhail, Carlos Esteves, Leonid Sigal, Ameesh Makadia
Abstract: Latent variable generative models have emerged as powerful tools for generative tasks including image and video synthesis. These models are enabled by pretrained autoencoders that map high resolution data into a compressed lower dimensional latent space, where the generative models can subsequently be developed while requiring fewer computational resources. Despite their effectiveness, the direct application of latent variable models to higher dimensional domains such as videos continues to pose challenges for efficient training and inference. In this paper, we propose an autoencoder that projects volumetric data onto a four-plane factorized latent space that grows sublinearly with the input size, making it ideal for higher dimensional data like videos. The design of our factorized model supports straightforward adoption in a number of conditional generation tasks with latent diffusion models (LDMs), such as class-conditional generation, frame prediction, and video interpolation. Our results show that the proposed four-plane latent space retains a rich representation needed for high-fidelity reconstructions despite the heavy compression, while simultaneously enabling LDMs to operate with significant improvements in speed and memory.
Authors: Yuto Matsubara, Ko Nishino
Abstract: We introduce a novel method for human shape and pose recovery that can fully leverage multiple static views. We target fixed-multiview people monitoring, including elderly care and safety monitoring, in which calibrated cameras can be installed at the corners of a room or an open space but whose configuration may vary depending on the environment. Our key idea is to formulate it as neural optimization. We achieve this with HeatFormer, a neural optimizer that iteratively refines the SMPL parameters given multiview images, which is fundamentally agonistic to the configuration of views. HeatFormer realizes this SMPL parameter estimation as heat map generation and alignment with a novel transformer encoder and decoder. We demonstrate the effectiveness of HeatFormer including its accuracy, robustness to occlusion, and generalizability through an extensive set of experiments. We believe HeatFormer can serve a key role in passive human behavior modeling.
Authors: Yiqing Liang, Mikhail Okunev, Mikaela Angelina Uy, Runfeng Li, Leonidas Guibas, James Tompkin, Adam W. Harley
Abstract: Gaussian splatting methods are emerging as a popular approach for converting multi-view image data into scene representations that allow view synthesis. In particular, there is interest in enabling view synthesis for dynamic scenes using only monocular input data -- an ill-posed and challenging problem. The fast pace of work in this area has produced multiple simultaneous papers that claim to work best, which cannot all be true. In this work, we organize, benchmark, and analyze many Gaussian-splatting-based methods, providing apples-to-apples comparisons that prior works have lacked. We use multiple existing datasets and a new instructive synthetic dataset designed to isolate factors that affect reconstruction quality. We systematically categorize Gaussian splatting methods into specific motion representation types and quantify how their differences impact performance. Empirically, we find that their rank order is well-defined in synthetic data, but the complexity of real-world data currently overwhelms the differences. Furthermore, the fast rendering speed of all Gaussian-based methods comes at the cost of brittleness in optimization. We summarize our experiments into a list of findings that can help to further progress in this lively problem setting. Project Webpage: https://lynl7130.github.io/MonoDyGauBench.github.io/
Authors: Justin Lazarow, David Griffiths, Gefen Kohavi, Francisco Crespo, Afshin Dehghan
Abstract: We consider indoor 3D object detection with respect to a single RGB(-D) frame acquired from a commodity handheld device. We seek to significantly advance the status quo with respect to both data and modeling. First, we establish that existing datasets have significant limitations to scale, accuracy, and diversity of objects. As a result, we introduce the Cubify-Anything 1M (CA-1M) dataset, which exhaustively labels over 400K 3D objects on over 1K highly accurate laser-scanned scenes with near-perfect registration to over 3.5K handheld, egocentric captures. Next, we establish Cubify Transformer (CuTR), a fully Transformer 3D object detection baseline which rather than operating in 3D on point or voxel-based representations, predicts 3D boxes directly from 2D features derived from RGB(-D) inputs. While this approach lacks any 3D inductive biases, we show that paired with CA-1M, CuTR outperforms point-based methods - accurately recalling over 62% of objects in 3D, and is significantly more capable at handling noise and uncertainty present in commodity LiDAR-derived depth maps while also providing promising RGB only performance without architecture changes. Furthermore, by pre-training on CA-1M, CuTR can outperform point-based methods on a more diverse variant of SUN RGB-D - supporting the notion that while inductive biases in 3D are useful at the smaller sizes of existing datasets, they fail to scale to the data-rich regime of CA-1M. Overall, this dataset and baseline model provide strong evidence that we are moving towards models which can effectively Cubify Anything.
Authors: Cheng Sun, Jaesung Choe, Charles Loop, Wei-Chiu Ma, Yu-Chiang Frank Wang
Abstract: We propose an efficient radiance field rendering algorithm that incorporates a rasterization process on sparse voxels without neural networks or 3D Gaussians. There are two key contributions coupled with the proposed system. The first is to render sparse voxels in the correct depth order along pixel rays by using dynamic Morton ordering. This avoids the well-known popping artifact found in Gaussian splatting. Second, we adaptively fit sparse voxels to different levels of detail within scenes, faithfully reproducing scene details while achieving high rendering frame rates. Our method improves the previous neural-free voxel grid representation by over 4db PSNR and more than 10x rendering FPS speedup, achieving state-of-the-art comparable novel-view synthesis results. Additionally, our neural-free sparse voxels are seamlessly compatible with grid-based 3D processing algorithms. We achieve promising mesh reconstruction accuracy by integrating TSDF-Fusion and Marching Cubes into our sparse grid system.
Authors: Yusuf Dalva, Yijun Li, Qing Liu, Nanxuan Zhao, Jianming Zhang, Zhe Lin, Pinar Yanardag
Abstract: Large-scale diffusion models have achieved remarkable success in generating high-quality images from textual descriptions, gaining popularity across various applications. However, the generation of layered content, such as transparent images with foreground and background layers, remains an under-explored area. Layered content generation is crucial for creative workflows in fields like graphic design, animation, and digital art, where layer-based approaches are fundamental for flexible editing and composition. In this paper, we propose a novel image generation pipeline based on Latent Diffusion Models (LDMs) that generates images with two layers: a foreground layer (RGBA) with transparency information and a background layer (RGB). Unlike existing methods that generate these layers sequentially, our approach introduces a harmonized generation mechanism that enables dynamic interactions between the layers for more coherent outputs. We demonstrate the effectiveness of our method through extensive qualitative and quantitative experiments, showing significant improvements in visual coherence, image quality, and layer consistency compared to baseline methods.
Authors: Chaoyang Wang, Peiye Zhuang, Tuan Duc Ngo, Willi Menapace, Aliaksandr Siarohin, Michael Vasilkovsky, Ivan Skorokhodov, Sergey Tulyakov, Peter Wonka, Hsin-Ying Lee
Abstract: We propose 4Real-Video, a novel framework for generating 4D videos, organized as a grid of video frames with both time and viewpoint axes. In this grid, each row contains frames sharing the same timestep, while each column contains frames from the same viewpoint. We propose a novel two-stream architecture. One stream performs viewpoint updates on columns, and the other stream performs temporal updates on rows. After each diffusion transformer layer, a synchronization layer exchanges information between the two token streams. We propose two implementations of the synchronization layer, using either hard or soft synchronization. This feedforward architecture improves upon previous work in three ways: higher inference speed, enhanced visual quality (measured by FVD, CLIP, and VideoScore), and improved temporal and viewpoint consistency (measured by VideoScore and Dust3R-Confidence).
Authors: Zhengqi Li, Richard Tucker, Forrester Cole, Qianqian Wang, Linyi Jin, Vickie Ye, Angjoo Kanazawa, Aleksander Holynski, Noah Snavely
Abstract: We present a system that allows for accurate, fast, and robust estimation of camera parameters and depth maps from casual monocular videos of dynamic scenes. Most conventional structure from motion and monocular SLAM techniques assume input videos that feature predominantly static scenes with large amounts of parallax. Such methods tend to produce erroneous estimates in the absence of these conditions. Recent neural network-based approaches attempt to overcome these challenges; however, such methods are either computationally expensive or brittle when run on dynamic videos with uncontrolled camera motion or unknown field of view. We demonstrate the surprising effectiveness of a deep visual SLAM framework: with careful modifications to its training and inference schemes, this system can scale to real-world videos of complex dynamic scenes with unconstrained camera paths, including videos with little camera parallax. Extensive experiments on both synthetic and real videos demonstrate that our system is significantly more accurate and robust at camera pose and depth estimation when compared with prior and concurrent work, with faster or comparable running times. See interactive results on our project page: https://mega-sam.github.io/
Authors: Ben Kaye, Tomas Jakab, Shangzhe Wu, Christian Rupprecht, Andrea Vedaldi
Abstract: The choice of data representation is a key factor in the success of deep learning in geometric tasks. For instance, DUSt3R has recently introduced the concept of viewpoint-invariant point maps, generalizing depth prediction, and showing that one can reduce all the key problems in the 3D reconstruction of static scenes to predicting such point maps. In this paper, we develop an analogous concept for a very different problem, namely, the reconstruction of the 3D shape and pose of deformable objects. To this end, we introduce the Dual Point Maps (DualPM), where a pair of point maps is extracted from the {same} image, one associating pixels to their 3D locations on the object, and the other to a canonical version of the object at rest pose. We also extend point maps to amodal reconstruction, seeing through self-occlusions to obtain the complete shape of the object. We show that 3D reconstruction and 3D pose estimation reduce to the prediction of the DualPMs. We demonstrate empirically that this representation is a good target for a deep network to predict; specifically, we consider modeling horses, showing that DualPMs can be trained purely on 3D synthetic data, consisting of a single model of a horse, while generalizing very well to real images. With this, we improve by a large margin previous methods for the 3D analysis and reconstruction of this type of objects.
Authors: Chang Liu, Viraj Shah, Aiyu Cui, Svetlana Lazebnik
Abstract: This paper introduces UnZipLoRA, a method for decomposing an image into its constituent subject and style, represented as two distinct LoRAs (Low-Rank Adaptations). Unlike existing personalization techniques that focus on either subject or style in isolation, or require separate training sets for each, UnZipLoRA disentangles these elements from a single image by training both the LoRAs simultaneously. UnZipLoRA ensures that the resulting LoRAs are compatible, i.e., they can be seamlessly combined using direct addition. UnZipLoRA enables independent manipulation and recontextualization of subject and style, including generating variations of each, applying the extracted style to new subjects, and recombining them to reconstruct the original image or create novel variations. To address the challenge of subject and style entanglement, UnZipLoRA employs a novel prompt separation technique, as well as column and block separation strategies to accurately preserve the characteristics of subject and style, and ensure compatibility between the learned LoRAs. Evaluation with human studies and quantitative metrics demonstrates UnZipLoRA's effectiveness compared to other state-of-the-art methods, including DreamBooth-LoRA, Inspiration Tree, and B-LoRA.
Authors: Senqiao Yang, Yukang Chen, Zhuotao Tian, Chengyao Wang, Jingyao Li, Bei Yu, Jiaya Jia
Abstract: Recent advancements in vision-language models have enhanced performance by increasing the length of visual tokens, making them much longer than text tokens and significantly raising computational costs. However, we observe that the visual tokens generated by popular vision encoders, such as CLIP and SigLIP, contain significant redundancy. To address this, we introduce VisionZip, a simple yet effective method that selects a set of informative tokens for input to the language model, reducing visual token redundancy and improving efficiency while maintaining model performance. The proposed VisionZip can be widely applied to image and video understanding tasks and is well-suited for multi-turn dialogues in real-world scenarios, where previous methods tend to underperform. Experimental results show that VisionZip outperforms the previous state-of-the-art method by at least 5% performance gains across nearly all settings. Moreover, our method significantly enhances model inference speed, improving the prefilling time by 8x and enabling the LLaVA-Next 13B model to infer faster than the LLaVA-Next 7B model while achieving better results. Furthermore, we analyze the causes of this redundancy and encourage the community to focus on extracting better visual features rather than merely increasing token length. Our code is available at https://github.com/dvlab-research/VisionZip .
Authors: Zhijian Liu, Ligeng Zhu, Baifeng Shi, Zhuoyang Zhang, Yuming Lou, Shang Yang, Haocheng Xi, Shiyi Cao, Yuxian Gu, Dacheng Li, Xiuyu Li, Yunhao Fang, Yukang Chen, Cheng-Yu Hsieh, De-An Huang, An-Chieh Cheng, Vishwesh Nath, Jinyi Hu, Sifei Liu, Ranjay Krishna, Daguang Xu, Xiaolong Wang, Pavlo Molchanov, Jan Kautz, Hongxu Yin, Song Han, Yao Lu
Abstract: Visual language models (VLMs) have made significant advances in accuracy in recent years. However, their efficiency has received much less attention. This paper introduces NVILA, a family of open VLMs designed to optimize both efficiency and accuracy. Building on top of VILA, we improve its model architecture by first scaling up the spatial and temporal resolutions, and then compressing visual tokens. This "scale-then-compress" approach enables NVILA to efficiently process high-resolution images and long videos. We also conduct a systematic investigation to enhance the efficiency of NVILA throughout its entire lifecycle, from training and fine-tuning to deployment. NVILA matches or surpasses the accuracy of many leading open and proprietary VLMs across a wide range of image and video benchmarks. At the same time, it reduces training costs by 4.5X, fine-tuning memory usage by 3.4X, pre-filling latency by 1.6-2.2X, and decoding latency by 1.2-2.8X. We will soon make our code and models available to facilitate reproducibility.
Authors: Sharath Girish, Tianye Li, Amrita Mazumdar, Abhinav Shrivastava, David Luebke, Shalini De Mello
Abstract: Online free-viewpoint video (FVV) streaming is a challenging problem, which is relatively under-explored. It requires incremental on-the-fly updates to a volumetric representation, fast training and rendering to satisfy real-time constraints and a small memory footprint for efficient transmission. If achieved, it can enhance user experience by enabling novel applications, e.g., 3D video conferencing and live volumetric video broadcast, among others. In this work, we propose a novel framework for QUantized and Efficient ENcoding (QUEEN) for streaming FVV using 3D Gaussian Splatting (3D-GS). QUEEN directly learns Gaussian attribute residuals between consecutive frames at each time-step without imposing any structural constraints on them, allowing for high quality reconstruction and generalizability. To efficiently store the residuals, we further propose a quantization-sparsity framework, which contains a learned latent-decoder for effectively quantizing attribute residuals other than Gaussian positions and a learned gating module to sparsify position residuals. We propose to use the Gaussian viewspace gradient difference vector as a signal to separate the static and dynamic content of the scene. It acts as a guide for effective sparsity learning and speeds up training. On diverse FVV benchmarks, QUEEN outperforms the state-of-the-art online FVV methods on all metrics. Notably, for several highly dynamic scenes, it reduces the model size to just 0.7 MB per frame while training in under 5 sec and rendering at 350 FPS. Project website is at https://research.nvidia.com/labs/amri/projects/queen
Authors: Hanzhe Hu, Tianwei Yin, Fujun Luan, Yiwei Hu, Hao Tan, Zexiang Xu, Sai Bi, Shubham Tulsiani, Kai Zhang
Abstract: We present Turbo3D, an ultra-fast text-to-3D system capable of generating high-quality Gaussian splatting assets in under one second. Turbo3D employs a rapid 4-step, 4-view diffusion generator and an efficient feed-forward Gaussian reconstructor, both operating in latent space. The 4-step, 4-view generator is a student model distilled through a novel Dual-Teacher approach, which encourages the student to learn view consistency from a multi-view teacher and photo-realism from a single-view teacher. By shifting the Gaussian reconstructor's inputs from pixel space to latent space, we eliminate the extra image decoding time and halve the transformer sequence length for maximum efficiency. Our method demonstrates superior 3D generation results compared to previous baselines, while operating in a fraction of their runtime.
Authors: Vinayak Gupta, Yunze Man, Yu-Xiong Wang
Abstract: Recent advances in diffusion models have revolutionized 2D and 3D content creation, yet generating photorealistic dynamic 4D scenes remains a significant challenge. Existing dynamic 4D generation methods typically rely on distilling knowledge from pre-trained 3D generative models, often fine-tuned on synthetic object datasets. Consequently, the resulting scenes tend to be object-centric and lack photorealism. While text-to-video models can generate more realistic scenes with motion, they often struggle with spatial understanding and provide limited control over camera viewpoints during rendering. To address these limitations, we present PaintScene4D, a novel text-to-4D scene generation framework that departs from conventional multi-view generative models in favor of a streamlined architecture that harnesses video generative models trained on diverse real-world datasets. Our method first generates a reference video using a video generation model, and then employs a strategic camera array selection for rendering. We apply a progressive warping and inpainting technique to ensure both spatial and temporal consistency across multiple viewpoints. Finally, we optimize multi-view images using a dynamic renderer, enabling flexible camera control based on user preferences. Adopting a training-free architecture, our PaintScene4D efficiently produces realistic 4D scenes that can be viewed from arbitrary trajectories. The code will be made publicly available. Our project page is at https://paintscene4d.github.io/
Authors: Luca Bartolomei, Fabio Tosi, Matteo Poggi, Stefano Mattoccia
Abstract: We introduce Stereo Anywhere, a novel stereo-matching framework that combines geometric constraints with robust priors from monocular depth Vision Foundation Models (VFMs). By elegantly coupling these complementary worlds through a dual-branch architecture, we seamlessly integrate stereo matching with learned contextual cues. Following this design, our framework introduces novel cost volume fusion mechanisms that effectively handle critical challenges such as textureless regions, occlusions, and non-Lambertian surfaces. Through our novel optical illusion dataset, MonoTrap, and extensive evaluation across multiple benchmarks, we demonstrate that our synthetic-only trained model achieves state-of-the-art results in zero-shot generalization, significantly outperforming existing solutions while showing remarkable robustness to challenging cases such as mirrors and transparencies.
Authors: Harsh Kumar
Abstract: Many methods have been proposed to find vector representation for words, but most rely on capturing context from the text to find semantic relationships between these vectors. We propose a novel method of using dictionary meanings and image depictions to find word vectors independent of any context. We use auto-encoder on the word images to find meaningful representations and use them to calculate the word vectors. We finally evaluate our method on word similarity, concept categorization and outlier detection tasks. Our method performs comparably to context-based methods while taking much less training time.
Authors: Caiwen Jiang, Mianxin Liu, Kaicong Sun, Dinggang Shen
Abstract: As a sensitive functional imaging technique, positron emission tomography (PET) plays a critical role in early disease diagnosis. However, obtaining a high-quality PET image requires injecting a sufficient dose (standard dose) of radionuclides into the body, which inevitably poses radiation hazards to patients. To mitigate radiation hazards, the reconstruction of standard-dose PET (SPET) from low-dose PET (LPET) is desired. According to imaging theory, PET reconstruction process involves multiple domains (e.g., projection domain and image domain), and a significant portion of the difference between SPET and LPET arises from variations in the noise levels introduced during the sampling of raw data as sinograms. In light of these two facts, we propose an end-to-end TriPle-domain LPET EnhancemenT (TriPLET) framework, by leveraging the advantages of a hybrid denoising-and-reconstruction process and a triple-domain representation (i.e., sinograms, frequency spectrum maps, and images) to reconstruct SPET images from LPET sinograms. Specifically, TriPLET consists of three sequentially coupled components including 1) a Transformer-assisted denoising network that denoises the inputted LPET sinograms in the projection domain, 2) a discrete-wavelet-transform-based reconstruction network that further reconstructs SPET from LPET in the wavelet domain, and 3) a pair-based adversarial network that evaluates the reconstructed SPET images in the image domain. Extensive experiments on the real PET dataset demonstrate that our proposed TriPLET can reconstruct SPET images with the highest similarity and signal-to-noise ratio to real data, compared with state-of-the-art methods.
Authors: Youssof Nawar, Nouran Soliman, Moustafa Wassel, Mohamed ElHabebe, Noha Adly, Marwan Torki, Ahmed Elmassry, Islam Ahmed
Abstract: Glaucoma is a progressive optic neuropathy characterized by structural damage to the optic nerve head and functional changes in the visual field. Detecting glaucoma early is crucial to preventing loss of eyesight. However, medical datasets often suffer from class imbalances, making detection more difficult for deep-learning algorithms. We use a generative-based framework to enhance glaucoma diagnosis, specifically addressing class imbalance through synthetic data generation. In addition, we collected the largest national dataset for glaucoma detection to support our study. The imbalance between normal and glaucomatous cases leads to performance degradation of classifier models. By combining our proposed framework leveraging diffusion models with a pretraining approach, we created a more robust classifier training process. This training process results in a better-performing classifier. The proposed approach shows promising results in improving the harmonic mean (sensitivity and specificity) and AUC for the roc for the glaucoma classifier. We report an improvement in the harmonic mean metric from 89.09% to 92.59% on the test set of our national dataset. We examine our method against other methods to overcome imbalance through extensive experiments. We report similar improvements on the AIROGS dataset. This study highlights that diffusion-based generation can be of great importance in tackling class imbalances in medical datasets to improve diagnostic performance.
