Authors: Jianyang Gu, Samuel Stevens, Elizabeth G Campolongo, Matthew J Thompson, Net Zhang, Jiaman Wu, Andrei Kopanev, Zheda Mai, Alexander E. White, James Balhoff, Wasila Dahdul, Daniel Rubenstein, Hilmar Lapp, Tanya Berger-Wolf, Wei-Lun Chao, Yu Su
Abstract: Foundation models trained at scale exhibit remarkable emergent behaviors, learning new capabilities beyond their initial training objectives. We find such emergent behaviors in biological vision models via large-scale contrastive vision-language training. To achieve this, we first curate TreeOfLife-200M, comprising 214 million images of living organisms, the largest and most diverse biological organism image dataset to date. We then train BioCLIP 2 on TreeOfLife-200M to distinguish different species. Despite the narrow training objective, BioCLIP 2 yields extraordinary accuracy when applied to various biological visual tasks such as habitat classification and trait prediction. We identify emergent properties in the learned embedding space of BioCLIP 2. At the inter-species level, the embedding distribution of different species aligns closely with functional and ecological meanings (e.g., beak sizes and habitats). At the intra-species level, instead of being diminished, the intra-species variations (e.g., life stages and sexes) are preserved and better separated in subspaces orthogonal to inter-species distinctions. We provide formal proof and analyses to explain why hierarchical supervision and contrastive objectives encourage these emergent properties. Crucially, our results reveal that these properties become increasingly significant with larger-scale training data, leading to a biologically meaningful embedding space.
Authors: Bowei Chen, Brian Curless, Ira Kemelmacher-Shlizerman, Steven M. Seitz
Abstract: Self-captured full-body videos are popular, but most deployments require mounted cameras, carefully-framed shots, and repeated practice. We propose a more convenient solution that enables full-body video capture using handheld mobile devices. Our approach takes as input two static photos (front and back) of you in a mirror, along with an IMU motion reference that you perform while holding your mobile phone, and synthesizes a realistic video of you performing a similar target motion. We enable rendering into a new scene, with consistent illumination and shadows. We propose a novel video diffusion-based model to achieve this. Specifically, we propose a parameter-free frame generation strategy, as well as a multi-reference attention mechanism, that effectively integrate appearance information from both the front and back selfies into the video diffusion model. Additionally, we introduce an image-based fine-tuning strategy to enhance frame sharpness and improve the generation of shadows and reflections, achieving a more realistic human-scene composition.
Authors: Amirhossein Almohammadi, Aryan Mikaeili, Sauradip Nag, Negar Hassanpour, Andrea Tagliasacchi, Ali Mahdavi-Amiri
Abstract: Image editing is an important task in computer graphics, vision, and VFX, with recent diffusion-based methods achieving fast and high-quality results. However, edits requiring significant structural changes, such as non-rigid deformations, object modifications, or content generation, remain challenging. Existing few step editing approaches produce artifacts such as irrelevant texture or struggle to preserve key attributes of the source image (e.g., pose). We introduce Cora, a novel editing framework that addresses these limitations by introducing correspondence-aware noise correction and interpolated attention maps. Our method aligns textures and structures between the source and target images through semantic correspondence, enabling accurate texture transfer while generating new content when necessary. Cora offers control over the balance between content generation and preservation. Extensive experiments demonstrate that, quantitatively and qualitatively, Cora excels in maintaining structure, textures, and identity across diverse edits, including pose changes, object addition, and texture refinements. User studies confirm that Cora delivers superior results, outperforming alternatives.
Authors: Neehar Kondapaneni, Oisin Mac Aodha, Pietro Perona
Abstract: We propose a method for discovering and visualizing the differences between two learned representations, enabling more direct and interpretable model comparisons. We validate our method, which we call Representational Differences Explanations (RDX), by using it to compare models with known conceptual differences and demonstrate that it recovers meaningful distinctions where existing explainable AI (XAI) techniques fail. Applied to state-of-the-art models on challenging subsets of the ImageNet and iNaturalist datasets, RDX reveals both insightful representational differences and subtle patterns in the data. Although comparison is a cornerstone of scientific analysis, current tools in machine learning, namely post hoc XAI methods, struggle to support model comparison effectively. Our work addresses this gap by introducing an effective and explainable tool for contrasting model representations.