Authors: Louis Airale, Adrien Pajot, Juliette Linossier
Abstract: The persisting threats on migratory bird populations highlights the urgent need for effective monitoring techniques that could assist in their conservation. Among these, passive acoustic monitoring is an essential tool, particularly for nocturnal migratory species that are difficult to track otherwise. This work presents the Nocturnal Bird Migration (NBM) dataset, a collection of 13,359 annotated vocalizations from 117 species of the Western Palearctic. The dataset includes precise time and frequency annotations, gathered by dozens of bird enthusiasts across France, enabling novel downstream acoustic analysis. In particular, we demonstrate that a two-stage object detection model, tailored for the processing of audio data, can be trained on our dataset to retrieve localized bounding box coordinates around each signal of interest in a spectrogram. This object detection approach, which is largely overlooked in the bird sound recognition literature, allows important applications by potentially differentiating individual birds within audio windows. Further, we show that the accuracy of our recognition model on the 45 main species of the dataset competes with state-of-the-art systems trained on much larger datasets. This highlights the interest of fostering similar open-science initiatives to acquire costly but valuable fine-grained annotations of audio files. All data and code are made openly available.
Authors: Jon Guti\'errez-Zaballa, Koldo Basterretxea, Javier Echanobe
Abstract: Machine learning-based embedded systems employed in safety-critical applications such as aerospace and autonomous driving need to be robust against perturbations produced by soft errors. Soft errors are an increasing concern in modern digital processors since smaller transistor geometries and lower voltages give electronic devices a higher sensitivity to background radiation. The resilience of deep neural network (DNN) models to perturbations in their parameters is determined, to a large extent, by the structure of the model itself, and also by the selected numerical representation and used arithmetic precision. When compression techniques such as model pruning and model quantization are applied to reduce memory footprint and computational complexity for deployment, both model structure and numerical representation are modified and thus, soft error robustness also changes. In this sense, although the choice of activation functions (AFs) in DNN models is frequently ignored, it conditions not only their accuracy and trainability, but also compressibility rates and numerical robustness. This paper investigates the suitability of using bounded AFs to improve model robustness against DNN parameter perturbations, assessing at the same time the impact of this choice on deployment in terms of model accuracy, compressibility, and computational burden. In particular, we analyze encoder-decoder fully convolutional models aimed at performing semantic segmentation tasks on hyperspectral images for scene understanding in autonomous driving. Deployment characterization is performed experimentally on an AMD-Xilinx's KV260 SoM.
Authors: Cheng-An Hsieh, Jing Zhang, Ava Yan
Abstract: In the game development process, creating character animations is a vital step that involves several stages. Typically for 2D games, illustrators begin by designing the main character image, which serves as the foundation for all subsequent animations. To create a smooth motion sequence, these subsequent animations involve drawing the character in different poses and actions, such as running, jumping, or attacking. This process requires significant manual effort from illustrators, as they must meticulously ensure consistency in design, proportions, and style across multiple motion frames. Each frame is drawn individually, making this a time-consuming and labor-intensive task. Generative models, such as diffusion models, have the potential to revolutionize this process by automating the creation of sprite sheets. Diffusion models, known for their ability to generate diverse images, can be adapted to create character animations. By leveraging the capabilities of diffusion models, we can significantly reduce the manual workload for illustrators, accelerate the animation creation process, and open up new creative possibilities in game development.
Authors: Abul Ehtesham, Saket Kumar, Aditi Singh, Tala Talaei Khoei
Abstract: Generative AI is reshaping the media landscape, enabling unprecedented capabilities in video creation, personalization, and scalability. This paper presents a comprehensive SWOT analysis of Metas Movie Gen, a cutting-edge generative AI foundation model designed to produce 1080p HD videos with synchronized audio from simple text prompts. We explore its strengths, including high-resolution video generation, precise editing, and seamless audio integration, which make it a transformative tool across industries such as filmmaking, advertising, and education. However, the analysis also addresses limitations, such as constraints on video length and potential biases in generated content, which pose challenges for broader adoption. In addition, we examine the evolving regulatory and ethical considerations surrounding generative AI, focusing on issues like content authenticity, cultural representation, and responsible use. Through comparative insights with leading models like DALL-E and Google Imagen, this paper highlights Movie Gens unique features, such as video personalization and multimodal synthesis, while identifying opportunities for innovation and areas requiring further research. Our findings provide actionable insights for stakeholders, emphasizing both the opportunities and challenges of deploying generative AI in media production. This work aims to guide future advancements in generative AI, ensuring scalability, quality, and ethical integrity in this rapidly evolving field.
Authors: Omar Elezabi, Marcos V. Conde, Zongwei Wu, Radu Timofte
Abstract: Professional photo editing remains challenging, requiring extensive knowledge of imaging pipelines and significant expertise. With the ubiquity of smartphone photography, there is an increasing demand for accessible yet sophisticated image editing solutions. While recent deep learning approaches, particularly style transfer methods, have attempted to automate this process, they often struggle with output fidelity, editing control, and complex retouching capabilities. We propose a novel retouch transfer approach that learns from professional edits through before-after image pairs, enabling precise replication of complex editing operations. To facilitate this research direction, we introduce a comprehensive Photo Retouching Dataset comprising 100,000 high-quality images edited using over 170 professional Adobe Lightroom presets. We develop a context-aware Implicit Neural Representation that learns to apply edits adaptively based on image content and context, requiring no pretraining and capable of learning from a single example. Our method extracts implicit transformations from reference edits and adaptively applies them to new images. Through extensive evaluation, we demonstrate that our approach not only surpasses existing methods in photo retouching but also enhances performance in related image reconstruction tasks like Gamut Mapping and Raw Reconstruction. By bridging the gap between professional editing capabilities and automated solutions, our work presents a significant step toward making sophisticated photo editing more accessible while maintaining high-fidelity results. Check the $\href{https://omaralezaby.github.io/inretouch}{Project\ Page}$ for more Results and information about Code and Dataset availability.
Authors: Hyesu Jang, Wooseong Yang, Hanguen Kim, Dongje Lee, Yongjin Kim, Jinbum Park, Minsoo Jeon, Jaeseong Koh, Yejin Kang, Minwoo Jung, Sangwoo Jung, Ayoung Kim
Abstract: Maritime environmental sensing requires overcoming challenges from complex conditions such as harsh weather, platform perturbations, large dynamic objects, and the requirement for long detection ranges. While cameras and LiDAR are commonly used in ground vehicle navigation, their applicability in maritime settings is limited by range constraints and hardware maintenance issues. Radar sensors, however, offer robust long-range detection capabilities and resilience to physical contamination from weather and saline conditions, making it a powerful sensor for maritime navigation. Among various radar types, X-band radar (e.g., marine radar) is widely employed for maritime vessel navigation, providing effective long-range detection essential for situational awareness and collision avoidance. Nevertheless, it exhibits limitations during berthing operations where close-range object detection is critical. To address this shortcoming, we incorporate W-band radar (e.g., Navtech imaging radar), which excels in detecting nearby objects with a higher update rate. We present a comprehensive maritime sensor dataset featuring multi-range detection capabilities. This dataset integrates short-range LiDAR data, medium-range W-band radar data, and long-range X-band radar data into a unified framework. Additionally, it includes object labels for oceanic object detection usage, derived from radar and stereo camera images. The dataset comprises seven sequences collected from diverse regions with varying levels of estimation difficulty, ranging from easy to challenging, and includes common locations suitable for global localization tasks. This dataset serves as a valuable resource for advancing research in place recognition, odometry estimation, SLAM, object detection, and dynamic object elimination within maritime environments. Dataset can be found in following link: https://sites.google.com/view/rpmmoana
Authors: Zhu Han, Jin Yang, Lianru Gao, Zhiqiang Zeng, Bing Zhang, Jocelyn Chanussot
Abstract: Deep learning (DL) has been widely applied into hyperspectral image (HSI) classification owing to its promising feature learning and representation capabilities. However, limited by the spatial resolution of sensors, existing DL-based classification approaches mainly focus on pixel-level spectral and spatial information extraction through complex network architecture design, while ignoring the existence of mixed pixels in actual scenarios. To tackle this difficulty, we propose a novel dual-branch subpixel-guided network for HSI classification, called DSNet, which automatically integrates subpixel information and convolutional class features by introducing a deep autoencoder unmixing architecture to enhance classification performance. DSNet is capable of fully considering physically nonlinear properties within subpixels and adaptively generating diagnostic abundances in an unsupervised manner to achieve more reliable decision boundaries for class label distributions. The subpixel fusion module is designed to ensure high-quality information fusion across pixel and subpixel features, further promoting stable joint classification. Experimental results on three benchmark datasets demonstrate the effectiveness and superiority of DSNet compared with state-of-the-art DL-based HSI classification approaches. The codes will be available at https://github.com/hanzhu97702/DSNet, contributing to the remote sensing community.
Authors: Alireza Maleki, Mahsa Lavaei, Mohsen Bagheritabar, Salar Beigzad, Zahra Abadi
Abstract: Deep learning techniques have proven highly effective in image classification, but their deployment in resourceconstrained environments remains challenging due to high computational demands. Furthermore, their interpretability is of high importance which demands even more available resources. In this work, we introduce an approach that combines saliency-guided training with quantization techniques to create an interpretable and resource-efficient model without compromising accuracy. We utilize Parameterized Clipping Activation (PACT) to perform quantization-aware training, specifically targeting activations and weights to optimize precision while minimizing resource usage. Concurrently, saliency-guided training is employed to enhance interpretability by iteratively masking features with low gradient values, leading to more focused and meaningful saliency maps. This training procedure helps in mitigating noisy gradients and yields models that provide clearer, more interpretable insights into their decision-making processes. To evaluate the impact of our approach, we conduct experiments using famous Convolutional Neural Networks (CNN) architecture on the MNIST and CIFAR-10 benchmark datasets as two popular datasets. We compare the saliency maps generated by standard and quantized models to assess the influence of quantization on both interpretability and classification accuracy. Our results demonstrate that the combined use of saliency-guided training and PACT-based quantization not only maintains classification performance but also produces models that are significantly more efficient and interpretable, making them suitable for deployment in resource-limited settings.
Authors: Sunyoung Jung, Yoonseok Choi, Mohammed A. Al-masni, Minyoung Jung, Dong-Hyun Kim
Abstract: Motion artifacts caused by prolonged acquisition time are a significant challenge in Magnetic Resonance Imaging (MRI), hindering accurate tissue segmentation. These artifacts appear as blurred images that mimic tissue-like appearances, making segmentation difficult. This study proposes a novel deep learning framework that demonstrates superior performance in both motion correction and robust brain tissue segmentation in the presence of artifacts. The core concept lies in a complementary process: a disentanglement learning network progressively removes artifacts, leading to cleaner images and consequently, more accurate segmentation by a jointly trained motion estimation and segmentation network. This network generates three outputs: a motioncorrected image, a motion deformation map that identifies artifact-affected regions, and a brain tissue segmentation mask. This deformation serves as a guidance mechanism for the disentanglement process, aiding the model in recovering lost information or removing artificial structures introduced by the artifacts. Extensive in-vivo experiments on pediatric motion data demonstrate that our proposed framework outperforms state-of-the-art methods in segmenting motion-corrupted MRI scans.
Authors: Genki Osada, Makoto Shing, Takashi Nishide
Abstract: The training of score-based diffusion models (SDMs) is based on score matching. The challenge of score matching is that it includes a computationally expensive Jacobian trace. While several methods have been proposed to avoid this computation, each has drawbacks, such as instability during training and approximating the learning as learning a denoising vector field rather than a true score. We propose a novel score matching variant, local curvature smoothing with Stein's identity (LCSS). The LCSS bypasses the Jacobian trace by applying Stein's identity, enabling regularization effectiveness and efficient computation. We show that LCSS surpasses existing methods in sample generation performance and matches the performance of denoising score matching, widely adopted by most SDMs, in evaluations such as FID, Inception score, and bits per dimension. Furthermore, we show that LCSS enables realistic image generation even at a high resolution of $1024 \times 1024$.
Authors: Yongjie Xu, Guangke Chen, Fu Song, Yuqi Chen
Abstract: Backdoor attacks embed hidden associations between triggers and targets in deep neural networks (DNNs), causing them to predict the target when a trigger is present while maintaining normal behavior otherwise. Physical backdoor attacks, which use physical objects as triggers, are feasible but lack remote control, temporal stealthiness, flexibility, and mobility. To overcome these limitations, in this work, we propose a new type of backdoor triggers utilizing lasers that feature long-distance transmission and instant-imaging properties. Based on the laser-based backdoor triggers, we present a physical backdoor attack, called LaserGuider, which possesses remote control ability and achieves high temporal stealthiness, flexibility, and mobility. We also introduce a systematic approach to optimize laser parameters for improving attack effectiveness. Our evaluation on traffic sign recognition DNNs, critical in autonomous vehicles, demonstrates that LaserGuider with three different laser-based triggers achieves over 90% attack success rate with negligible impact on normal inputs. Additionally, we release LaserMark, the first dataset of real world traffic signs stamped with physical laser spots, to support further research in backdoor attacks and defenses.
Authors: Zhifan Jiang, Daniel Capell\'an-Mart\'in, Abhijeet Parida, Austin Tapp, Xinyang Liu, Mar\'ia J. Ledesma-Carbayo, Syed Muhammad Anwar, Marius George Linguraru
Abstract: Accurate and automatic segmentation of brain tumors in multi-parametric magnetic resonance imaging (mpMRI) is essential for quantitative measurements, which play an increasingly important role in clinical diagnosis and prognosis. The International Brain Tumor Segmentation (BraTS) Challenge 2024 offers a unique benchmarking opportunity, including various types of brain tumors in both adult and pediatric populations, such as pediatric brain tumors (PED), meningiomas (MEN-RT) and brain metastases (MET), among others. Compared to previous editions, BraTS 2024 has implemented changes to substantially increase clinical relevance, such as refined tumor regions for evaluation. We propose a deep learning-based ensemble approach that integrates state-of-the-art segmentation models. Additionally, we introduce innovative, adaptive pre- and post-processing techniques that employ MRI-based radiomic analyses to differentiate tumor subtypes. Given the heterogeneous nature of the tumors present in the BraTS datasets, this approach enhances the precision and generalizability of segmentation models. On the final testing sets, our method achieved mean lesion-wise Dice similarity coefficients of 0.926, 0.801, and 0.688 for the whole tumor in PED, MEN-RT, and MET, respectively. These results demonstrate the effectiveness of our approach in improving segmentation performance and generalizability for various brain tumor types.
Authors: Abhijeet Parida, Daniel Capell\'an-Mart\'in, Zhifan Jiang, Austin Tapp, Xinyang Liu, Syed Muhammad Anwar, Mar\'ia J. Ledesma-Carbayo, Marius George Linguraru
Abstract: Gliomas, a kind of brain tumor characterized by high mortality, present substantial diagnostic challenges in low- and middle-income countries, particularly in Sub-Saharan Africa. This paper introduces a novel approach to glioma segmentation using transfer learning to address challenges in resource-limited regions with minimal and low-quality MRI data. We leverage pre-trained deep learning models, nnU-Net and MedNeXt, and apply a stratified fine-tuning strategy using the BraTS2023-Adult-Glioma and BraTS-Africa datasets. Our method exploits radiomic analysis to create stratified training folds, model training on a large brain tumor dataset, and transfer learning to the Sub-Saharan context. A weighted model ensembling strategy and adaptive post-processing are employed to enhance segmentation accuracy. The evaluation of our proposed method on unseen validation cases on the BraTS-Africa 2024 task resulted in lesion-wise mean Dice scores of 0.870, 0.865, and 0.926, for enhancing tumor, tumor core, and whole tumor regions and was ranked first for the challenge. Our approach highlights the ability of integrated machine-learning techniques to bridge the gap between the medical imaging capabilities of resource-limited countries and established developed regions. By tailoring our methods to a target population's specific needs and constraints, we aim to enhance diagnostic capabilities in isolated environments. Our findings underscore the importance of approaches like local data integration and stratification refinement to address healthcare disparities, ensure practical applicability, and enhance impact.
Authors: S\'ebastien Pi\'erard, Ana\"is Halin, Anthony Cioppa, Adrien Deli\`ege, Marc Van Droogenbroeck
Abstract: Ranking entities such as algorithms, devices, methods, or models based on their performances, while accounting for application-specific preferences, is a challenge. To address this challenge, we establish the foundations of a universal theory for performance-based ranking. First, we introduce a rigorous framework built on top of both the probability and order theories. Our new framework encompasses the elements necessary to (1) manipulate performances as mathematical objects, (2) express which performances are worse than or equivalent to others, (3) model tasks through a variable called satisfaction, (4) consider properties of the evaluation, (5) define scores, and (6) specify application-specific preferences through a variable called importance. On top of this framework, we propose the first axiomatic definition of performance orderings and performance-based rankings. Then, we introduce a universal parametric family of scores, called ranking scores, that can be used to establish rankings satisfying our axioms, while considering application-specific preferences. Finally, we show, in the case of two-class classification, that the family of ranking scores encompasses well-known performance scores, including the accuracy, the true positive rate (recall, sensitivity), the true negative rate (specificity), the positive predictive value (precision), and F1. However, we also show that some other scores commonly used to compare classifiers are unsuitable to derive performance orderings satisfying the axioms. Therefore, this paper provides the computer vision and machine learning communities with a rigorous framework for evaluating and ranking entities.
Authors: Elliot Chane-Sane, Constant Roux, Olivier Stasse, Nicolas Mansard
Abstract: We propose to learn legged robot locomotion skills by watching thousands of wild animal videos from the internet, such as those featured in nature documentaries. Indeed, such videos offer a rich and diverse collection of plausible motion examples, which could inform how robots should move. To achieve this, we introduce Reinforcement Learning from Wild Animal Videos (RLWAV), a method to ground these motions into physical robots. We first train a video classifier on a large-scale animal video dataset to recognize actions from RGB clips of animals in their natural habitats. We then train a multi-skill policy to control a robot in a physics simulator, using the classification score of a third-person camera capturing videos of the robot's movements as a reward for reinforcement learning. Finally, we directly transfer the learned policy to a real quadruped Solo. Remarkably, despite the extreme gap in both domain and embodiment between animals in the wild and robots, our approach enables the policy to learn diverse skills such as walking, jumping, and keeping still, without relying on reference trajectories nor skill-specific rewards.
Authors: Jie Bao, Zhixin Zhou, Wen Jung Li, Rui Luo
Abstract: Accurate medical image segmentation is essential for effective diagnosis and treatment planning but is often challenged by domain shifts caused by variations in imaging devices, acquisition conditions, and patient-specific attributes. Traditional domain generalization methods typically require inclusion of parts of the test domain within the training set, which is not always feasible in clinical settings with limited diverse data. Additionally, although diffusion models have demonstrated strong capabilities in image generation and style transfer, they often fail to preserve the critical structural information necessary for precise medical analysis. To address these issues, we propose a novel medical image segmentation method that combines diffusion models and Structure-Preserving Network for structure-aware one-shot image stylization. Our approach effectively mitigates domain shifts by transforming images from various sources into a consistent style while maintaining the location, size, and shape of lesions. This ensures robust and accurate segmentation even when the target domain is absent from the training data. Experimental evaluations on colonoscopy polyp segmentation and skin lesion segmentation datasets show that our method enhances the robustness and accuracy of segmentation models, achieving superior performance metrics compared to baseline models without style transfer. This structure-aware stylization framework offers a practical solution for improving medical image segmentation across diverse domains, facilitating more reliable clinical diagnoses.
Authors: George Webber, Yuya Mizuno, Oliver D. Howes, Alexander Hammers, Andrew P. King, Andrew J. Reader
Abstract: Score-based generative models (SGMs) have recently shown promising results for image reconstruction on simulated positron emission tomography (PET) datasets. In this work we have developed and implemented practical methodology for 3D image reconstruction with SGMs, and perform (to our knowledge) the first SGM-based reconstruction of real fully 3D PET data. We train an SGM on full-count reference brain images, and extend methodology to allow SGM-based reconstructions at very low counts (1% of original, to simulate low-dose or short-duration scanning). We then perform reconstructions for multiple independent realisations of 1% count data, allowing us to analyse the bias and variance characteristics of the method. We sample from the learned posterior distribution of the generative algorithm to calculate uncertainty images for our reconstructions. We evaluate the method's performance on real full- and low-count PET data and compare with conventional OSEM and MAP-EM baselines, showing that our SGM-based low-count reconstructions match full-dose reconstructions more closely and in a bias-variance trade-off comparison, our SGM-reconstructed images have lower variance than existing baselines. Future work will compare to supervised deep-learned methods, with other avenues for investigation including how data conditioning affects the SGM's posterior distribution and the algorithm's performance with different tracers.
Authors: George Webber, Yuya Mizuno, Oliver D. Howes, Alexander Hammers, Andrew P. King, Andrew J. Reader
Abstract: Large high-quality medical image datasets are difficult to acquire but necessary for many deep learning applications. For positron emission tomography (PET), reconstructed image quality is limited by inherent Poisson noise. We propose a novel method for synthesising diverse and realistic pseudo-PET images with improved signal-to-noise ratio. We also show how our pseudo-PET images may be exploited as a generative prior for single-subject PET image reconstruction. Firstly, we perform deep-learned deformable registration of multi-subject magnetic resonance (MR) images paired to multi-subject PET images. We then use the anatomically-learned deformation fields to transform multiple PET images to the same reference space, before averaging random subsets of the transformed multi-subject data to form a large number of varying pseudo-PET images. We observe that using MR information for registration imbues the resulting pseudo-PET images with improved anatomical detail compared to the originals. We consider applications to PET image reconstruction, by generating pseudo-PET images in the same space as the intended single-subject reconstruction and using them as training data for a diffusion model-based reconstruction method. We show visual improvement and reduced background noise in our 2D reconstructions as compared to OSEM, MAP-EM and an existing state-of-the-art diffusion model-based approach. Our method shows the potential for utilising highly subject-specific prior information within a generative reconstruction framework. Future work may compare the benefits of our approach to explicitly MR-guided reconstruction methodologies.
Authors: George Webber, Yuya Mizuno, Oliver D. Howes, Alexander Hammers, Andrew P. King, Andrew J. Reader
Abstract: Medical image reconstruction with pre-trained score-based generative models (SGMs) has advantages over other existing state-of-the-art deep-learned reconstruction methods, including improved resilience to different scanner setups and advanced image distribution modeling. SGM-based reconstruction has recently been applied to simulated positron emission tomography (PET) datasets, showing improved contrast recovery for out-of-distribution lesions relative to the state-of-the-art. However, existing methods for SGM-based reconstruction from PET data suffer from slow reconstruction, burdensome hyperparameter tuning and slice inconsistency effects (in 3D). In this work, we propose a practical methodology for fully 3D reconstruction that accelerates reconstruction and reduces the number of critical hyperparameters by matching the likelihood of an SGM's reverse diffusion process to a current iterate of the maximum-likelihood expectation maximization algorithm. Using the example of low-count reconstruction from simulated $[^{18}$F]DPA-714 datasets, we show our methodology can match or improve on the NRMSE and SSIM of existing state-of-the-art SGM-based PET reconstruction while reducing reconstruction time and the need for hyperparameter tuning. We evaluate our methodology against state-of-the-art supervised and conventional reconstruction algorithms. Finally, we demonstrate a first-ever implementation of SGM-based reconstruction for real 3D PET data, specifically $[^{18}$F]DPA-714 data, where we integrate perpendicular pre-trained SGMs to eliminate slice inconsistency issues.
Authors: Pranab Sahoo, Ashutosh Tripathi, Sriparna Saha, Samrat Mondal
Abstract: Federated Learning (FL) marks a transformative approach to distributed model training by combining locally optimized models from various clients into a unified global model. While FL preserves data privacy by eliminating centralized storage, it encounters significant challenges such as performance degradation, slower convergence, and reduced robustness of the global model due to the heterogeneity in client data distributions. Among the various forms of data heterogeneity, label skew emerges as a particularly formidable and prevalent issue, especially in domains such as image classification. To address these challenges, we begin with comprehensive experiments to pinpoint the underlying issues in the FL training process. Based on our findings, we then introduce an innovative dual-strategy approach designed to effectively resolve these issues. First, we introduce an adaptive loss function for client-side training, meticulously crafted to preserve previously acquired knowledge while maintaining an optimal equilibrium between local optimization and global model coherence. Secondly, we develop a dynamic aggregation strategy for aggregating client models at the server. This approach adapts to each client's unique learning patterns, effectively addressing the challenges of diverse data across the network. Our comprehensive evaluation, conducted across three diverse real-world datasets, coupled with theoretical convergence guarantees, demonstrates the superior efficacy of our method compared to several established state-of-the-art approaches.
Authors: Yi Chen, Yuying Ge, Yizhuo Li, Yixiao Ge, Mingyu Ding, Ying Shan, Xihui Liu
Abstract: Recent developments in Large Language Models pre-trained on extensive corpora have shown significant success in various natural language processing tasks with minimal fine-tuning. This success offers new promise for robotics, which has long been constrained by the high cost of action-labeled data. We ask: given the abundant video data containing interaction-related knowledge available as a rich "corpus", can a similar generative pre-training approach be effectively applied to enhance robot learning? The key challenge is to identify an effective representation for autoregressive pre-training that benefits robot manipulation tasks. Inspired by the way humans learn new skills through observing dynamic environments, we propose that effective robotic learning should emphasize motion-related knowledge, which is closely tied to low-level actions and is hardware-agnostic, facilitating the transfer of learned motions to actual robot actions. To this end, we introduce Moto, which converts video content into latent Motion Token sequences by a Latent Motion Tokenizer, learning a bridging "language" of motion from videos in an unsupervised manner. We pre-train Moto-GPT through motion token autoregression, enabling it to capture diverse visual motion knowledge. After pre-training, Moto-GPT demonstrates the promising ability to produce semantically interpretable motion tokens, predict plausible motion trajectories, and assess trajectory rationality through output likelihood. To transfer learned motion priors to real robot actions, we implement a co-fine-tuning strategy that seamlessly bridges latent motion token prediction and real robot control. Extensive experiments show that the fine-tuned Moto-GPT exhibits superior robustness and efficiency on robot manipulation benchmarks, underscoring its effectiveness in transferring knowledge from video data to downstream visual manipulation tasks.
Authors: Lu Qiu, Yuying Ge, Yi Chen, Yixiao Ge, Ying Shan, Xihui Liu
Abstract: The advent of Multimodal Large Language Models, leveraging the power of Large Language Models, has recently demonstrated superior multimodal understanding and reasoning abilities, heralding a new era for artificial general intelligence. However, achieving AGI necessitates more than just comprehension and reasoning. A crucial capability required is effective planning in diverse scenarios, which involves making reasonable decisions based on complex environments to solve real-world problems. Despite its importance, the planning abilities of current MLLMs in varied scenarios remain underexplored. In this paper, we introduce EgoPlan-Bench2, a rigorous and comprehensive benchmark designed to assess the planning capabilities of MLLMs across a wide range of real-world scenarios. EgoPlan-Bench2 encompasses everyday tasks spanning 4 major domains and 24 detailed scenarios, closely aligned with human daily life. EgoPlan-Bench2 is constructed through a semi-automatic process utilizing egocentric videos, complemented by manual verification. Grounded in a first-person perspective, it mirrors the way humans approach problem-solving in everyday life. We evaluate 21 competitive MLLMs and provide an in-depth analysis of their limitations, revealing that they face significant challenges in real-world planning. To further improve the planning proficiency of current MLLMs, we propose a training-free approach using multimodal Chain-of-Thought (CoT) prompting through investigating the effectiveness of various multimodal prompts in complex planning. Our approach enhances the performance of GPT-4V by 10.24 on EgoPlan-Bench2 without additional training. Our work not only sheds light on the current limitations of MLLMs in planning, but also provides insights for future enhancements in this critical area. We have made data and code available at https://qiulu66.github.io/egoplanbench2/.
Authors: An-Chieh Cheng, Yandong Ji, Zhaojing Yang, Xueyan Zou, Jan Kautz, Erdem B{\i}y{\i}k, Hongxu Yin, Sifei Liu, Xiaolong Wang
Abstract: This paper proposes to solve the problem of Vision-and-Language Navigation with legged robots, which not only provides a flexible way for humans to command but also allows the robot to navigate through more challenging and cluttered scenes. However, it is non-trivial to translate human language instructions all the way to low-level leg joint actions. We propose NaVILA, a 2-level framework that unifies a Vision-Language-Action model (VLA) with locomotion skills. Instead of directly predicting low-level actions from VLA, NaVILA first generates mid-level actions with spatial information in the form of language, (e.g., "moving forward 75cm"), which serves as an input for a visual locomotion RL policy for execution. NaVILA substantially improves previous approaches on existing benchmarks. The same advantages are demonstrated in our newly developed benchmarks with IsaacLab, featuring more realistic scenes, low-level controls, and real-world robot experiments. We show more results at https://navila-bot.github.io/
Authors: Enshen Zhou, Qi Su, Cheng Chi, Zhizheng Zhang, Zhongyuan Wang, Tiejun Huang, Lu Sheng, He Wang
Abstract: Automatic detection and prevention of open-set failures are crucial in closed-loop robotic systems. Recent studies often struggle to simultaneously identify unexpected failures reactively after they occur and prevent foreseeable ones proactively. To this end, we propose Code-as-Monitor (CaM), a novel paradigm leveraging the vision-language model (VLM) for both open-set reactive and proactive failure detection. The core of our method is to formulate both tasks as a unified set of spatio-temporal constraint satisfaction problems and use VLM-generated code to evaluate them for real-time monitoring. To enhance the accuracy and efficiency of monitoring, we further introduce constraint elements that abstract constraint-related entities or their parts into compact geometric elements. This approach offers greater generality, simplifies tracking, and facilitates constraint-aware visual programming by leveraging these elements as visual prompts. Experiments show that CaM achieves a 28.7% higher success rate and reduces execution time by 31.8% under severe disturbances compared to baselines across three simulators and a real-world setting. Moreover, CaM can be integrated with open-loop control policies to form closed-loop systems, enabling long-horizon tasks in cluttered scenes with dynamic environments.
Authors: Xumeng Han, Xuehui Yu, Guorong Li, Jian Zhao, Gang Pan, Qixiang Ye, Jianbin Jiao, Zhenjun Han
Abstract: Unsupervised person re-identification (re-ID) remains a challenging task. While extensive research has focused on the framework design and loss function, this paper shows that sampling strategy plays an equally important role. We analyze the reasons for the performance differences between various sampling strategies under the same framework and loss function. We suggest that deteriorated over-fitting is an important factor causing poor performance, and enhancing statistical stability can rectify this problem. Inspired by that, a simple yet effective approach is proposed, termed group sampling, which gathers samples from the same class into groups. The model is thereby trained using normalized group samples, which helps alleviate the negative impact of individual samples. Group sampling updates the pipeline of pseudo-label generation by guaranteeing that samples are more efficiently classified into the correct classes. It regulates the representation learning process, enhancing statistical stability for feature representation in a progressive fashion. Extensive experiments on Market-1501, DukeMTMC-reID and MSMT17 show that group sampling achieves performance comparable to state-of-the-art methods and outperforms the current techniques under purely camera-agnostic settings. Code has been available at https://github.com/ucas-vg/GroupSampling.
Authors: Anni Tang, Tianyu He, Xu Tan, Jun Ling, Li Song
Abstract: Talking face generation aims at generating photo-realistic video portraits of a target person driven by input audio. Due to its nature of one-to-many mapping from the input audio to the output video (e.g., one speech content may have multiple feasible visual appearances), learning a deterministic mapping like previous works brings ambiguity during training, and thus causes inferior visual results. Although this one-to-many mapping could be alleviated in part by a two-stage framework (i.e., an audio-to-expression model followed by a neural-rendering model), it is still insufficient since the prediction is produced without enough information (e.g., emotions, wrinkles, etc.). In this paper, we propose MemFace to complement the missing information with an implicit memory and an explicit memory that follow the sense of the two stages respectively. More specifically, the implicit memory is employed in the audio-to-expression model to capture high-level semantics in the audio-expression shared space, while the explicit memory is employed in the neural-rendering model to help synthesize pixel-level details. Our experimental results show that our proposed MemFace surpasses all the state-of-the-art results across multiple scenarios consistently and significantly.
Authors: Gyeongrok Oh, Sungjune Kim, Heon Gu, Sang Ho Yoon, Jinkyu Kim, Sangpil Kim
Abstract: Moire patterns, created by the interference between overlapping grid patterns in the pixel space, degrade the visual quality of images and videos. Therefore, removing such patterns~(demoireing) is crucial, yet remains a challenge due to their complexities in sizes and distortions. Conventional methods mainly tackle this task by only exploiting the spatial domain of the input images, limiting their capabilities in removing large-scale moire patterns. Therefore, this work proposes FPANet, an image-video demoireing network that learns filters in both frequency and spatial domains, improving the restoration quality by removing various sizes of moire patterns. To further enhance, our model takes multiple consecutive frames, learning to extract frame-invariant content features and outputting better quality temporally consistent images. We demonstrate the effectiveness of our proposed method with a publicly available large-scale dataset, observing that ours outperforms the state-of-the-art approaches in terms of image and video quality metrics and visual experience.
Authors: Jing Liu, Donglai Wei, Yang Liu, Sipeng Zhang, Tong Yang, Victor C. M. Leung
Abstract: Text-Based Person Search (TBPS) is a crucial task that enables accurate retrieval of target individuals from large-scale galleries with only given textual caption. For cross-modal TBPS tasks, it is critical to obtain well-distributed representation in the common embedding space to reduce the inter-modal gap. Furthermore, learning detailed image-text correspondences is essential to discriminate similar targets and enable fine-grained search. To address these challenges, we present a simple yet effective method named Sew Calibration and Masked Modeling (SCMM) that calibrates cross-modal representations by learning compact and well-aligned embeddings. SCMM is distinguished by two novel losses to provide fine-grained cross-modal representations: 1) a Sew calibration loss that takes the quality of textual captions as guidance and aligns features between image and text modalities, and 2) a Masked Caption Modeling (MCM) loss that leverages a masked caption prediction task to establish detailed and generic relationships between textual and visual parts. The dual-pronged strategy refines feature alignment and enriches cross-modal correspondences, enabling the accurate distinction of similar individuals. Consequently, its streamlined dual-encoder architecture avoids complex branches and interactions and facilitates high-speed inference suitable for real-time requirements. Consequently, high-speed inference is achieved, which is essential for resource-limited applications often demanding real-time processing. Extensive experiments on three popular TBPS benchmarks demonstrate the superiority of SCMM, achieving top results with 73.81%, 64.25%, and 57.35% Rank-1 accuracy on CUHK-PEDES, ICFG-PEDES, and RSTPReID, respectively. We hope SCMM's scalable and cost-effective design will serve as a strong baseline and facilitate future research in this field.
Authors: Zinan Lin, Sivakanth Gopi, Janardhan Kulkarni, Harsha Nori, Sergey Yekhanin
Abstract: Generating differentially private (DP) synthetic data that closely resembles the original private data is a scalable way to mitigate privacy concerns in the current data-driven world. In contrast to current practices that train customized models for this task, we aim to generate DP Synthetic Data via APIs (DPSDA), where we treat foundation models as blackboxes and only utilize their inference APIs. Such API-based, training-free approaches are easier to deploy as exemplified by the recent surge in the number of API-based apps. These approaches can also leverage the power of large foundation models which are only accessible via their inference APIs. However, this comes with greater challenges due to strictly more restrictive model access and the need to protect privacy from the API provider. In this paper, we present a new framework called Private Evolution (PE) to solve this problem and show its initial promise on synthetic images. Surprisingly, PE can match or even outperform state-of-the-art (SOTA) methods without any model training. For example, on CIFAR10 (with ImageNet as the public data), we achieve FID <= 7.9 with privacy cost {\epsilon} = 0.67, significantly improving the previous SOTA from {\epsilon} = 32. We further demonstrate the promise of applying PE on large foundation models such as Stable Diffusion to tackle challenging private datasets with a small number of high-resolution images. The code and data are released at https://github.com/microsoft/DPSDA.
Authors: Guoyao Shen, Yancheng Zhu, Mengyu Li, Ryan McNaughton, Hernan Jara, Sean B. Andersson, Chad W. Farris, Stephan Anderson, Xin Zhang
Abstract: Recent advances in MRI reconstruction have achieved remarkable success with deep learning-based models. However, most methods depend on large-scale, task-specific datasets, leaving reconstruction in data-limited settings as a critical but underexplored challenge. Regularization by denoising (RED) is a general pipeline that incorporates a denoiser as a prior for image reconstruction, showing promising results in various image processing tasks, including denoising, deblurring, and super-resolution. In this work, we propose a regularization by neural style transfer (RNST) method to further leverage the priors from the neural transfer and denoising engine. RNST effectively reconstructs high-quality images from noisy, low-quality inputs across varying image styles, even with limited data. We validate RNST on clinical MRI scans, demonstrating its ability to significantly improve image quality. These findings underline the potential of RNST for MRI field-transfer reconstruction and its promise in addressing reconstruction tasks in data-constrained scenarios.
Authors: Rohit Bharadwaj, Muzammal Naseer, Salman Khan, Fahad Shahbaz Khan
Abstract: In this work, we address the challenging and emergent problem of novel object detection (NOD), focusing on the accurate detection of both known and novel object categories during inference. Traditional object detection algorithms are inherently closed-set, limiting their capability to handle NOD. We present a novel approach to transform existing closed-set detectors into open-set detectors. This transformation is achieved by leveraging the complementary strengths of pre-trained foundational models, specifically CLIP and SAM, through our cooperative mechanism. Furthermore, by integrating this mechanism with state-of-the-art open-set detectors such as GDINO, we establish new benchmarks in object detection performance. Our method achieves 17.42 mAP in novel object detection and 42.08 mAP for known objects on the challenging LVIS dataset. Adapting our approach to the COCO OVD split, we surpass the current state-of-the-art by a margin of 7.2 $ \text{AP}_{50} $ for novel classes. Our code is available at https://rohit901.github.io/coop-foundation-models/ .
Authors: Agelos Kratimenos, Jiahui Lei, Kostas Daniilidis
Abstract: Accurately and efficiently modeling dynamic scenes and motions is considered so challenging a task due to temporal dynamics and motion complexity. To address these challenges, we propose DynMF, a compact and efficient representation that decomposes a dynamic scene into a few neural trajectories. We argue that the per-point motions of a dynamic scene can be decomposed into a small set of explicit or learned trajectories. Our carefully designed neural framework consisting of a tiny set of learned basis queried only in time allows for rendering speed similar to 3D Gaussian Splatting, surpassing 120 FPS, while at the same time, requiring only double the storage compared to static scenes. Our neural representation adequately constrains the inherently underconstrained motion field of a dynamic scene leading to effective and fast optimization. This is done by biding each point to motion coefficients that enforce the per-point sharing of basis trajectories. By carefully applying a sparsity loss to the motion coefficients, we are able to disentangle the motions that comprise the scene, independently control them, and generate novel motion combinations that have never been seen before. We can reach state-of-the-art render quality within just 5 minutes of training and in less than half an hour, we can synthesize novel views of dynamic scenes with superior photorealistic quality. Our representation is interpretable, efficient, and expressive enough to offer real-time view synthesis of complex dynamic scene motions, in monocular and multi-view scenarios.
Authors: Ryan Po, Guandao Yang, Kfir Aberman, Gordon Wetzstein
Abstract: Customization techniques for text-to-image models have paved the way for a wide range of previously unattainable applications, enabling the generation of specific concepts across diverse contexts and styles. While existing methods facilitate high-fidelity customization for individual concepts or a limited, pre-defined set of them, they fall short of achieving scalability, where a single model can seamlessly render countless concepts. In this paper, we address a new problem called Modular Customization, with the goal of efficiently merging customized models that were fine-tuned independently for individual concepts. This allows the merged model to jointly synthesize concepts in one image without compromising fidelity or incurring any additional computational costs. To address this problem, we introduce Orthogonal Adaptation, a method designed to encourage the customized models, which do not have access to each other during fine-tuning, to have orthogonal residual weights. This ensures that during inference time, the customized models can be summed with minimal interference. Our proposed method is both simple and versatile, applicable to nearly all optimizable weights in the model architecture. Through an extensive set of quantitative and qualitative evaluations, our method consistently outperforms relevant baselines in terms of efficiency and identity preservation, demonstrating a significant leap toward scalable customization of diffusion models.
Authors: Xu Shi, Wei Yao, Chuanchen Luo, Junran Peng, Hongwen Zhang, Yunlian Sun
Abstract: Recently, significant progress has been made in text-based motion generation, enabling the generation of diverse and high-quality human motions that conform to textual descriptions. However, generating motions beyond the distribution of original datasets remains challenging, i.e., zero-shot generation. By adopting a divide-and-conquer strategy, we propose a new framework named Fine-Grained Human Motion Diffusion Model (FG-MDM) for zero-shot human motion generation. Specifically, we first parse previous vague textual annotations into fine-grained descriptions of different body parts by leveraging a large language model. We then use these fine-grained descriptions to guide a transformer-based diffusion model, which further adopts a design of part tokens. FG-MDM can generate human motions beyond the scope of original datasets owing to descriptions that are closer to motion essence. Our experimental results demonstrate the superiority of FG-MDM over previous methods in zero-shot settings. We will release our fine-grained textual annotations for HumanML3D and KIT.