Authors: David Ma, Huaqing Yuan, Xingjian Wang, Qianbo Zang, Tianci Liu, Xinyang He, Yanbin Wei, Jiawei Guo, Ni Jiahui, Zhenzhu Yang, Meng Cao, Shanghaoran Quan, Yizhi Li, Wangchunshu Zhou, Jiaheng Liu, Wenhao Huang, Ge Zhang, Shiwen Ni, Xiaojie Jin
Abstract: Although long-video understanding demands that models capture hierarchical temporal information -- from clip (seconds) and shot (tens of seconds) to event (minutes) and story (hours) -- existing benchmarks either neglect this multi-scale design or scatter scale-specific questions across different videos, preventing direct comparison of model performance across timescales on the same content. To address this, we introduce ScaleLong, the first benchmark to disentangle these factors by embedding questions targeting four hierarchical timescales -- clip (seconds), shot (tens of seconds), event (minutes), and story (hours) -- all within the same video content. This within-content multi-timescale questioning design enables direct comparison of model performance across timescales on identical videos. ScaleLong features 269 long videos (avg.\ 86\,min) from 5 main categories and 36 sub-categories, with 4--8 carefully designed questions, including at least one question for each timescale. Evaluating 23 MLLMs reveals a U-shaped performance curve, with higher accuracy at the shortest and longest timescales and a dip at intermediate levels. Furthermore, ablation studies show that increased visual token capacity consistently enhances reasoning across all timescales. ScaleLong offers a fine-grained, multi-timescale benchmark for advancing MLLM capabilities in long-video understanding. The code and dataset are available https://github.com/multimodal-art-projection/ScaleLong.
URLs: https://github.com/multimodal-art-projection/ScaleLong.
Authors: Xuweiyi Chen, Wentao Zhou, Aruni RoyChowdhury, Zezhou Cheng
Abstract: While scaling laws have transformed natural language processing and computer vision, 3D point cloud understanding has yet to reach that stage. This can be attributed to both the comparatively smaller scale of 3D datasets, as well as the disparate sources of the data itself. Point clouds are captured by diverse sensors (e.g., depth cameras, LiDAR) across varied domains (e.g., indoor, outdoor), each introducing unique scanning patterns, sampling densities, and semantic biases. Such domain heterogeneity poses a major barrier towards training unified models at scale, especially under the realistic constraint that domain labels are typically inaccessible at inference time. In this work, we propose Point-MoE, a Mixture-of-Experts architecture designed to enable large-scale, cross-domain generalization in 3D perception. We show that standard point cloud backbones degrade significantly in performance when trained on mixed-domain data, whereas Point-MoE with a simple top-k routing strategy can automatically specialize experts, even without access to domain labels. Our experiments demonstrate that Point-MoE not only outperforms strong multi-domain baselines but also generalizes better to unseen domains. This work highlights a scalable path forward for 3D understanding: letting the model discover structure in diverse 3D data, rather than imposing it via manual curation or domain supervision.
Authors: Adriano Fragomeni, Dima Damen, Michael Wray
Abstract: Text-to-Video (T2V) retrieval aims to identify the most relevant item from a gallery of videos based on a user's text query. Traditional methods rely solely on aligning video and text modalities to compute the similarity and retrieve relevant items. However, recent advancements emphasise incorporating auxiliary information extracted from video and text modalities to improve retrieval performance and bridge the semantic gap between these modalities. Auxiliary information can include visual attributes, such as objects; temporal and spatial context; and textual descriptions, such as speech and rephrased captions. This survey comprehensively reviews 81 research papers on Text-to-Video retrieval that utilise such auxiliary information. It provides a detailed analysis of their methodologies; highlights state-of-the-art results on benchmark datasets; and discusses available datasets and their auxiliary information. Additionally, it proposes promising directions for future research, focusing on different ways to further enhance retrieval performance using this information.