Authors: Enis Simsar, Alessio Tonioni, Yongqin Xian, Thomas Hofmann, Federico Tombari
Abstract: Diffusion models (DMs) have gained prominence due to their ability to generate high-quality varied images with recent advancements in text-to-image generation. The research focus is now shifting towards the controllability of DMs. A significant challenge within this domain is localized editing, where specific areas of an image are modified without affecting the rest of the content. This paper introduces LIME for localized image editing in diffusion models. LIME does not require user-specified regions of interest (RoI) or additional text input, but rather employs features from pre-trained methods and a straightforward clustering method to obtain precise editing mask. Then, by leveraging cross-attention maps, it refines these segments for finding regions to obtain localized edits. Finally, we propose a novel cross-attention regularization technique that penalizes unrelated cross-attention scores in the RoI during the denoising steps, ensuring localized edits. Our approach, without re-training, fine-tuning and additional user inputs, consistently improves the performance of existing methods in various editing benchmarks. The project page can be found at https://enisimsar.github.io/LIME/.
Authors: Yiqing Liang, Numair Khan, Zhengqin Li, Thu Nguyen-Phuoc, Douglas Lanman, James Tompkin, Lei Xiao
Abstract: We propose a method that achieves state-of-the-art rendering quality and efficiency on monocular dynamic scene reconstruction using deformable 3D Gaussians. Implicit deformable representations commonly model motion with a canonical space and time-dependent backward-warping deformation field. Our method, GauFRe, uses a forward-warping deformation to explicitly model non-rigid transformations of scene geometry. Specifically, we propose a template set of 3D Gaussians residing in a canonical space, and a time-dependent forward-warping deformation field to model dynamic objects. Additionally, we tailor a 3D Gaussian-specific static component supported by an inductive bias-aware initialization approach which allows the deformation field to focus on moving scene regions, improving the rendering of complex real-world motion. The differentiable pipeline is optimized end-to-end with a self-supervised rendering loss. Experiments show our method achieves competitive results and higher efficiency than both previous state-of-the-art NeRF and Gaussian-based methods. For real-world scenes, GauFRe can train in ~20 mins and offer 96 FPS real-time rendering on an RTX 3090 GPU. Project website: https://lynl7130.github.io/gaufre/index.html
Authors: Mehran Hosseini, Peyman Hosseini
Abstract: The enduring inability of image generative models to recreate intricate geometric features, such as those present in human hands and fingers has been an ongoing problem in image generation for nearly a decade. While strides have been made by increasing model sizes and diversifying training datasets, this issue remains prevalent across all models, from denoising diffusion models to Generative Adversarial Networks (GAN), pointing to a fundamental shortcoming in the underlying architectures. In this paper, we demonstrate how this problem can be mitigated by augmenting convolution layers geometric capabilities through providing them with a single input channel incorporating the relative n-dimensional Cartesian coordinate system. We show this drastically improves quality of images generated by Diffusion Models, GANs, and Variational AutoEncoders (VAE).
Authors: Shoubin Yu, Jaehong Yoon, Mohit Bansal
Abstract: Despite impressive advancements in recent multimodal reasoning approaches, they are still limited in flexibility and efficiency, as these models typically process only a few fixed modality inputs and require updates to numerous parameters. This paper tackles these critical challenges and proposes CREMA, a generalizable, highly efficient, and modular modality-fusion framework that can incorporate any new modality to enhance video reasoning. We first augment multiple informative modalities (such as optical flow, 3D point cloud, audio, thermal heatmap, and touch map) from given videos without extra human annotation by leveraging sensors or existing pre-trained models. Next, we introduce a query transformer with multiple parameter-efficient modules associated with each accessible modality. It projects diverse modality features to the LLM token embedding space, allowing the model to integrate different data types for response generation. Furthermore, we propose a novel progressive multimodal fusion design supported by a lightweight fusion module and modality-sequential training strategy. It helps compress information across various assisting modalities, maintaining computational efficiency in the LLM while improving performance. We validate our method on 7 video-language reasoning tasks assisted by diverse modalities, including conventional VideoQA and Video-Audio/3D/Touch/Thermal QA, and achieve better/equivalent performance against strong multimodal LLMs, including OneLLM, BLIP-2, and SeViLA while reducing over 90% trainable parameters. We provide extensive analyses of CREMA, including the impact of each modality on reasoning domains, the design of the fusion module, and example visualizations.
Authors: Zongyu Wu, Hongcheng Gao, Yueze Wang, Xiang Zhang, Suhang Wang
Abstract: Text-to-Image (T2I) models have shown great performance in generating images based on textual prompts. However, these models are vulnerable to unsafe input to generate unsafe content like sexual, harassment and illegal-activity images. Existing studies based on image checker, model fine-tuning and embedding blocking are impractical in real-world applications. Hence, we propose the first universal prompt optimizer for safe T2I (POSI) generation in black-box scenario. We first construct a dataset consisting of toxic-clean prompt pairs by GPT-3.5 Turbo. To guide the optimizer to have the ability of converting toxic prompt to clean prompt while preserving semantic information, we design a novel reward function measuring toxicity and text alignment of generated images and train the optimizer through Proximal Policy Optimization. Experiments show that our approach can effectively reduce the likelihood of various T2I models in generating inappropriate images, with no significant impact on text alignment. It is also flexible to be combined with methods to achieve better performance. Our code is available at https://github.com/wu-zongyu/POSI.
Authors: Alexander H. Berger, Laurin Lux, Suprosanna Shit, Ivan Ezhov, Georgios Kaissis, Martin J. Menten, Daniel Rueckert, Johannes C. Paetzold
Abstract: Direct image-to-graph transformation is a challenging task that involves solving object detection and relationship prediction in a single model. Due to this task's complexity, large training datasets are rare in many domains, making the training of deep-learning methods challenging. This data sparsity necessitates transfer learning strategies akin to the state-of-the-art in general computer vision. In this work, we introduce a set of methods enabling cross-domain and cross-dimension learning for image-to-graph transformers. We propose (1) a regularized edge sampling loss to effectively learn object relations in multiple domains with different numbers of edges, (2) a domain adaptation framework for image-to-graph transformers aligning image- and graph-level features from different domains, and (3) a projection function that allows using 2D data for training 3D transformers. We demonstrate our method's utility in cross-domain and cross-dimension experiments, where we utilize labeled data from 2D road networks for simultaneous learning in vastly different target domains. Our method consistently outperforms standard transfer learning and self-supervised pretraining on challenging benchmarks, such as retinal or whole-brain vessel graph extraction.
Authors: Haiyang Zheng, Nan Pu, Wenjing Li, Nicu Sebe, Zhun Zhong
Abstract: In this paper, we study the problem of Generalized Category Discovery (GCD), which aims to cluster unlabeled data from both known and unknown categories using the knowledge of labeled data from known categories. Current GCD methods rely on only visual cues, which however neglect the multi-modality perceptive nature of human cognitive processes in discovering novel visual categories. To address this, we propose a two-phase TextGCD framework to accomplish multi-modality GCD by exploiting powerful Visual-Language Models. TextGCD mainly includes a retrieval-based text generation (RTG) phase and a cross-modality co-teaching (CCT) phase. First, RTG constructs a visual lexicon using category tags from diverse datasets and attributes from Large Language Models, generating descriptive texts for images in a retrieval manner. Second, CCT leverages disparities between textual and visual modalities to foster mutual learning, thereby enhancing visual GCD. In addition, we design an adaptive class aligning strategy to ensure the alignment of category perceptions between modalities as well as a soft-voting mechanism to integrate multi-modality cues. Experiments on eight datasets show the large superiority of our approach over state-of-the-art methods. Notably, our approach outperforms the best competitor, by 7.7% and 10.8% in All accuracy on ImageNet-1k and CUB, respectively.
Authors: Tingtian Li, Zixun Sun, Xinyu Xiao
Abstract: Identifying highlight moments of raw video materials is crucial for improving the efficiency of editing videos that are pervasive on internet platforms. However, the extensive work of manually labeling footage has created obstacles to applying supervised methods to videos of unseen categories. The absence of an audio modality that contains valuable cues for highlight detection in many videos also makes it difficult to use multimodal strategies. In this paper, we propose a novel model with cross-modal perception for unsupervised highlight detection. The proposed model learns representations with visual-audio level semantics from image-audio pair data via a self-reconstruction task. To achieve unsupervised highlight detection, we investigate the latent representations of the network and propose the representation activation sequence learning (RASL) module with k-point contrastive learning to learn significant representation activations. To connect the visual modality with the audio modality, we use the symmetric contrastive learning (SCL) module to learn the paired visual and audio representations. Furthermore, an auxiliary task of masked feature vector sequence (FVS) reconstruction is simultaneously conducted during pretraining for representation enhancement. During inference, the cross-modal pretrained model can generate representations with paired visual-audio semantics given only the visual modality. The RASL module is used to output the highlight scores. The experimental results show that the proposed framework achieves superior performance compared to other state-of-the-art approaches.
Authors: Lingdong Kong, Xiang Xu, Jun Cen, Wenwei Zhang, Liang Pan, Kai Chen, Ziwei Liu
Abstract: Safety-critical 3D scene understanding tasks necessitate not only accurate but also confident predictions from 3D perception models. This study introduces Calib3D, a pioneering effort to benchmark and scrutinize the reliability of 3D scene understanding models from an uncertainty estimation viewpoint. We comprehensively evaluate 28 state-of-the-art models across 10 diverse 3D datasets, uncovering insightful phenomena that cope with both the aleatoric and epistemic uncertainties in 3D scene understanding. We discover that despite achieving impressive levels of accuracy, existing models frequently fail to provide reliable uncertainty estimates -- a pitfall that critically undermines their applicability in safety-sensitive contexts. Through extensive analysis of key factors such as network capacity, LiDAR representations, rasterization resolutions, and 3D data augmentation techniques, we correlate these aspects directly with the model calibration efficacy. Furthermore, we introduce DeptS, a novel depth-aware scaling approach aimed at enhancing 3D model calibration. Extensive experiments across a wide range of configurations validate the superiority of our method. We hope this work could serve as a cornerstone for fostering reliable 3D scene understanding. Code and benchmark toolkit are publicly available.
Authors: Guanhua Ding, Zexi Ye, Zhen Zhong, Gang Li, David Shao
Abstract: Block pruning, which eliminates contiguous blocks of weights, is a structural pruning method that can significantly enhance the performance of neural processing units (NPUs). In industrial applications, an ideal block pruning algorithm should meet three key requirements: (1) maintain high accuracy across diverse models and tasks, as machine learning deployments on edge devices are typically accuracy-critical; (2) offer precise control over resource constraints to facilitate user adoption; and (3) provide convergence guarantees to prevent performance instability. However, to the best of our knowledge, no existing block pruning algorithm satisfies all three requirements simultaneously. In this paper, we introduce SMART (Separate, Dynamic, and Differentiable) pruning, a novel algorithm designed to address this gap. SMART leverages both weight and activation information to enhance accuracy, employs a differentiable top-k operator for precise control of resource constraints, and offers convergence guarantees under mild conditions. Extensive experiments involving seven models, four datasets, three different block types, and three computer vision tasks demonstrate that SMART pruning achieves state-of-the-art performance in block pruning.
Authors: Akshita Gupta, Gaurav Mittal, Ahmed Magooda, Ye Yu, Graham W. Taylor, Mei Chen
Abstract: Temporal Action Localization (TAL) involves localizing and classifying action snippets in an untrimmed video. The emergence of large video foundation models has led RGB-only video backbones to outperform previous methods needing both RGB and optical flow modalities. Leveraging these large models is often limited to training only the TAL head due to the prohibitively large GPU memory required to adapt the video backbone for TAL. To overcome this limitation, we introduce LoSA, the first memory-and-parameter-efficient backbone adapter designed specifically for TAL to handle untrimmed videos. LoSA specializes for TAL by introducing Long-Short-range Adapters that adapt the intermediate layers of the video backbone over different temporal ranges. These adapters run parallel to the video backbone to significantly reduce memory footprint. LoSA also includes Long-Short-range Gated Fusion that strategically combines the output of these adapters from the video backbone layers to enhance the video features provided to the TAL head. Experiments show that LoSA significantly outperforms all existing methods on standard TAL benchmarks, THUMOS-14 and ActivityNet-v1.3, by scaling end-to-end backbone adaptation to billion-parameter-plus models like VideoMAEv2~(ViT-g) and leveraging them beyond head-only transfer learning.
Authors: Yuliang Liu, Mingxin Huang, Hao Yan, Linger Deng, Weijia Wu, Hao Lu, Chunhua Shen, Lianwen Jin, Xiang Bai
Abstract: Text spotting, a task involving the extraction of textual information from image or video sequences, faces challenges in cross-domain adaption, such as image-to-image and image-to-video generalization. In this paper, we introduce a new method, termed VimTS, which enhances the generalization ability of the model by achieving better synergy among different tasks. Typically, we propose a Prompt Queries Generation Module and a Tasks-aware Adapter to effectively convert the original single-task model into a multi-task model suitable for both image and video scenarios with minimal additional parameters. The Prompt Queries Generation Module facilitates explicit interaction between different tasks, while the Tasks-aware Adapter helps the model dynamically learn suitable features for each task. Additionally, to further enable the model to learn temporal information at a lower cost, we propose a synthetic video text dataset (VTD-368k) by leveraging the Content Deformation Fields (CoDeF) algorithm. Notably, our method outperforms the state-of-the-art method by an average of 2.6% in six cross-domain benchmarks such as TT-to-IC15, CTW1500-to-TT, and TT-to-CTW1500. For video-level cross-domain adaption, our method even surpasses the previous end-to-end video spotting method in ICDAR2015 video and DSText v2 by an average of 5.5% on the MOTA metric, using only image-level data. We further demonstrate that existing Large Multimodal Models exhibit limitations in generating cross-domain scene text spotting, in contrast to our VimTS model which requires significantly fewer parameters and data. The code and datasets will be made available at the https://VimTextSpotter.github.io.
Authors: Jonathan Roberts, Kai Han, Neil Houlsby, Samuel Albanie
Abstract: Large multimodal models (LMMs) have proven flexible and generalisable across many tasks and fields. Although they have strong potential to aid scientific research, their capabilities in this domain are not well characterised. A key aspect of scientific research is the ability to understand and interpret figures, which serve as a rich, compressed source of complex information. In this work, we present SciFIBench, a scientific figure interpretation benchmark consisting of 2000 questions split between two tasks across 8 categories. The questions are curated from arXiv paper figures and captions, using adversarial filtering to find hard negatives and human verification for quality control. We evaluate 28 LMMs on SciFIBench, finding it to be a challenging benchmark. Finally, we investigate the alignment and reasoning faithfulness of the LMMs on augmented question sets from our benchmark. We release SciFIBench to encourage progress in this domain.
Authors: Xiaolin Qin, Jiacen Liu, Qianlei Wang, Shaolin Zhang, Fei Zhu, Zhang Yi
Abstract: Glass largely blurs the boundary between the real world and the reflection. The special transmittance and reflectance quality have confused the semantic tasks related to machine vision. Therefore, how to clear the boundary built by glass, and avoid over-capturing features as false positive information in deep structure, matters for constraining the segmentation of reflection surface and penetrating glass. We proposed the Fourier Boundary Features Network with Wider Catchers (FBWC), which might be the first attempt to utilize sufficiently wide horizontal shallow branches without vertical deepening for guiding the fine granularity segmentation boundary through primary glass semantic information. Specifically, we designed the Wider Coarse-Catchers (WCC) for anchoring large area segmentation and reducing excessive extraction from a structural perspective. We embed fine-grained features by Cross Transpose Attention (CTA), which is introduced to avoid the incomplete area within the boundary caused by reflection noise. For excavating glass features and balancing high-low layers context, a learnable Fourier Convolution Controller (FCC) is proposed to regulate information integration robustly. The proposed method has been validated on three different public glass segmentation datasets. Experimental results reveal that the proposed method yields better segmentation performance compared with the state-of-the-art (SOTA) methods in glass image segmentation.
Authors: Maryam Hosseini, Marco Cipriano, Sedigheh Eslami, Daniel Hodczak, Liu Liu, Andres Sevtsuk, Gerard de Melo
Abstract: Existing Open Vocabulary Detection (OVD) models exhibit a number of challenges. They often struggle with semantic consistency across diverse inputs, and are often sensitive to slight variations in input phrasing, leading to inconsistent performance. The calibration of their predictive confidence, especially in complex multi-label scenarios, remains suboptimal, frequently resulting in overconfident predictions that do not accurately reflect their context understanding. To understand these limitations, multi-label detection benchmarks are needed. A particularly challenging domain for such benchmarking is social activities. Due to the lack of multi-label benchmarks for social interactions, in this work we present ELSA: Evaluating Localization of Social Activities. ELSA draws on theoretical frameworks in urban sociology and design and uses in-the-wild street-level imagery, where the size of groups and the types of activities vary significantly. ELSA includes more than 900 manually annotated images with more than 4,300 multi-labeled bounding boxes for individual and group activities. We introduce a novel confidence score computation method NLSE and a novel Dynamic Box Aggregation (DBA) algorithm to assess semantic consistency in overlapping predictions. We report our results on the widely-used SOTA models Grounding DINO, Detic, OWL, and MDETR. Our evaluation protocol considers semantic stability and localization accuracy and further exposes the limitations of existing approaches.
Authors: Jikai Wang, Qifan Zhang, Yu-Wei Chao, Bowen Wen, Xiaohu Guo, Yu Xiang
Abstract: We introduce a data capture system and a new dataset, HO-Cap, for 3D reconstruction and pose tracking of hands and objects in videos. The system leverages multiple RGB-D cameras and a HoloLens headset for data collection, avoiding the use of expensive 3D scanners or mocap systems. We propose a semi-automatic method for annotating the shape and pose of hands and objects in the collected videos, significantly reducing the annotation time compared to manual labeling. With this system, we captured a video dataset of humans interacting with objects to perform various tasks, including simple pick-and-place actions, handovers between hands, and using objects according to their affordance, which can serve as human demonstrations for research in embodied AI and robot manipulation. Our data capture setup and annotation framework will be available for the community to use in reconstructing 3D shapes of objects and human hands and tracking their poses in videos.
Authors: Rohit Jena, Pratik Chaudhari, James C. Gee
Abstract: Deep Learning in Image Registration (DLIR) methods have been tremendously successful in image registration due to their speed and ability to incorporate weak label supervision at training time. However, existing DLIR methods forego many of the benefits and invariances of optimization methods. The lack of a task-specific inductive bias in DLIR methods leads to suboptimal performance, especially in the presence of domain shift. Our method aims to bridge this gap between statistical learning and optimization by explicitly incorporating optimization as a layer in a deep network. A deep network is trained to predict multi-scale dense feature images that are registered using a black box iterative optimization solver. This optimal warp is then used to minimize image and label alignment errors. By implicitly differentiating end-to-end through an iterative optimization solver, we explicitly exploit invariances of the correspondence matching problem induced by the optimization, while learning registration and label-aware features, and guaranteeing the warp functions to be a local minima of the registration objective in the feature space. Our framework shows excellent performance on in-domain datasets, and is agnostic to domain shift such as anisotropy and varying intensity profiles. For the first time, our method allows switching between arbitrary transformation representations (free-form to diffeomorphic) at test time with zero retraining. End-to-end feature learning also facilitates interpretability of features and arbitrary test-time regularization, which is not possible with existing DLIR methods.
Authors: Javier Montalvo, Juan Ignacio Bravo P\'erez-Villar, \'Alvaro Garc\'ia-Mart\'in, Pablo Carballeira, Jes\'us Besc\'os
Abstract: The scarcity of data acquired under actual space operational conditions poses a significant challenge for developing learning-based visual navigation algorithms crucial for autonomous spacecraft navigation. This data shortage is primarily due to the prohibitive costs and inherent complexities of space operations. While existing datasets, predominantly relying on computer-simulated data, have partially addressed this gap, they present notable limitations. Firstly, these datasets often utilize proprietary image generation tools, restricting the evaluation of navigation methods in novel, unseen scenarios. Secondly, they provide limited ground-truth data, typically focusing solely on the spacecraft's translation and rotation relative to the camera. To address these limitations, we present SPIN (SPacecraft Imagery for Navigation), an open-source spacecraft image generation tool designed to support a wide range of visual navigation scenarios in space, with a particular focus on relative navigation tasks. SPIN provides multiple modalities of ground-truth data and allows researchers to employ custom 3D models of satellites, define specific camera-relative poses, and adjust settings such as camera parameters or environmental illumination conditions. We also propose a method for exploiting our tool as a data augmentation module. We validate our tool on the spacecraft pose estimation task by training with a SPIN-generated replica of SPEED+, reaching a 47% average error reduction on SPEED+ testbed data (that simulates realistic space conditions), further reducing it to a 60% error reduction when using SPIN as a data augmentation method. Both the SPIN tool (and source code) and our SPIN-generated version of SPEED+ will be publicly released upon paper acceptance on GitHub. https://github.com/vpulab/SPIN
Authors: Alex Hanson, Allen Tu, Vasu Singla, Mayuka Jayawardhana, Matthias Zwicker, Tom Goldstein
Abstract: Recent advances in novel view synthesis have enabled real-time rendering speeds with high reconstruction accuracy. 3D Gaussian Splatting (3D-GS), a foundational point-based parametric 3D scene representation, models scenes as large sets of 3D Gaussians. However, complex scenes can consist of millions of Gaussians, resulting in high storage and memory requirements that limit the viability of 3D-GS on devices with limited resources. Current techniques for compressing these pretrained models by pruning Gaussians rely on combining heuristics to determine which Gaussians to remove. At high compression ratios, these pruned scenes suffer from heavy degradation of visual fidelity and loss of foreground details. In this paper, we propose a principled sensitivity pruning score that preserves visual fidelity and foreground details at significantly higher compression ratios than existing approaches. It is computed as a second-order approximation of the reconstruction error on the training views with respect to the spatial parameters of each Gaussian. Additionally, we propose a multi-round prune-refine pipeline that can be applied to any pretrained 3D-GS model without changing its training pipeline. After pruning 90% of Gaussians, a substantially higher percentage than previous methods, our PUP 3D-GS pipeline increases average rendering speed by 3.56$\times$ while retaining more salient foreground information and achieving higher image quality metrics than existing techniques on scenes from the Mip-NeRF 360, Tanks & Temples, and Deep Blending datasets.
Authors: Bingqi Ma, Zhuofan Zong, Guanglu Song, Hongsheng Li, Yu Liu
Abstract: Large language models (LLMs) based on decoder-only transformers have demonstrated superior text understanding capabilities compared to CLIP and T5-series models. However, the paradigm for utilizing current advanced LLMs in text-to-image diffusion models remains to be explored. We observed an unusual phenomenon: directly using a large language model as the prompt encoder significantly degrades the prompt-following ability in image generation. We identified two main obstacles behind this issue. One is the misalignment between the next token prediction training in LLM and the requirement for discriminative prompt features in diffusion models. The other is the intrinsic positional bias introduced by the decoder-only architecture. To deal with this issue, we propose a novel framework to fully harness the capabilities of LLMs. Through the carefully designed usage guidance, we effectively enhance the text representation capability for prompt encoding and eliminate its inherent positional bias. This allows us to integrate state-of-the-art LLMs into the text-to-image generation model flexibly. Furthermore, we also provide an effective manner to fuse multiple LLMs into our framework. Considering the excellent performance and scaling capabilities demonstrated by the transformer architecture, we further design an LLM-Infused Diffusion Transformer (LI-DiT) based on the framework. We conduct extensive experiments to validate LI-DiT across model size and data size. Benefiting from the inherent ability of the LLMs and our innovative designs, the prompt understanding performance of LI-DiT easily surpasses state-of-the-art open-source models as well as mainstream closed-source commercial models including Stable Diffusion 3, DALL-E 3, and Midjourney V6. The LLM-Infused Diffuser framework is also one of the core technologies powering SenseMirage, a highly advanced text-to-image model.
Authors: Reza Akbarian Bafghi, Nidhin Harilal, Claire Monteleoni, Maziar Raissi
Abstract: This paper introduces MixDiff, a new self-supervised learning (SSL) pre-training framework that combines real and synthetic images. Unlike traditional SSL methods that predominantly use real images, MixDiff uses a variant of Stable Diffusion to replace an augmented instance of a real image, facilitating the learning of cross real-synthetic image representations. Our key insight is that while models trained solely on synthetic images underperform, combining real and synthetic data leads to more robust and adaptable representations. Experiments show MixDiff enhances SimCLR, BarlowTwins, and DINO across various robustness datasets and domain transfer tasks, boosting SimCLR's ImageNet-1K accuracy by 4.56%. Our framework also demonstrates comparable performance without needing any augmentations, a surprising finding in SSL where augmentations are typically crucial.
Authors: Tao Lian, Jose L. G\'omez, Antonio M. L\'opez
Abstract: The last mile of unsupervised domain adaptation (UDA) for semantic segmentation is the challenge of solving the syn-to-real domain gap. Recent UDA methods have progressed significantly, yet they often rely on strategies customized for synthetic single-source datasets (e.g., GTA5), which limits their generalisation to multi-source datasets. Conversely, synthetic multi-source datasets hold promise for advancing the last mile of UDA but remain underutilized in current research. Thus, we propose DEC, a flexible UDA framework for multi-source datasets. Following a divide-and-conquer strategy, DEC simplifies the task by categorizing semantic classes, training models for each category, and fusing their outputs by an ensemble model trained exclusively on synthetic datasets to obtain the final segmentation mask. DEC can integrate with existing UDA methods, achieving state-of-the-art performance on Cityscapes, BDD100K, and Mapillary Vistas, significantly narrowing the syn-to-real domain gap.
Authors: Kun Wu, Zhiguo Jiang, Kunming Tang, Jun Shi, Fengying Xie, Wei Wang, Haibo Wu, Yushan Zheng
Abstract: Large-scale pre-training models have promoted the development of histopathology image analysis. However, existing self-supervised methods for histopathology images focus on learning patch features, while there is still a lack of available pre-training models for WSI-level feature learning. In this paper, we propose a novel self-supervised learning framework for pan-cancer WSI-level representation pre-training with the designed position-aware masked autoencoder (PAMA). Meanwhile, we propose the position-aware cross-attention (PACA) module with a kernel reorientation (KRO) strategy and an anchor dropout (AD) mechanism. The KRO strategy can capture the complete semantic structure and eliminate ambiguity in WSIs, and the AD contributes to enhancing the robustness and generalization of the model. We evaluated our method on 6 large-scale datasets from multiple organs for pan-cancer classification tasks. The results have demonstrated the effectiveness of PAMA in generalized and discriminative WSI representation learning and pan-cancer WSI pre-training. The proposed method was also compared with 7 WSI analysis methods. The experimental results have indicated that our proposed PAMA is superior to the state-of-the-art methods.The code and checkpoints are available at https://github.com/WkEEn/PAMA.
Authors: Cherie Ho, Jiaye Zou, Omar Alama, Sai Mitheran Jagadesh Kumar, Benjamin Chiang, Taneesh Gupta, Chen Wang, Nikhil Keetha, Katia Sycara, Sebastian Scherer
Abstract: Top-down Bird's Eye View (BEV) maps are a popular representation for ground robot navigation due to their richness and flexibility for downstream tasks. While recent methods have shown promise for predicting BEV maps from First-Person View (FPV) images, their generalizability is limited to small regions captured by current autonomous vehicle-based datasets. In this context, we show that a more scalable approach towards generalizable map prediction can be enabled by using two large-scale crowd-sourced mapping platforms, Mapillary for FPV images and OpenStreetMap for BEV semantic maps. We introduce Map It Anywhere (MIA), a data engine that enables seamless curation and modeling of labeled map prediction data from existing open-source map platforms. Using our MIA data engine, we display the ease of automatically collecting a dataset of 1.2 million pairs of FPV images & BEV maps encompassing diverse geographies, landscapes, environmental factors, camera models & capture scenarios. We further train a simple camera model-agnostic model on this data for BEV map prediction. Extensive evaluations using established benchmarks and our dataset show that the data curated by MIA enables effective pretraining for generalizable BEV map prediction, with zero-shot performance far exceeding baselines trained on existing datasets by 35%. Our analysis highlights the promise of using large-scale public maps for developing & testing generalizable BEV perception, paving the way for more robust autonomous navigation. Website: https://mapitanywhere.github.io/
Authors: Ahmet Serdar Karadeniz, Dimitrios Mallis, Nesryne Mejri, Kseniya Cherenkova, Anis Kacem, Djamila Aouada
Abstract: This work introduces PICASSO, a framework for the parameterization of 2D CAD sketches from hand-drawn and precise sketch images. PICASSO converts a given CAD sketch image into parametric primitives that can be seamlessly integrated into CAD software. Our framework leverages rendering self-supervision to enable the pre-training of a CAD sketch parameterization network using sketch renderings only, thereby eliminating the need for corresponding CAD parameterization. Thus, we significantly reduce reliance on parameter-level annotations, which are often unavailable, particularly for hand-drawn sketches. The two primary components of PICASSO are (1) a Sketch Parameterization Network (SPN) that predicts a series of parametric primitives from CAD sketch images, and (2) a Sketch Rendering Network (SRN) that renders parametric CAD sketches in a differentiable manner and facilitates the computation of a rendering (image-level) loss for self-supervision. We demonstrate that the proposed PICASSO can achieve reasonable performance even when finetuned with only a small number of parametric CAD sketches. Extensive evaluation on the widely used SketchGraphs and CAD as Language datasets validates the effectiveness of the proposed approach on zero- and few-shot learning scenarios.
Authors: Jae Joong Lee, Bedrich Benes
Abstract: We introduce RGB2Point, an unposed single-view RGB image to a 3D point cloud generation based on Transformer. RGB2Point takes an input image of an object and generates a dense 3D point cloud. Contrary to prior works based on CNN layers and diffusion denoising approaches, we use pre-trained Transformer layers that are fast and generate high-quality point clouds with consistent quality over available categories. Our generated point clouds demonstrate high quality on a real-world dataset, as evidenced by improved Chamfer distance (51.15%) and Earth Mover's distance (45.96%) metrics compared to the current state-of-the-art. Additionally, our approach shows a better quality on a synthetic dataset, achieving better Chamfer distance (39.26%), Earth Mover's distance (26.95%), and F-score (47.16%). Moreover, our method produces 63.1% more consistent high-quality results across various object categories compared to prior works. Furthermore, RGB2Point is computationally efficient, requiring only 2.3GB of VRAM to reconstruct a 3D point cloud from a single RGB image, and our implementation generates the results 15,133x faster than a SOTA diffusion-based model.
Authors: Orest Kupyn, Eugene Khvedchenia, Christian Rupprecht
Abstract: Human head detection, keypoint estimation, and 3D head model fitting are essential tasks with many applications. However, traditional real-world datasets often suffer from bias, privacy, and ethical concerns, and they have been recorded in laboratory environments, which makes it difficult for trained models to generalize. Here, we introduce \method -- a large-scale synthetic dataset generated with diffusion models for human head detection and 3D mesh estimation. Our dataset comprises over 1 million high-resolution images, each annotated with detailed 3D head meshes, facial landmarks, and bounding boxes. Using this dataset, we introduce a new model architecture capable of simultaneous head detection and head mesh reconstruction from a single image in a single step. Through extensive experimental evaluations, we demonstrate that models trained on our synthetic data achieve strong performance on real images. Furthermore, the versatility of our dataset makes it applicable across a broad spectrum of tasks, offering a general and comprehensive representation of human heads.
Authors: Jiayuan Zhu, Abdullah Hamdi, Yunli Qi, Yueming Jin, Junde Wu
Abstract: Medical image segmentation plays a pivotal role in clinical diagnostics and treatment planning, yet existing models often face challenges in generalization and in handling both 2D and 3D data uniformly. In this paper, we introduce Medical SAM 2 (MedSAM-2), a generalized auto-tracking model for universal 2D and 3D medical image segmentation. The core concept is to leverage the Segment Anything Model 2 (SAM2) pipeline to treat all 2D and 3D medical segmentation tasks as a video object tracking problem. To put it into practice, we propose a novel \emph{self-sorting memory bank} mechanism that dynamically selects informative embeddings based on confidence and dissimilarity, regardless of temporal order. This mechanism not only significantly improves performance in 3D medical image segmentation but also unlocks a \emph{One-Prompt Segmentation} capability for 2D images, allowing segmentation across multiple images from a single prompt without temporal relationships. We evaluated MedSAM-2 on five 2D tasks and nine 3D tasks, including white blood cells, optic cups, retinal vessels, mandibles, coronary arteries, kidney tumors, liver tumors, breast cancer, nasopharynx cancer, vestibular schwannoma, mediastinal lymph nodules, cerebral artery, inferior alveolar nerve, and abdominal organs, comparing it against state-of-the-art (SOTA) models in task-tailored, general and interactive segmentation settings. Our findings demonstrate that MedSAM-2 surpasses a wide range of existing models and updates new SOTA on several benchmarks. The code is released on the project page: https://supermedintel.github.io/Medical-SAM2/.
Authors: Yiming Zhong, Xiaolin Zhang, Yao Zhao, Yunchao Wei
Abstract: Recently, the text-to-3D task has developed rapidly due to the appearance of the SDS method. However, the SDS method always generates 3D objects with poor quality due to the over-smooth issue. This issue is attributed to two factors: 1) the DDPM single-step inference produces poor guidance gradients; 2) the randomness from the input noises and timesteps averages the details of the 3D contents. In this paper, to address the issue, we propose DreamLCM which incorporates the Latent Consistency Model (LCM). DreamLCM leverages the powerful image generation capabilities inherent in LCM, enabling generating consistent and high-quality guidance, i.e., predicted noises or images. Powered by the improved guidance, the proposed method can provide accurate and detailed gradients to optimize the target 3D models. In addition, we propose two strategies to enhance the generation quality further. Firstly, we propose a guidance calibration strategy, utilizing Euler Solver to calibrate the guidance distribution to accelerate 3D models to converge. Secondly, we propose a dual timestep strategy, increasing the consistency of guidance and optimizing 3D models from geometry to appearance in DreamLCM. Experiments show that DreamLCM achieves state-of-the-art results in both generation quality and training efficiency. The code is available at https://github.com/1YimingZhong/DreamLCM.
Authors: Chuangchuang Tan, Renshuai Tao, Huan Liu, Guanghua Gu, Baoyuan Wu, Yao Zhao, Yunchao Wei
Abstract: This work focuses on AIGC detection to develop universal detectors capable of identifying various types of forgery images. Recent studies have found large pre-trained models, such as CLIP, are effective for generalizable deepfake detection along with linear classifiers. However, two critical issues remain unresolved: 1) understanding why CLIP features are effective on deepfake detection through a linear classifier; and 2) exploring the detection potential of CLIP. In this study, we delve into the underlying mechanisms of CLIP's detection capabilities by decoding its detection features into text and performing word frequency analysis. Our finding indicates that CLIP detects deepfakes by recognizing similar concepts (Fig. \ref{fig:fig1} a). Building on this insight, we introduce Category Common Prompt CLIP, called C2P-CLIP, which integrates the category common prompt into the text encoder to inject category-related concepts into the image encoder, thereby enhancing detection performance (Fig. \ref{fig:fig1} b). Our method achieves a 12.41\% improvement in detection accuracy compared to the original CLIP, without introducing additional parameters during testing. Comprehensive experiments conducted on two widely-used datasets, encompassing 20 generation models, validate the efficacy of the proposed method, demonstrating state-of-the-art performance. The code is available at \url{https://github.com/chuangchuangtan/C2P-CLIP-DeepfakeDetection}
URLs: https://github.com/chuangchuangtan/C2P-CLIP-DeepfakeDetection
Authors: Mohsen Asghari Ilani, Leila Amini, Hossein Karimi, Maryam Shavali Kuhshuri
Abstract: Numerous image processing techniques (IPTs) have been employed to detect crack defects, offering an alternative to human-conducted onsite inspections. These IPTs manipulate images to extract defect features, particularly cracks in surfaces produced through Additive Manufacturing (AM). This article presents a vision-based approach that utilizes deep convolutional neural networks (CNNs) for crack detection in AM surfaces. Traditional image processing techniques face challenges with diverse real-world scenarios and varying crack types. To overcome these challenges, our proposed method leverages CNNs, eliminating the need for extensive feature extraction. Annotation for CNN training is facilitated by LabelImg without the requirement for additional IPTs. The trained CNN, enhanced by OpenCV preprocessing techniques, achieves an outstanding 99.54% accuracy on a dataset of 14,982 annotated images with resolutions of 1536 x 1103 pixels. Evaluation metrics exceeding 96% precision, 98% recall, and a 97% F1-score highlight the precision and effectiveness of the entire process.
Authors: Mohsen Asghari Ilani, Saba Moftakhar Tehran, Ashkan Kavei, Arian Radmehr
Abstract: In response to the burgeoning global demand for premium agricultural products, particularly within the competitive nut market, this paper introduces an innovative methodology aimed at enhancing the grading process for almonds and their shells. Leveraging state-of-the-art Deep Convolutional Neural Networks (CNNs), specifically the AlmondNet-20 architecture, our study achieves exceptional accuracy exceeding 99%, facilitated by the utilization of a 20-layer CNN model. To bolster robustness in differentiating between almonds and shells, data augmentation techniques are employed, ensuring the reliability and accuracy of our classification system. Our model, meticulously trained over 1000 epochs, demonstrates remarkable performance, boasting an accuracy rate of 99% alongside a minimal loss function of 0.0567. Rigorous evaluation through test datasets further validates the efficacy of our approach, revealing impeccable precision, recall, and F1-score metrics for almond detection. Beyond its technical prowess, this advanced classification system offers tangible benefits to both industry experts and non-specialists alike, ensuring globally reliable almond classification. The application of deep learning algorithms, as showcased in our study, not only enhances grading accuracy but also presents opportunities for product patents, thereby contributing to the economic value of our nation. Through the adoption of cutting-edge technologies such as the AlmondNet-20 model, we pave the way for future advancements in agricultural product classification, ultimately enriching global trade and economic prosperity.
Authors: Xiaoyu Guo, Pengzhi Zhong, Hao Zhang, Defeng Huang, Huikai Shao, Qijun Zhao, Shuiwang Li
Abstract: Visual tracking has seen remarkable advancements, largely driven by the availability of large-scale training datasets that have enabled the development of highly accurate and robust algorithms. While significant progress has been made in tracking general objects, research on more challenging scenarios, such as tracking camouflaged objects, remains limited. Camouflaged objects, which blend seamlessly with their surroundings or other objects, present unique challenges for detection and tracking in complex environments. This challenge is particularly critical in applications such as military, security, agriculture, and marine monitoring, where precise tracking of camouflaged objects is essential. To address this gap, we introduce the Camouflaged Object Tracking Dataset (COTD), a specialized benchmark designed specifically for evaluating camouflaged object tracking methods. The COTD dataset comprises 200 sequences and approximately 80,000 frames, each annotated with detailed bounding boxes. Our evaluation of 20 existing tracking algorithms reveals significant deficiencies in their performance with camouflaged objects. To address these issues, we propose a novel tracking framework, HiPTrack-MLS, which demonstrates promising results in improving tracking performance for camouflaged objects. COTD and code are avialable at https://github.com/openat25/HIPTrack-MLS.
Authors: Wei-Jhe Huang, Min-Hung Chen, Shang-Hong Lai
Abstract: Spatio-temporal action detection encompasses the tasks of localizing and classifying individual actions within a video. Recent works aim to enhance this process by incorporating interaction modeling, which captures the relationship between people and their surrounding context. However, these approaches have primarily focused on fully-supervised learning, and the current limitation lies in the lack of generalization capability to recognize unseen action categories. In this paper, we aim to adapt the pretrained image-language models to detect unseen actions. To this end, we propose a method which can effectively leverage the rich knowledge of visual-language models to perform Person-Context Interaction. Meanwhile, our Context Prompting module will utilize contextual information to prompt labels, thereby enhancing the generation of more representative text features. Moreover, to address the challenge of recognizing distinct actions by multiple people at the same timestamp, we design the Interest Token Spotting mechanism which employs pretrained visual knowledge to find each person's interest context tokens, and then these tokens will be used for prompting to generate text features tailored to each individual. To evaluate the ability to detect unseen actions, we propose a comprehensive benchmark on J-HMDB, UCF101-24, and AVA datasets. The experiments show that our method achieves superior results compared to previous approaches and can be further extended to multi-action videos, bringing it closer to real-world applications. The code and data can be found in https://webber2933.github.io/ST-CLIP-project-page.
Authors: Jia-Wei Liao, Winston Wang, Tzu-Sian Wang, Li-Xuan Peng, Ju-Hsuan Weng, Cheng-Fu Chou, Jun-Cheng Chen
Abstract: With the success of Diffusion Models for image generation, the technologies also have revolutionized the aesthetic Quick Response (QR) code generation. Despite significant improvements in visual attractiveness for the beautified codes, their scannabilities are usually sacrificed and thus hinder their practical uses in real-world scenarios. To address this issue, we propose a novel training-free Diffusion-based QR Code generator (DiffQRCoder) to effectively craft both scannable and visually pleasing QR codes. The proposed approach introduces Scanning-Robust Perceptual Guidance (SRPG), a new diffusion guidance for Diffusion Models to guarantee the generated aesthetic codes to obey the ground-truth QR codes while maintaining their attractiveness during the denoising process. Additionally, we present another post-processing technique, Scanning Robust Manifold Projected Gradient Descent (SR-MPGD), to further enhance their scanning robustness through iterative latent space optimization. With extensive experiments, the results demonstrate that our approach not only outperforms other compared methods in Scanning Success Rate (SSR) with better or comparable CLIP aesthetic score (CLIP-aes.) but also significantly improves the SSR of the ControlNet-only approach from 60% to 99%. The subjective evaluation indicates that our approach achieves promising visual attractiveness to users as well. Finally, even with different scanning angles and the most rigorous error tolerance settings, our approach robustly achieves over 95% SSR, demonstrating its capability for real-world applications. Our project page is available at https://jwliao1209.github.io/DiffQRCoder.
Authors: Shaode Yu, Ze Chen, Zhimu Yang, Jiacheng Gu, Bizu Feng
Abstract: Score prediction is crucial in evaluating realistic image sharpness based on collected informative features. Recently, Kolmogorov-Arnold networks (KANs) have been developed and witnessed remarkable success in data fitting. This study introduces the Taylor series-based KAN (TaylorKAN). Then, different KANs are explored in four realistic image databases (BID2011, CID2013, CLIVE, and KonIQ-10k) to predict the scores by using 15 mid-level features and 2048 high-level features. Compared to support vector regression, results show that KANs are generally competitive or superior, and TaylorKAN is the best one when mid-level features are used. This is the first study to investigate KANs on image quality assessment that sheds some light on how to select and further improve KANs in related tasks.
Authors: Matt Deitke, Christopher Clark, Sangho Lee, Rohun Tripathi, Yue Yang, Jae Sung Park, Mohammadreza Salehi, Niklas Muennighoff, Kyle Lo, Luca Soldaini, Jiasen Lu, Taira Anderson, Erin Bransom, Kiana Ehsani, Huong Ngo, YenSung Chen, Ajay Patel, Mark Yatskar, Chris Callison-Burch, Andrew Head, Rose Hendrix, Favyen Bastani, Eli VanderBilt, Nathan Lambert, Yvonne Chou, Arnavi Chheda, Jenna Sparks, Sam Skjonsberg, Michael Schmitz, Aaron Sarnat, Byron Bischoff, Pete Walsh, Chris Newell, Piper Wolters, Tanmay Gupta, Kuo-Hao Zeng, Jon Borchardt, Dirk Groeneveld, Crystal Nam, Sophie Lebrecht, Caitlin Wittlif, Carissa Schoenick, Oscar Michel, Ranjay Krishna, Luca Weihs, Noah A. Smith, Hannaneh Hajishirzi, Ross Girshick, Ali Farhadi, Aniruddha Kembhavi
Abstract: Today's most advanced vision-language models (VLMs) remain proprietary. The strongest open-weight models rely heavily on synthetic data from proprietary VLMs to achieve good performance, effectively distilling these closed VLMs into open ones. As a result, the community has been missing foundational knowledge about how to build performant VLMs from scratch. We present Molmo, a new family of VLMs that are state-of-the-art in their class of openness. Our key contribution is a collection of new datasets called PixMo, including a dataset of highly detailed image captions for pre-training, a free-form image Q&A dataset for fine-tuning, and an innovative 2D pointing dataset, all collected without the use of external VLMs. The success of our approach relies on careful modeling choices, a well-tuned training pipeline, and, most critically, the quality of our newly collected datasets. Our best-in-class 72B model not only outperforms others in the class of open weight and data models, but also outperforms larger proprietary models including Claude 3.5 Sonnet, and Gemini 1.5 Pro and Flash, second only to GPT-4o based on both academic benchmarks and on a large human evaluation. Our model weights, new datasets, and source code are available at https://molmo.allenai.org/blog.
Authors: Janek Haberer, Ali Hojjat, Olaf Landsiedel
Abstract: The architecture of Vision Transformers (ViTs), particularly the Multi-head Attention (MHA) mechanism, imposes substantial hardware demands. Deploying ViTs on devices with varying constraints, such as mobile phones, requires multiple models of different sizes. However, this approach has limitations, such as training and storing each required model separately. This paper introduces HydraViT, a novel approach that addresses these limitations by stacking attention heads to achieve a scalable ViT. By repeatedly changing the size of the embedded dimensions throughout each layer and their corresponding number of attention heads in MHA during training, HydraViT induces multiple subnetworks. Thereby, HydraViT achieves adaptability across a wide spectrum of hardware environments while maintaining performance. Our experimental results demonstrate the efficacy of HydraViT in achieving a scalable ViT with up to 10 subnetworks, covering a wide range of resource constraints. HydraViT achieves up to 5 p.p. more accuracy with the same GMACs and up to 7 p.p. more accuracy with the same throughput on ImageNet-1K compared to the baselines, making it an effective solution for scenarios where hardware availability is diverse or varies over time. Source code available at https://github.com/ds-kiel/HydraViT.
Authors: Jacopo Dapueto, Nicoletta Noceti, Francesca Odone
Abstract: Developing meaningful and efficient representations that separate the fundamental structure of the data generation mechanism is crucial in representation learning. However, Disentangled Representation Learning has not fully shown its potential on real images, because of correlated generative factors, their resolution and limited access to ground truth labels. Specifically on the latter, we investigate the possibility of leveraging synthetic data to learn general-purpose disentangled representations applicable to real data, discussing the effect of fine-tuning and what properties of disentanglement are preserved after the transfer. We provide an extensive empirical study to address these issues. In addition, we propose a new interpretable intervention-based metric, to measure the quality of factors encoding in the representation. Our results indicate that some level of disentanglement, transferring a representation from synthetic to real data, is possible and effective.
Authors: Conghan Yue, Zhengwei Peng, Shiyan Du, Zhi Ji, Chuangjian Cai, Le Wan, Dongyu Zhang
Abstract: While many diffusion models perform well when controlling for particular aspect among style, character, and interaction, they struggle with fine-grained control due to dataset limitations and intricate model architecture design. This paper introduces a novel algorithm, Aggregation of Multiple Diffusion Models (AMDM), which synthesizes features from multiple diffusion models into a specified model, activating specific features for fine-grained control. Experimental results demonstrate that AMDM significantly improves fine-grained control without training, proving its effectiveness. Additionally, it reveals that diffusion models initially focus on features such as position, attributes, and style, with later stages improving generation quality and consistency. AMDM offers a new perspective for tackling the challenges of fine-grained conditional control generation in diffusion models: We can fully utilize existing or develop new conditional diffusion models that control specific aspects, and then aggregate them using AMDM algorithm. This eliminates the need for constructing complex datasets, designing intricate model architectures, and incurring high training costs. Code is available at: https://github.com/Hammour-steak/AMDM.
Authors: Sihyun Yu, Sangkyung Kwak, Huiwon Jang, Jongheon Jeong, Jonathan Huang, Jinwoo Shin, Saining Xie
Abstract: Recent studies have shown that the denoising process in (generative) diffusion models can induce meaningful (discriminative) representations inside the model, though the quality of these representations still lags behind those learned through recent self-supervised learning methods. We argue that one main bottleneck in training large-scale diffusion models for generation lies in effectively learning these representations. Moreover, training can be made easier by incorporating high-quality external visual representations, rather than relying solely on the diffusion models to learn them independently. We study this by introducing a straightforward regularization called REPresentation Alignment (REPA), which aligns the projections of noisy input hidden states in denoising networks with clean image representations obtained from external, pretrained visual encoders. The results are striking: our simple strategy yields significant improvements in both training efficiency and generation quality when applied to popular diffusion and flow-based transformers, such as DiTs and SiTs. For instance, our method can speed up SiT training by over 17.5$\times$, matching the performance (without classifier-free guidance) of a SiT-XL model trained for 7M steps in less than 400K steps. In terms of final generation quality, our approach achieves state-of-the-art results of FID=1.42 using classifier-free guidance with the guidance interval.
Authors: Daniel Gallo Fern\'andez, R\u{a}zvan-Andrei Mati\c{s}an, Alejandro Monroy Mu\~noz, Ana-Maria Vasilcoiu, Janusz Partyka, Tin Had\v{z}i Veljkovi\'c, Metod Jazbec
Abstract: Diffusion models have achieved unprecedented performance in image generation, yet they suffer from slow inference due to their iterative sampling process. To address this, early-exiting has recently been proposed, where the depth of the denoising network is made adaptive based on the (estimated) difficulty of each sampling step. Here, we discover an interesting "phase transition" in the sampling process of current adaptive diffusion models: the denoising network consistently exits early during the initial sampling steps, until it suddenly switches to utilizing the full network. Based on this, we propose accelerating generation by employing a shallower denoising network in the initial sampling steps and a deeper network in the later steps. We demonstrate empirically that our dual-backbone approach, DuoDiff, outperforms existing early-exit diffusion methods in both inference speed and generation quality. Importantly, DuoDiff is easy to implement and complementary to existing approaches for accelerating diffusion.
Authors: Daniel Gallo Fern\'andez, Robert van der Klis, R\u{a}zvan-Andrei Mati\c{s}an, Janusz Partyka, Efstratios Gavves, Samuele Papa, Phillip Lippe
Abstract: While vision transformers are able to solve a wide variety of computer vision tasks, no pre-training method has yet demonstrated the same scaling laws as observed in language models. Autoregressive models show promising results, but are commonly trained on images that are cropped or transformed into square images, which distorts or destroys information present in the input. To overcome this limitation, we propose NARAIM, a vision model pre-trained with an autoregressive objective that uses images in their native aspect ratio. By maintaining the native aspect ratio, we preserve the original spatial context, thereby enhancing the model's ability to interpret visual information. In our experiments, we show that maintaining the aspect ratio improves performance on a downstream classification task.
Authors: Long Le, Jason Xie, William Liang, Hung-Ju Wang, Yue Yang, Yecheng Jason Ma, Kyle Vedder, Arjun Krishna, Dinesh Jayaraman, Eric Eaton
Abstract: Interactive 3D simulated objects are crucial in AR/VR, animations, and robotics, driving immersive experiences and advanced automation. However, creating these articulated objects requires extensive human effort and expertise, limiting their broader applications. To overcome this challenge, we present Articulate-Anything, a system that automates the articulation of diverse, complex objects from many input modalities, including text, images, and videos. Articulate-Anything leverages vision-language models (VLMs) to generate code that can be compiled into an interactable digital twin for use in standard 3D simulators. Our system exploits existing 3D asset datasets via a mesh retrieval mechanism, along with an actor-critic system that iteratively proposes, evaluates, and refines solutions for articulating the objects, self-correcting errors to achieve a robust outcome. Qualitative evaluations demonstrate Articulate-Anything's capability to articulate complex and even ambiguous object affordances by leveraging rich grounded inputs. In extensive quantitative experiments on the standard PartNet-Mobility dataset, Articulate-Anything substantially outperforms prior work, increasing the success rate from 8.7-11.6% to 75% and setting a new bar for state-of-the-art performance. We further showcase the utility of our system by generating 3D assets from in-the-wild video inputs, which are then used to train robotic policies for fine-grained manipulation tasks in simulation that go beyond basic pick and place. These policies are then transferred to a real robotic system.
Authors: Juliette Marrie, Romain Menegaux, Michael Arbel, Diane Larlus, Julien Mairal
Abstract: We address the problem of extending the capabilities of vision foundation models such as DINO, SAM, and CLIP, to 3D tasks. Specifically, we introduce a novel method to uplift 2D image features into 3D Gaussian Splatting scenes. Unlike traditional approaches that rely on minimizing a reconstruction loss, our method employs a simpler and more efficient feature aggregation technique, augmented by a graph diffusion mechanism. Graph diffusion enriches features from a given model, such as CLIP, by leveraging pairwise similarities that encode 3D geometry or similarities induced by another embedding like DINOv2. Our approach achieves performance comparable to the state of the art on multiple downstream tasks while delivering significant speed-ups. Notably, we obtain competitive segmentation results using generic DINOv2 features, despite DINOv2 not being trained on millions of annotated segmentation masks like SAM. When applied to CLIP features, our method demonstrates strong performance in open-vocabulary, language-based object detection tasks, highlighting the versatility of our approach.
Authors: Pengcheng Shi, Shaocheng Yan, Yilin Xiao, Xinyi Liu, Yongjun Zhang, Jiayuan Li
Abstract: Correspondence-based point cloud registration (PCR) plays a key role in robotics and computer vision. However, challenges like sensor noises, object occlusions, and descriptor limitations inevitably result in numerous outliers. RANSAC family is the most popular outlier removal solution. However, the requisite iterations escalate exponentially with the outlier ratio, rendering it far inferior to existing methods (SC2PCR [1], MAC [2], etc.) in terms of accuracy or speed. Thus, we propose a two-stage consensus filtering (TCF) that elevates RANSAC to state-of-the-art (SOTA) speed and accuracy. Firstly, one-point RANSAC obtains a consensus set based on length consistency. Subsequently, two-point RANSAC refines the set via angle consistency. Then, three-point RANSAC computes a coarse pose and removes outliers based on transformed correspondence's distances. Drawing on optimizations from one-point and two-point RANSAC, three-point RANSAC requires only a few iterations. Eventually, an iterative reweighted least squares (IRLS) is applied to yield the optimal pose. Experiments on the large-scale KITTI and ETH datasets demonstrate our method achieves up to three-orders-of-magnitude speedup compared to MAC while maintaining registration accuracy and recall. Our code is available at https://github.com/ShiPC-AI/TCF.
Authors: Ryozo Masukawa, Sanggeon Yun, Yoshiki Yamaguchi, Mohsen Imani
Abstract: Video crime detection is a significant application of computer vision and artificial intelligence. However, existing datasets primarily focus on detecting severe crimes by analyzing entire video clips, often neglecting the precursor activities (i.e., privacy violations) that could potentially prevent these crimes. To address this limitation, we present PV-VTT (Privacy Violation Video To Text), a unique multimodal dataset aimed at identifying privacy violations. PV-VTT provides detailed annotations for both video and text in scenarios. To ensure the privacy of individuals in the videos, we only provide video feature vectors, avoiding the release of any raw video data. This privacy-focused approach allows researchers to use the dataset while protecting participant confidentiality. Recognizing that privacy violations are often ambiguous and context-dependent, we propose a Graph Neural Network (GNN)-based video description model. Our model generates a GNN-based prompt with image for Large Language Model (LLM), which deliver cost-effective and high-quality video descriptions. By leveraging a single video frame along with relevant text, our method reduces the number of input tokens required, maintaining descriptive quality while optimizing LLM API-usage. Extensive experiments validate the effectiveness and interpretability of our approach in video description tasks and flexibility of our PV-VTT dataset.
Authors: Xianghui Yang, Huiwen Shi, Bowen Zhang, Fan Yang, Jiacheng Wang, Hongxu Zhao, Xinhai Liu, Xinzhou Wang, Qingxiang Lin, Jiaao Yu, Lifu Wang, Zhuo Chen, Sicong Liu, 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: Koichi Namekata, Sherwin Bahmani, Ziyi Wu, Yash Kant, Igor Gilitschenski, David B. Lindell
Abstract: Methods for image-to-video generation have achieved impressive, photo-realistic quality. However, adjusting specific elements in generated videos, such as object motion or camera movement, is often a tedious process of trial and error, e.g., involving re-generating videos with different random seeds. Recent techniques address this issue by fine-tuning a pre-trained model to follow conditioning signals, such as bounding boxes or point trajectories. Yet, this fine-tuning procedure can be computationally expensive, and it requires datasets with annotated object motion, which can be difficult to procure. In this work, we introduce SG-I2V, a framework for controllable image-to-video generation that is self-guided$\unicode{x2013}$offering zero-shot control by relying solely on the knowledge present in a pre-trained image-to-video diffusion model without the need for fine-tuning or external knowledge. Our zero-shot method outperforms unsupervised baselines while significantly narrowing down the performance gap with supervised models in terms of visual quality and motion fidelity.
Authors: Ansh Shah, K Madhava Krishna
Abstract: Recovering metric depth from a single image remains a fundamental challenge in computer vision, requiring both scene understanding and accurate scaling. While deep learning has advanced monocular depth estimation, current models often struggle with unfamiliar scenes and layouts, particularly in zero-shot scenarios and when predicting scale-ergodic metric depth. We present MetricGold, a novel approach that harnesses generative diffusion model's rich priors to improve metric depth estimation. Building upon recent advances in MariGold, DDVM and Depth Anything V2 respectively, our method combines latent diffusion, log-scaled metric depth representation, and synthetic data training. MetricGold achieves efficient training on a single RTX 3090 within two days using photo-realistic synthetic data from HyperSIM, VirtualKitti, and TartanAir. Our experiments demonstrate robust generalization across diverse datasets, producing sharper and higher quality metric depth estimates compared to existing approaches.
Authors: Ruichuan An, Sihan Yang, Ming Lu, Kai Zeng, Yulin Luo, Ying Chen, Jiajun Cao, Hao Liang, Qi She, Shanghang Zhang, Wentao Zhang
Abstract: Current vision-language models (VLMs) show exceptional abilities across diverse tasks including visual question answering. To enhance user experience in practical applications, recent studies investigate VLM personalization to understand user-provided concepts. However, existing studies mainly focus on single-concept personalization, neglecting the existence and interplay of multiple concepts, which limits the real-world applicability of personalized VLMs. In this paper, we propose the first multi-concept personalization method named MC-LLaVA along with a high-quality multi-concept personalization dataset. Specifically, MC-LLaVA uses a joint training strategy incorporating multiple concepts in a single training step, allowing VLMs to perform accurately in multi-concept personalization. To reduce the cost of joint training, MC-LLaVA leverages visual token information for concept token initialization, yielding improved concept representation and accelerating joint training. To advance multi-concept personalization research, we further contribute a high-quality dataset. We carefully collect images from various movies that contain multiple characters and manually generate the multi-concept question-answer samples. Our dataset features diverse movie types and question-answer types. We conduct comprehensive qualitative and quantitative experiments to demonstrate that MC-LLaVA can achieve impressive multi-concept personalized responses, paving the way for VLMs to become better user-specific assistants. The code and dataset will be publicly available at https://github.com/arctanxarc/MC-LLaVA.
Authors: Jooyoung Choi, Chaehun Shin, Yeongtak Oh, Heeseung Kim, Sungroh Yoon
Abstract: Recent large-scale diffusion models generate high-quality images but struggle to learn new, personalized artistic styles, which limits the creation of unique style templates. Fine-tuning with reference images is the most promising approach, but it often blindly utilizes objectives and noise level distributions used for pre-training, leading to suboptimal style alignment. We propose the Style-friendly SNR sampler, which aggressively shifts the signal-to-noise ratio (SNR) distribution toward higher noise levels during fine-tuning to focus on noise levels where stylistic features emerge. This enables models to better capture unique styles and generate images with higher style alignment. Our method allows diffusion models to learn and share new "style templates", enhancing personalized content creation. We demonstrate the ability to generate styles such as personal watercolor paintings, minimal flat cartoons, 3D renderings, multi-panel images, and memes with text, thereby broadening the scope of style-driven generation.
Authors: Julian Bigge, Maite Ogueta, Luis Garcia, Benjamin Risse
Abstract: In this paper we propose the Hatching-Box, a novel imaging and analysis system to automatically monitor and quantify the developmental behavior of Drosophila in standard rearing vials and during regular rearing routines, rendering explicit experiments obsolete. This is achieved by combining custom tailored imaging hardware with dedicated detection and tracking algorithms, enabling the quantification of larvae, filled/empty pupae and flies over multiple days. Given the affordable and reproducible design of the Hatching-Box in combination with our generic client/server-based software, the system can easily be scaled to monitor an arbitrary amount of rearing vials simultaneously. We evaluated our system on a curated image dataset comprising nearly 470,000 annotated objects and performed several studies on real world experiments. We successfully reproduced results from well-established circadian experiments by comparing the eclosion periods of wild type flies to the clock mutants $\textit{per}^{short}$, $\textit{per}^{long}$ and $\textit{per}^0$ without involvement of any manual labor. Furthermore we show, that the Hatching-Box is able to extract additional information about group behavior as well as to reconstruct the whole life-cycle of the individual specimens. These results not only demonstrate the applicability of our system for long-term experiments but also indicate its benefits for automated monitoring in the general cultivation process.
Authors: Wuzheng Dong, Yujuan Zhu
Abstract: This paper presents a method for object recognition and automatic labeling in large-area remote sensing images called LRSAA. The method integrates YOLOv11 and MobileNetV3-SSD object detection algorithms through ensemble learning to enhance model performance. Furthermore, it employs Poisson disk sampling segmentation techniques and the EIOU metric to optimize the training and inference processes of segmented images, followed by the integration of results. This approach not only reduces the demand for computational resources but also achieves a good balance between accuracy and speed. The source code for this project has been made publicly available on https://github.com/anaerovane/LRSAA.
Authors: Shenghai Yuan, Jinfa Huang, Xianyi He, Yunyuan Ge, Yujun Shi, Liuhan Chen, Jiebo Luo, Li Yuan
Abstract: Identity-preserving text-to-video (IPT2V) generation aims to create high-fidelity videos with consistent human identity. It is an important task in video generation but remains an open problem for generative models. This paper pushes the technical frontier of IPT2V in two directions that have not been resolved in literature: (1) A tuning-free pipeline without tedious case-by-case finetuning, and (2) A frequency-aware heuristic identity-preserving DiT-based control scheme. We propose ConsisID, a tuning-free DiT-based controllable IPT2V model to keep human identity consistent in the generated video. Inspired by prior findings in frequency analysis of diffusion transformers, it employs identity-control signals in the frequency domain, where facial features can be decomposed into low-frequency global features and high-frequency intrinsic features. First, from a low-frequency perspective, we introduce a global facial extractor, which encodes reference images and facial key points into a latent space, generating features enriched with low-frequency information. These features are then integrated into shallow layers of the network to alleviate training challenges associated with DiT. Second, from a high-frequency perspective, we design a local facial extractor to capture high-frequency details and inject them into transformer blocks, enhancing the model's ability to preserve fine-grained features. We propose a hierarchical training strategy to leverage frequency information for identity preservation, transforming a vanilla pre-trained video generation model into an IPT2V model. Extensive experiments demonstrate that our frequency-aware heuristic scheme provides an optimal control solution for DiT-based models. Thanks to this scheme, our ConsisID generates high-quality, identity-preserving videos, making strides towards more effective IPT2V.
Authors: Rongchang Xie, Chen Du, Ping Song, Chang Liu
Abstract: We introduce MUSE-VL, a Unified Vision-Language Model through Semantic discrete Encoding for multimodal understanding and generation. Recently, the research community has begun exploring unified models for visual generation and understanding. However, existing vision tokenizers (e.g., VQGAN) only consider low-level information, which makes it difficult to align with texture semantic features. This results in high training complexity and necessitates a large amount of training data to achieve optimal performance. Additionally, their performance is still far from dedicated understanding models. This paper proposes Semantic Discrete Encoding (SDE), which effectively aligns the information of visual tokens and language tokens by adding semantic constraints to the visual tokenizer. This greatly reduces training difficulty and improves the performance of the unified model. The proposed model significantly surpasses the previous state-of-the-art in various vision-language benchmarks and achieves better performance than dedicated understanding models.
Authors: Xue Song, Jiequan Cui, Hanwang Zhang, Jiaxin Shi, Jingjing Chen, Chi Zhang, Yu-Gang Jiang
Abstract: In this paper, we propose the LoRA of Change (LoC) framework for image editing with visual instructions, i.e., before-after image pairs. Compared to the ambiguities, insufficient specificity, and diverse interpretations of natural language, visual instructions can accurately reflect users' intent. Building on the success of LoRA in text-based image editing and generation, we dynamically learn an instruction-specific LoRA to encode the "change" in a before-after image pair, enhancing the interpretability and reusability of our model. Furthermore, generalizable models for image editing with visual instructions typically require quad data, i.e., a before-after image pair, along with query and target images. Due to the scarcity of such quad data, existing models are limited to a narrow range of visual instructions. To overcome this limitation, we introduce the LoRA Reverse optimization technique, enabling large-scale training with paired data alone. Extensive qualitative and quantitative experiments demonstrate that our model produces high-quality images that align with user intent and support a broad spectrum of real-world visual instructions.
Authors: Chi Su, Xiaoxuan Ma, Jiajun Su, Yizhou Wang
Abstract: We propose a one-stage framework for real-time multi-person 3D human mesh estimation from a single RGB image. While current one-stage methods, which follow a DETR-style pipeline, achieve state-of-the-art (SOTA) performance with high-resolution inputs, we observe that this particularly benefits the estimation of individuals in smaller scales of the image (e.g., those far from the camera), but at the cost of significantly increased computation overhead. To address this, we introduce scale-adaptive tokens that are dynamically adjusted based on the relative scale of each individual in the image within the DETR framework. Specifically, individuals in smaller scales are processed at higher resolutions, larger ones at lower resolutions, and background regions are further distilled. These scale-adaptive tokens more efficiently encode the image features, facilitating subsequent decoding to regress the human mesh, while allowing the model to allocate computational resources more effectively and focus on more challenging cases. Experiments show that our method preserves the accuracy benefits of high-resolution processing while substantially reducing computational cost, achieving real-time inference with performance comparable to SOTA methods.
Authors: Simon Mielke, Anthony Stein
Abstract: Animal excretions in form of urine puddles and feces are a significant source of emissions in livestock farming. Automated detection of soiled floor in barns can contribute to improved management processes but also the derived information can be used to model emission dynamics. Previous research approaches to determine the puddle area require manual detection of the puddle in the barn. While humans can detect animal excretions on thermal images of a livestock barn, automated approaches using thresholds fail due to other objects of the same temperature, such as the animals themselves. In addition, various parameters such as the type of housing, animal species, age, sex, weather and unknown factors can influence the type and shape of excretions. Due to this heterogeneity, a method for automated detection of excretions must therefore be not only be accurate but also robust to varying conditions. These requirements can be met by using contemporary deep learning models from the field of artificial intelligence. This work is the first to investigate the suitability of different deep learning models for the detection of excretions in pigsties, thereby comparing established convolutional architectures with recent transformer-based approaches. The detection models Faster R-CNN, YOLOv8, DETR and DAB-DETR are compared and statistically assessed on two created training datasets representing two pig houses. We apply a method derived from nested cross-validation and report on the results in terms of eight common detection metrics. Our work demonstrates that all investigated deep learning models are generally suitable for reliably detecting excretions with an average precision of over 90%. The models also show robustness on out of distribution data that possesses differences from the conditions in the training data, however, with expected slight decreases in the overall detection performance.
Authors: Junli Deng, Yihao Luo
Abstract: Dynamic scene rendering has taken a leap forward with the rise of 4D Gaussian Splatting, but there's still one elusive challenge: how to make 3D Gaussians move through time as naturally as they would in the real world, all while keeping the motion smooth and consistent. In this paper, we unveil a fresh approach that blends state-space modeling with Wasserstein geometry, paving the way for a more fluid and coherent representation of dynamic scenes. We introduce a State Consistency Filter that merges prior predictions with the current observations, enabling Gaussians to stay true to their way over time. We also employ Wasserstein distance regularization to ensure smooth, consistent updates of Gaussian parameters, reducing motion artifacts. Lastly, we leverage Wasserstein geometry to capture both translational motion and shape deformations, creating a more physically plausible model for dynamic scenes. Our approach guides Gaussians along their natural way in the Wasserstein space, achieving smoother, more realistic motion and stronger temporal coherence. Experimental results show significant improvements in rendering quality and efficiency, outperforming current state-of-the-art techniques.
Authors: Jiahao Cui, Hui Li, Yun Zhan, Hanlin Shang, Kaihui Cheng, Yuqi Ma, Shan Mu, Hang Zhou, Jingdong Wang, Siyu Zhu
Abstract: Existing methodologies for animating portrait images face significant challenges, particularly in handling non-frontal perspectives, rendering dynamic objects around the portrait, and generating immersive, realistic backgrounds. In this paper, we introduce the first application of a pretrained transformer-based video generative model that demonstrates strong generalization capabilities and generates highly dynamic, realistic videos for portrait animation, effectively addressing these challenges. The adoption of a new video backbone model makes previous U-Net-based methods for identity maintenance, audio conditioning, and video extrapolation inapplicable. To address this limitation, we design an identity reference network consisting of a causal 3D VAE combined with a stacked series of transformer layers, ensuring consistent facial identity across video sequences. Additionally, we investigate various speech audio conditioning and motion frame mechanisms to enable the generation of continuous video driven by speech audio. Our method is validated through experiments on benchmark and newly proposed wild datasets, demonstrating substantial improvements over prior methods in generating realistic portraits characterized by diverse orientations within dynamic and immersive scenes. Further visualizations and the source code are available at: https://fudan-generative-vision.github.io/hallo3/.
Authors: Wooseok Jang, Youngjun Hong, Gunho Cha, Seungryong Kim
Abstract: Manipulation of facial images to meet specific controls such as pose, expression, and lighting, also known as face rigging, is a complex task in computer vision. Existing methods are limited by their reliance on image datasets, which necessitates individual-specific fine-tuning and limits their ability to retain fine-grained identity and semantic details, reducing practical usability. To overcome these limitations, we introduce ControlFace, a novel face rigging method conditioned on 3DMM renderings that enables flexible, high-fidelity control. We employ a dual-branch U-Nets: one, referred to as FaceNet, captures identity and fine details, while the other focuses on generation. To enhance control precision, the control mixer module encodes the correlated features between the target-aligned control and reference-aligned control, and a novel guidance method, reference control guidance, steers the generation process for better control adherence. By training on a facial video dataset, we fully utilize FaceNet's rich representations while ensuring control adherence. Extensive experiments demonstrate ControlFace's superior performance in identity preservation and control precision, highlighting its practicality. Please see the project website: https://cvlab-kaist.github.io/ControlFace/.
Authors: Jaskirat Singh, Lindsey Li, Weijia Shi, Ranjay Krishna, Yejin Choi, Pang Wei Koh, Michael F. Cohen, Stephen Gould, Liang Zheng, Luke Zettlemoyer
Abstract: Text-based adversarial guidance using a negative prompt has emerged as a widely adopted approach to steer diffusion models away from producing undesired concepts. While useful, performing adversarial guidance using text alone can be insufficient to capture complex visual concepts or avoid specific visual elements like copyrighted characters. In this paper, for the first time we explore an alternate modality in this direction by performing adversarial guidance directly using visual features from a reference image or other images in a batch. We introduce negative token merging (NegToMe), a simple but effective training-free approach which performs adversarial guidance through images by selectively pushing apart matching visual features between reference and generated images during the reverse diffusion process. By simply adjusting the used reference, NegToMe enables a diverse range of applications. Notably, when using other images in same batch as reference, we find that NegToMe significantly enhances output diversity (e.g., racial, gender, visual) by guiding features of each image away from others. Similarly, when used w.r.t. copyrighted reference images, NegToMe reduces visual similarity to copyrighted content by 34.57%. NegToMe is simple to implement using just few-lines of code, uses only marginally higher (<4%) inference time and is compatible with different diffusion architectures, including those like Flux, which don't natively support the use of a negative prompt. Code is available at https://negtome.github.io
Authors: Lan Wang, Yujia Chen, Du Tran, Vishnu Naresh Boddeti, Wen-Sheng Chu
Abstract: Long video understanding presents challenges due to the inherent high computational complexity and redundant temporal information. An effective representation for long videos must process such redundancy efficiently while preserving essential contents for downstream tasks. This paper introduces SEmantic Attention Learning (SEAL), a novel unified representation for long videos. To reduce computational complexity, long videos are decomposed into three distinct types of semantic entities: scenes, objects, and actions, allowing models to operate on a handful of entities rather than a large number of frames or pixels. To further address redundancy, we propose an attention learning module that balances token relevance with diversity formulated as a subset selection optimization problem. Our representation is versatile, enabling applications across various long video understanding tasks. Extensive experiments show that SEAL significantly outperforms state-of-the-art methods in video question answering and temporal grounding tasks and benchmarks including LVBench, MovieChat-1K, and Ego4D.
Authors: Anton Voronov, Denis Kuznedelev, Mikhail Khoroshikh, Valentin Khrulkov, Dmitry Baranchuk
Abstract: This work presents Switti, a scale-wise transformer for text-to-image generation. Starting from existing next-scale prediction AR models, we first explore them for T2I generation and propose architectural modifications to improve their convergence and overall performance. We then argue that scale-wise transformers do not require causality and propose a non-causal counterpart facilitating ~11% faster sampling and lower memory usage while also achieving slightly better generation quality. Furthermore, we reveal that classifier-free guidance at high-resolution scales is often unnecessary and can even degrade performance. By disabling guidance at these scales, we achieve an additional sampling acceleration of ~20% and improve the generation of fine-grained details. Extensive human preference studies and automated evaluations show that Switti outperforms existing T2I AR models and competes with state-of-the-art T2I diffusion models while being up to 7 times faster.
Authors: Zhibo Yang, Jun Tang, Zhaohai Li, Pengfei Wang, Jianqiang Wan, Humen Zhong, Xuejing Liu, Mingkun Yang, Peng Wang, Yuliang Liu, LianWen Jin, Xiang Bai, Shuai Bai, Junyang Lin
Abstract: Large Multimodal Models (LMMs) have demonstrated impressive performance on recognizing document images with natural language instructions. However, it remains unclear to what extent capabilities in literacy with rich structure and fine-grained visual challenges. The current landscape lacks a comprehensive benchmark to effectively measure the literate capabilities of LMMs. Existing benchmarks are often limited by narrow scenarios and specified tasks. To this end, we introduce CC-OCR, a comprehensive benchmark that possess a diverse range of scenarios, tasks, and challenges. CC-OCR comprises four OCR-centric tracks: multi-scene text reading, multilingual text reading, document parsing, and key information extraction. It includes 39 subsets with 7,058 full annotated images, of which 41% are sourced from real applications, being released for the first time. Furthermore, we evaluate nine prominent LMMs and reveal both the strengths and weaknesses of these models, particularly in text grounding, multi-orientation, and hallucination of repetition. CC-OCR aims to comprehensively evaluate the capabilities of LMMs on OCR-centered tasks, driving advancement in LMMs.
Authors: Kefan Chen, Chaerin Min, Linguang Zhang, Shreyas Hampali, Cem Keskin, Srinath Sridhar
Abstract: Despite remarkable progress in image generation models, generating realistic hands remains a persistent challenge due to their complex articulation, varying viewpoints, and frequent occlusions. We present FoundHand, a large-scale domain-specific diffusion model for synthesizing single and dual hand images. To train our model, we introduce FoundHand-10M, a large-scale hand dataset with 2D keypoints and segmentation mask annotations. Our insight is to use 2D hand keypoints as a universal representation that encodes both hand articulation and camera viewpoint. FoundHand learns from image pairs to capture physically plausible hand articulations, natively enables precise control through 2D keypoints, and supports appearance control. Our model exhibits core capabilities that include the ability to repose hands, transfer hand appearance, and even synthesize novel views. This leads to zero-shot capabilities for fixing malformed hands in previously generated images, or synthesizing hand video sequences. We present extensive experiments and evaluations that demonstrate state-of-the-art performance of our method.
Authors: Trung-Anh Dang, Vincent Nguyen, Ngoc-Son Vu, Christel Vrain
Abstract: Contrastive learning has significantly improved representation quality, enhancing knowledge transfer across tasks in continual learning (CL). However, catastrophic forgetting remains a key challenge, as contrastive based methods primarily focus on "soft relationships" or "softness" between samples, which shift with changing data distributions and lead to representation overlap across tasks. Recently, the newly identified Neural Collapse phenomenon has shown promise in CL by focusing on "hard relationships" or "hardness" between samples and fixed prototypes. However, this approach overlooks "softness", crucial for capturing intra-class variability, and this rigid focus can also pull old class representations toward current ones, increasing forgetting. Building on these insights, we propose Focal Neural Collapse Contrastive (FNC2), a novel representation learning loss that effectively balances both soft and hard relationships. Additionally, we introduce the Hardness-Softness Distillation (HSD) loss to progressively preserve the knowledge gained from these relationships across tasks. Our method outperforms state-of-the-art approaches, particularly in minimizing memory reliance. Remarkably, even without the use of memory, our approach rivals rehearsal-based methods, offering a compelling solution for data privacy concerns.
Authors: Tianyu Chang, Xiaohao Chen. Zhichao Wei, Xuanpu Zhang, Qing-Guo Chen, Weihua Luo, Xun Yang
Abstract: Video Virtual Try-on aims to fluently transfer the garment image to a semantically aligned try-on area in the source person video. Previous methods leveraged the inpainting mask to remove the original garment in the source video, thus achieving accurate garment transfer on simple model videos. However, when these methods are applied to realistic video data with more complex scene changes and posture movements, the overly large and incoherent agnostic masks will destroy the essential spatial-temporal information of the original video, thereby inhibiting the fidelity and coherence of the try-on video. To alleviate this problem, we propose a novel point-enhanced mask-free video virtual try-on framework (PEMF-VVTO). Specifically, we first leverage the pre-trained mask-based try-on model to construct large-scale paired training data (pseudo-person samples). Training on these mask-free data enables our model to perceive the original spatial-temporal information while realizing accurate garment transfer. Then, based on the pre-acquired sparse frame-cloth and frame-frame point alignments, we design the point-enhanced spatial attention (PSA) and point-enhanced temporal attention (PTA) to further improve the try-on accuracy and video coherence of the mask-free model. Concretely, PSA explicitly guides the garment transfer to desirable locations through the sparse semantic alignments of video frames and cloth. PTA exploits the temporal attention on sparse point correspondences to enhance the smoothness of generated videos. Extensive qualitative and quantitative experiments clearly illustrate that our PEMF-VVTO can generate more natural and coherent try-on videos than existing state-of-the-art methods.
Authors: Jiahao Lu, Tianyu Huang, Peng Li, Zhiyang Dou, Cheng Lin, Zhiming Cui, Zhen Dong, Sai-Kit Yeung, Wenping Wang, Yuan Liu
Abstract: Recent developments in monocular depth estimation methods enable high-quality depth estimation of single-view images but fail to estimate consistent video depth across different frames. Recent works address this problem by applying a video diffusion model to generate video depth conditioned on the input video, which is training-expensive and can only produce scale-invariant depth values without camera poses. In this paper, we propose a novel video-depth estimation method called Align3R to estimate temporal consistent depth maps for a dynamic video. Our key idea is to utilize the recent DUSt3R model to align estimated monocular depth maps of different timesteps. First, we fine-tune the DUSt3R model with additional estimated monocular depth as inputs for the dynamic scenes. Then, we apply optimization to reconstruct both depth maps and camera poses. Extensive experiments demonstrate that Align3R estimates consistent video depth and camera poses for a monocular video with superior performance than baseline methods.
Authors: Shuai Wang, Huiyan Kong, Baotian Li, Fa Zheng
Abstract: Effective defect detection is critical for ensuring the quality, functionality, and economic value of textile products. However, existing methods face challenges in achieving high accuracy, real-time performance, and efficient global information extraction. To address these issues, we propose Fab-ME, an advanced framework based on YOLOv8s, specifically designed for the accurate detection of 20 fabric defect types. Our contributions include the introduction of the cross-stage partial bottleneck with two convolutions (C2F) vision state-space (C2F-VMamba) module, which integrates visual state-space (VSS) blocks into the YOLOv8s feature fusion network neck, enhancing the capture of intricate details and global context while maintaining high processing speeds. Additionally, we incorporate an enhanced multi-scale channel attention (EMCA) module into the final layer of the feature extraction network, significantly improving sensitivity to small targets. Experimental results on the Tianchi fabric defect detection dataset demonstrate that Fab-ME achieves a 3.5% improvement in mAP@0.5 compared to the original YOLOv8s, validating its effectiveness for precise and efficient fabric defect detection.
Authors: Gin\'es Carreto Pic\'on, Illia Oleksiienko, Lukas Hedegaard, Arian Bakhtiarnia, Alexandros Iosifidis
Abstract: Transformers are widely used for their ability to capture data relations in sequence processing, with great success for a wide range of static tasks. However, the computational and memory footprint of their main component, i.e., the Scaled Dot-product Attention, is commonly overlooked. This makes their adoption in applications involving stream data processing with constraints in response latency, computational and memory resources infeasible. Some works have proposed methods to lower the computational cost of transformers, i.e. low-rank approximations, sparsity in attention, and efficient formulations for Continual Inference. In this paper, we introduce a new formulation of the Scaled Dot-product Attention based on the Nystr\"om approximation that is suitable for Continual Inference. In experiments on Online Audio Classification and Online Action Detection tasks, the proposed Continual Scaled Dot-product Attention can lower the number of operations by up to three orders of magnitude compared to the original Transformers while retaining the predictive performance of competing models.
Authors: Wangbo Zhao, Yizeng Han, Jiasheng Tang, Zhikai Li, Yibing Song, Kai Wang, Zhangyang Wang, Yang You
Abstract: Vision-language models (VLMs) have shown remarkable success across various multi-modal tasks, yet large VLMs encounter significant efficiency challenges due to processing numerous visual tokens. A promising approach to accelerating large VLM inference is using partial information, such as attention maps from specific layers, to assess token importance and prune less essential tokens. However, our study reveals three key insights: (i) Partial attention information is insufficient for accurately identifying critical visual tokens, resulting in suboptimal performance, especially at low token retention ratios; (ii) Global attention information, such as the attention map aggregated across all layers, more effectively preserves essential tokens and maintains comparable performance under aggressive pruning. However, the attention maps from all layers requires a full inference pass, which increases computational load and is therefore impractical in existing methods; and (iii) The global attention map aggregated from a small VLM closely resembles that of a large VLM, suggesting an efficient alternative. Based on these findings, we introduce a \textbf{training-free} method, \underline{\textbf{S}}mall VLM \underline{\textbf{G}}uidance for accelerating \underline{\textbf{L}}arge VLMs (\textbf{SGL}). Specifically, we employ the attention map aggregated from a small VLM to guide visual token pruning in a large VLM. Additionally, an early exiting mechanism is developed to fully use the small VLM's predictions, dynamically invoking the larger VLM only when necessary, yielding a superior trade-off between accuracy and computation. Extensive evaluations across 11 benchmarks demonstrate the effectiveness and generalizability of SGL, achieving up to 91\% pruning ratio for visual tokens while retaining competitive performance.
Authors: Zhaopeng Gu, Bingke Zhu, Guibo Zhu, Yingying Chen, Ming Tang, Jinqiao Wang
Abstract: Visual Anomaly Detection (VAD) aims to identify abnormal samples in images that deviate from normal patterns, covering multiple domains, including industrial, logical, and medical fields. Due to the domain gaps between these fields, existing VAD methods are typically tailored to each domain, with specialized detection techniques and model architectures that are difficult to generalize across different domains. Moreover, even within the same domain, current VAD approaches often follow a "one-category-one-model" paradigm, requiring large amounts of normal samples to train class-specific models, resulting in poor generalizability and hindering unified evaluation across domains. To address this issue, we propose a generalized few-shot VAD method, UniVAD, capable of detecting anomalies across various domains, such as industrial, logical, and medical anomalies, with a training-free unified model. UniVAD only needs few normal samples as references during testing to detect anomalies in previously unseen objects, without training on the specific domain. Specifically, UniVAD employs a Contextual Component Clustering ($C^3$) module based on clustering and vision foundation models to segment components within the image accurately, and leverages Component-Aware Patch Matching (CAPM) and Graph-Enhanced Component Modeling (GECM) modules to detect anomalies at different semantic levels, which are aggregated to produce the final detection result. We conduct experiments on nine datasets spanning industrial, logical, and medical fields, and the results demonstrate that UniVAD achieves state-of-the-art performance in few-shot anomaly detection tasks across multiple domains, outperforming domain-specific anomaly detection models. The code will be made publicly available.
Authors: Proma Hossain Progga, Md. Jobayer Rahman, Swapnil Biswas, Md. Shakil Ahmed, Arif Reza Anwary, Swakkhar Shatabda
Abstract: Gait recognition is a significant biometric technique for person identification, particularly in scenarios where other physiological biometrics are impractical or ineffective. In this paper, we address the challenges associated with gait recognition and present a novel approach to improve its accuracy and reliability. The proposed method leverages advanced techniques, including sequential gait landmarks obtained through the Mediapipe pose estimation model, Procrustes analysis for alignment, and a Siamese biGRU-dualStack Neural Network architecture for capturing temporal dependencies. Extensive experiments were conducted on large-scale cross-view datasets to demonstrate the effectiveness of the approach, achieving high recognition accuracy compared to other models. The model demonstrated accuracies of 95.7%, 94.44%, 87.71%, and 86.6% on CASIA-B, SZU RGB-D, OU-MVLP, and Gait3D datasets respectively. The results highlight the potential applications of the proposed method in various practical domains, indicating its significant contribution to the field of gait recognition.
Authors: Junqi Tang, Guixian Xu, Subhadip Mukherjee, Carola-Bibiane Sch\"onlieb
Abstract: We propose a new operator-sketching paradigm for designing efficient iterative data-driven reconstruction (IDR) schemes, e.g. Plug-and-Play algorithms and deep unrolling networks. These IDR schemes are currently the state-of-the-art solutions for imaging inverse problems. However, for high-dimensional imaging tasks, especially X-ray CT and MRI imaging, these IDR schemes typically become inefficient both in terms of computation, due to the need of computing multiple times the high-dimensional forward and adjoint operators. In this work, we explore and propose a universal dimensionality reduction framework for accelerating IDR schemes in solving imaging inverse problems, based on leveraging the sketching techniques from stochastic optimization. Using this framework, we derive a number of accelerated IDR schemes, such as the plug-and-play multi-stage sketched gradient (PnP-MS2G) and sketching-based primal-dual (LSPD and Sk-LSPD) deep unrolling networks. Meanwhile, for fully accelerating PnP schemes when the denoisers are computationally expensive, we provide novel stochastic lazy denoising schemes (Lazy-PnP and Lazy-PnP-EQ), leveraging the ProxSkip scheme in optimization and equivariant image denoisers, which can massively accelerate the PnP algorithms with improved practicality. We provide theoretical analysis for recovery guarantees of instances of the proposed framework. Our numerical experiments on natural image processing and tomographic image reconstruction demonstrate the remarkable effectiveness of our sketched IDR schemes.
Authors: Weiwen Zhang, Dawei Yang, Haoxuan Che, An Ran Ran, Carol Y. Cheung, Hao Chen
Abstract: For optical coherence tomography angiography (OCTA) images, a limited scanning rate leads to a trade-off between field-of-view (FOV) and imaging resolution. Although larger FOV images may reveal more parafoveal vascular lesions, their application is greatly hampered due to lower resolution. To increase the resolution, previous works only achieved satisfactory performance by using paired data for training, but real-world applications are limited by the challenge of collecting large-scale paired images. Thus, an unpaired approach is highly demanded. Generative Adversarial Network (GAN) has been commonly used in the unpaired setting, but it may struggle to accurately preserve fine-grained capillary details, which are critical biomarkers for OCTA. In this paper, our approach aspires to preserve these details by leveraging the frequency information, which represents details as high-frequencies ($\textbf{hf}$) and coarse-grained backgrounds as low-frequencies ($\textbf{lf}$). In general, we propose a GAN-based unpaired super-resolution method for OCTA images and exceptionally emphasize $\textbf{hf}$ fine capillaries through a dual-path generator. To facilitate a precise spectrum of the reconstructed image, we also propose a frequency-aware adversarial loss for the discriminator and introduce a frequency-aware focal consistency loss for end-to-end optimization. Experiments show that our method outperforms other state-of-the-art unpaired methods both quantitatively and visually.
Authors: Benjamin Salmon, Alexander Krull
Abstract: Accurate analysis of microscopy images is hindered by the presence of noise. This noise is usually signal-dependent and often additionally correlated along rows or columns of pixels. Current self- and unsupervised denoisers can address signal-dependent noise, but none can reliably remove noise that is also row- or column-correlated. Here, we present the first fully unsupervised deep learning-based denoiser capable of handling imaging noise that is row-correlated as well as signal-dependent. Our approach uses a Variational Autoencoder (VAE) with a specially designed autoregressive decoder. This decoder is capable of modeling row-correlated and signal-dependent noise but is incapable of independently modeling underlying clean signal. The VAE therefore produces latent variables containing only clean signal information, and these are mapped back into image space using a proposed second decoder network. Our method does not require a pre-trained noise model and can be trained from scratch using unpaired noisy data. We benchmark our approach on microscopy datatsets from a range of imaging modalities and sensor types, each with row- or column-correlated, signal-dependent noise, and show that it outperforms existing self- and unsupervised denoisers.
Authors: Guoyao Shen, Mengyu Li, Chad W. Farris, Stephan Anderson, Xin Zhang
Abstract: Deep learning-based MRI reconstruction models have achieved superior performance these days. Most recently, diffusion models have shown remarkable performance in image generation, in-painting, super-resolution, image editing and more. As a generalized diffusion model, cold diffusion further broadens the scope and considers models built around arbitrary image transformations such as blurring, down-sampling, etc. In this paper, we propose a k-space cold diffusion model that performs image degradation and restoration in k-space without the need for Gaussian noise. We provide comparisons with multiple deep learning-based MRI reconstruction models and perform tests on a well-known large open-source MRI dataset. Our results show that this novel way of performing degradation can generate high-quality reconstruction images for accelerated MRI.
Authors: Hikaru Shindo, Manuel Brack, Gopika Sudhakaran, Devendra Singh Dhami, Patrick Schramowski, Kristian Kersting
Abstract: Large-scale, pre-trained neural networks have demonstrated strong capabilities in various tasks, including zero-shot image segmentation. To identify concrete objects in complex scenes, humans instinctively rely on deictic descriptions in natural language, i.e., referring to something depending on the context such as "The object that is on the desk and behind the cup.". However, deep learning approaches cannot reliably interpret such deictic representations due to their lack of reasoning capabilities in complex scenarios. To remedy this issue, we propose DeiSAM -- a combination of large pre-trained neural networks with differentiable logic reasoners -- for deictic promptable segmentation. Given a complex, textual segmentation description, DeiSAM leverages Large Language Models (LLMs) to generate first-order logic rules and performs differentiable forward reasoning on generated scene graphs. Subsequently, DeiSAM segments objects by matching them to the logically inferred image regions. As part of our evaluation, we propose the Deictic Visual Genome (DeiVG) dataset, containing paired visual input and complex, deictic textual prompts. Our empirical results demonstrate that DeiSAM is a substantial improvement over purely data-driven baselines for deictic promptable segmentation.
Authors: Kung-Hsiang Huang, Hou Pong Chan, Yi R. Fung, Haoyi Qiu, Mingyang Zhou, Shafiq Joty, Shih-Fu Chang, Heng Ji
Abstract: Data visualization in the form of charts plays a pivotal role in data analysis, offering critical insights and aiding in informed decision-making. Automatic chart understanding has witnessed significant advancements with the rise of large foundation models in recent years. Foundation models, such as large language models, have revolutionized various natural language processing tasks and are increasingly being applied to chart understanding tasks. This survey paper provides a comprehensive overview of the recent developments, challenges, and future directions in chart understanding within the context of these foundation models. We review fundamental building blocks crucial for studying chart understanding tasks. Additionally, we explore various tasks and their evaluation metrics and sources of both charts and textual inputs. Various modeling strategies are then examined, encompassing both classification-based and generation-based approaches, along with tool augmentation techniques that enhance chart understanding performance. Furthermore, we discuss the state-of-the-art performance of each task and discuss how we can improve the performance. Challenges and future directions are addressed, highlighting the importance of several topics, such as domain-specific charts, lack of efforts in developing evaluation metrics, and agent-oriented settings. This survey paper serves as a comprehensive resource for researchers and practitioners in the fields of natural language processing, computer vision, and data analysis, providing valuable insights and directions for future research in chart understanding leveraging large foundation models. The studies mentioned in this paper, along with emerging new research, will be continually updated at: https://github.com/khuangaf/Awesome-Chart-Understanding.
URLs: https://github.com/khuangaf/Awesome-Chart-Understanding.
Authors: Reza Esfandiarpoor, Cristina Menghini, Stephen H. Bach
Abstract: Recent works often assume that Vision-Language Model (VLM) representations are based on visual attributes like shape. However, it is unclear to what extent VLMs prioritize this information to represent concepts. We propose Extract and Explore (EX2), a novel approach to characterize textual features that are important for VLMs. EX2 uses reinforcement learning to align a large language model with VLM preferences and generates descriptions that incorporate features that are important for the VLM. Then, we inspect the descriptions to identify features that contribute to VLM representations. Using EX2, we find that spurious descriptions have a major role in VLM representations despite providing no helpful information, e.g., Click to enlarge photo of CONCEPT. More importantly, among informative descriptions, VLMs rely significantly on non-visual attributes like habitat (e.g., North America) to represent visual concepts. Also, our analysis reveals that different VLMs prioritize different attributes in their representations. Overall, we show that VLMs do not simply match images to scene descriptions and that non-visual or even spurious descriptions significantly influence their representations.
Authors: Omer Sahin Tas, Royden Wagner
Abstract: Transformer-based models generate hidden states that are difficult to interpret. In this work, we aim to interpret these hidden states and control them at inference, with a focus on motion forecasting. We use linear probes to measure neural collapse towards interpretable motion features in hidden states. High probing accuracy implies meaningful directions and distances between hidden states of opposing features, which we use to fit interpretable control vectors for activation steering at inference. To optimize our control vectors, we use sparse autoencoders with fully-connected, convolutional, MLPMixer layers and various activation functions. Notably, we show that enforcing sparsity in hidden states leads to a more linear relationship between control vector temperatures and forecasts. Our approach enables mechanistic interpretability and zero-shot generalization to unseen dataset characteristics with negligible computational overhead. Our implementation is available at https://github.com/kit-mrt/future-motion
Authors: Vito Paolo Pastore, Massimiliano Ciranni, Davide Marinelli, Francesca Odone, Vittorio Murino
Abstract: It is widely recognized that deep neural networks are sensitive to bias in the data. This means that during training these models are likely to learn spurious correlations between data and labels, resulting in limited generalization abilities and low performance. In this context, model debiasing approaches can be devised aiming at reducing the model's dependency on such unwanted correlations, either leveraging the knowledge of bias information or not. In this work, we focus on the latter and more realistic scenario, showing the importance of accurately predicting the bias-conflicting and bias-aligned samples to obtain compelling performance in bias mitigation. On this ground, we propose to conceive the problem of model bias from an out-of-distribution perspective, introducing a new bias identification method based on anomaly detection. We claim that when data is mostly biased, bias-conflicting samples can be regarded as outliers with respect to the bias-aligned distribution in the feature space of a biased model, thus allowing for precisely detecting them with an anomaly detection method. Coupling the proposed bias identification approach with bias-conflicting data upsampling and augmentation in a two-step strategy, we reach state-of-the-art performance on synthetic and real benchmark datasets. Ultimately, our proposed approach shows that the data bias issue does not necessarily require complex debiasing methods, given that an accurate bias identification procedure is defined. Source code is available at https://github.com/Malga-Vision/MoDAD
Authors: Junming Wang, Xiuxian Guan, Zekai Sun, Tianxiang Shen, Dong Huang, Fangming Liu, Heming Cui
Abstract: Air-ground robots (AGRs) are widely used in surveillance and disaster response due to their exceptional mobility and versatility (i.e., flying and driving). Current AGR navigation systems perform well in static occlusion-prone environments (e.g., indoors) by using 3D semantic occupancy networks to predict occlusions for complete local mapping and then computing Euclidean Signed Distance Field (ESDF) for path planning. However, these systems face challenges in dynamic, severe occlusion scenes (e.g., crowds) due to limitations in perception networks' low prediction accuracy and path planners' high computation overhead. In this paper, we propose OMEGA, which contains OccMamba with an Efficient AGR-Planner to address the above-mentioned problems. OccMamba adopts a novel architecture that separates semantic and occupancy prediction into independent branches, incorporating two mamba blocks within these branches. These blocks efficiently extract semantic and geometric features in 3D environments with linear complexity, ensuring that the network can learn long-distance dependencies to improve prediction accuracy. Semantic and geometric features are combined within the Bird's Eye View (BEV) space to minimise computational overhead during feature fusion. The resulting semantic occupancy map is then seamlessly integrated into the local map, providing occlusion awareness of the dynamic environment. Our AGR-Planner utilizes this local map and employs kinodynamic A* search and gradient-based trajectory optimization to guarantee planning is ESDF-free and energy-efficient. Extensive experiments demonstrate that OccMamba outperforms the state-of-the-art 3D semantic occupancy network with 25.0% mIoU. End-to-end navigation experiments in dynamic scenes verify OMEGA's efficiency, achieving a 96% average planning success rate. Code and video are available at https://jmwang0117.github.io/OMEGA/.
Authors: Suwichaya Suwanwimolkul, Natanon Tongamrak, Nuttamon Thungka, Naebboon Hoonchareon, Jitkomut Songsiri
Abstract: This paper presents an online platform showing Thailand solar irradiance map every 30 minutes, available at https://www.cusolarforecast.com. The methodology for estimating global horizontal irradiance (GHI) across Thailand relies on cloud index extracted from Himawari-8 satellite imagery, Ineichen clear-sky model with locally-tuned Linke turbidity, and machine learning models. The methods take clear-sky irradiance, cloud index, re-analyzed GHI and temperature data from the MERRA-2 database, and date-time as inputs for GHI estimation models, including LightGBM, LSTM, Informer, and Transformer. These are benchmarked with the estimate from a commercial service X by evaluation of 15-minute ground GHI data from 53 ground stations over 1.5 years during 2022-2023. The results show that the four models exhibit comparable overall MAE performance to the service X. The best model is LightGBM with an overall MAE of 78.58 W/sqm and RMSE of 118.97 W/sqm, while the service X achieves the lowest MAE, RMSE, and MBE in cloudy condition. Obtaining re-analyzed MERRA-2 data for the whole Thailand region is not economically feasible for deployment. When removing these features, the Informer model has a winning performance in MAE of 78.67 W/sqm. The obtained performance aligns with existing literature by taking the climate zone and time granularity of data into consideration. As the map shows an estimate of GHI over 93,000 grids with a frequent update, the paper also describes a computational framework for displaying the entire map. It tests the runtime performance of deep learning models in the GHI estimation process.
Authors: Eliahu Horwitz, Bar Cavia, Jonathan Kahana, Yedid Hoshen
Abstract: The increasing availability of public models begs the question: can we train neural networks that use other networks as input? Such models allow us to study different aspects of a given neural network, for example, determining the categories in a model's training dataset. However, machine learning on model weights is challenging as they often exhibit significant variation unrelated to the models' semantic properties (nuisance variation). Here, we identify a key property of real-world models: most public models belong to a small set of Model Trees, where all models within a tree are fine-tuned from a common ancestor (e.g., a foundation model). Importantly, we find that within each tree there is less nuisance variation between models. Concretely, while learning across Model Trees requires complex architectures, even a linear classifier trained on a single model layer often works within trees. While effective, these linear classifiers are computationally expensive, especially when dealing with larger models that have many parameters. To address this, we introduce Probing Experts (ProbeX), a theoretically motivated and lightweight method. Notably, ProbeX is the first probing method specifically designed to learn from the weights of a single hidden model layer. We demonstrate the effectiveness of ProbeX by predicting the categories in a model's training dataset based only on its weights. Excitingly, ProbeX can also map the weights of Stable Diffusion into a shared weight-language embedding space, enabling zero-shot model classification.
Authors: Wen Jiang, Boshu Lei, Katrina Ashton, Kostas Daniilidis
Abstract: We present an active mapping system that could plan for long-horizon exploration goals and short-term actions with a 3D Gaussian Splatting (3DGS) representation. Existing methods either did not take advantage of recent developments in multimodal Large Language Models (LLM) or did not consider challenges in localization uncertainty, which is critical in embodied agents. We propose employing multimodal LLMs for long-horizon planning in conjunction with detailed motion planning using our information-based algorithm. By leveraging high-quality view synthesis from our 3DGS representation, our method employs a multimodal LLM as a zero-shot planner for long-horizon exploration goals from the semantic perspective. We also introduce an uncertainty-aware path proposal and selection algorithm that balances the dual objectives of maximizing the information gain for the environment while minimizing the cost of localization errors. Experiments conducted on the Gibson and Habitat-Matterport 3D datasets demonstrate state-of-the-art results of the proposed method.
Authors: Abdulkadir Gokce, Martin Schrimpf
Abstract: When trained on large-scale object classification datasets, certain artificial neural network models begin to approximate core object recognition (COR) behaviors and neural response patterns in the primate visual ventral stream (VVS). While recent machine learning advances suggest that scaling model size, dataset size, and compute resources improve task performance, the impact of scaling on brain alignment remains unclear. In this study, we explore scaling laws for modeling the primate VVS by systematically evaluating over 600 models trained under controlled conditions on benchmarks spanning V1, V2, V4, IT and COR behaviors. We observe that while behavioral alignment continues to scale with larger models, neural alignment saturates. This observation remains true across model architectures and training datasets, even though models with stronger inductive bias and datasets with higher-quality images are more compute-efficient. Increased scaling is especially beneficial for higher-level visual areas, where small models trained on few samples exhibit only poor alignment. Finally, we develop a scaling recipe, indicating that a greater proportion of compute should be allocated to data samples over model size. Our results suggest that while scaling alone might suffice for alignment with human core object recognition behavior, it will not yield improved models of the brain's visual ventral stream with current architectures and datasets, highlighting the need for novel strategies in building brain-like models.
Authors: Ching-Yi Wang
Abstract: Multi-modal large language models (MLLMs), such as GPT-4o, excel at integrating text and visual data but face systematic challenges when interpreting ambiguous or incomplete visual stimuli [9]. This study leverages statistical modeling to analyze the factors driving these errors, using a dataset of geometric stimuli characterized by features like 3D, rotation, and missing face/side. We applied parametric methods, non-parametric methods, and ensemble techniques to predict classification errors, with the non-linear gradient boosting model achieving the highest performance (AUC=0.85) during cross-validation. Feature importance analysis highlighted difficulties in depth perception and reconstructing incomplete structures as key contributors to misclassification. These findings demonstrate the effectiveness of statistical approaches for uncovering limitations in MLLMs and offer actionable insights for enhancing model architectures by integrating contextual reasoning mechanisms.