Authors: Rafi Hassan Chowdhury, Sabbir Ahmed
Abstract: Ensuring food safety is critical due to its profound impact on public health, economic stability, and global supply chains. Cultivation of Mango, a major agricultural product in several South Asian countries, faces high financial losses due to different diseases, affecting various aspects of the entire supply chain. While deep learning-based methods have been explored for mango leaf disease classification, there remains a gap in designing solutions that are computationally efficient and compatible with low-end devices. In this work, we propose a lightweight Vision Transformer-based pipeline with a self-attention mechanism to classify mango leaf diseases, achieving state-of-the-art performance with minimal computational overhead. Our approach leverages global attention to capture intricate patterns among disease types and incorporates runtime augmentation for enhanced performance. Evaluation on the MangoLeafBD dataset demonstrates a 99.43% accuracy, outperforming existing methods in terms of model size, parameter count, and FLOPs count.
Authors: Yichen Feng, Zhangchen Xu, Fengqing Jiang, Yuetai Li, Bhaskar Ramasubramanian, Luyao Niu, Bill Yuchen Lin, Radha Poovendran
Abstract: Vision language models (VLMs) are expected to perform effective multimodal reasoning and make logically coherent decisions, which is critical to tasks such as diagram understanding and spatial problem solving. However, current VLM reasoning lacks large-scale and well-structured training datasets. To bridge this gap, we propose VisualSphinx, a first-of-its-kind large-scale synthetic visual logical reasoning training data. To tackle the challenge of image synthesis with grounding answers, we propose a rule-to-image synthesis pipeline, which extracts and expands puzzle rules from seed questions and generates the code of grounding synthesis image synthesis for puzzle sample assembly. Experiments demonstrate that VLM trained using GRPO on VisualSphinx benefit from logical coherence and readability of our dataset and exhibit improved performance on logical reasoning tasks. The enhanced reasoning capabilities developed from VisualSphinx also benefit other reasoning tasks such as algebraic reasoning, arithmetic reasoning and geometry reasoning.
Authors: Bayu Adhi Tama, Mansa Krishna, Homayra Alam, Mostafa Cham, Omar Faruque, Gong Cheng, Jianwu Wang, Mathieu Morlighem, Vandana Janeja
Abstract: Understanding Greenland's subglacial topography is critical for projecting the future mass loss of the ice sheet and its contribution to global sea-level rise. However, the complex and sparse nature of observational data, particularly information about the bed topography under the ice sheet, significantly increases the uncertainty in model projections. Bed topography is traditionally measured by airborne ice-penetrating radar that measures the ice thickness directly underneath the aircraft, leaving data gap of tens of kilometers in between flight lines. This study introduces a deep learning framework, which we call as DeepTopoNet, that integrates radar-derived ice thickness observations and BedMachine Greenland data through a novel dynamic loss-balancing mechanism. Among all efforts to reconstruct bed topography, BedMachine has emerged as one of the most widely used datasets, combining mass conservation principles and ice thickness measurements to generate high-resolution bed elevation estimates. The proposed loss function adaptively adjusts the weighting between radar and BedMachine data, ensuring robustness in areas with limited radar coverage while leveraging the high spatial resolution of BedMachine predictions i.e. bed estimates. Our approach incorporates gradient-based and trend surface features to enhance model performance and utilizes a CNN architecture designed for subgrid-scale predictions. By systematically testing on the Upernavik Isstr{\o}m) region, the model achieves high accuracy, outperforming baseline methods in reconstructing subglacial terrain. This work demonstrates the potential of deep learning in bridging observational gaps, providing a scalable and efficient solution to inferring subglacial topography.
Authors: Vaishnav Ramesh, Junliang Liu, Haining Wang, Md Jahidul Islam
Abstract: A long-held challenge in no-reference image quality assessment (NR-IQA) learning from human subjective perception is the lack of objective generalization to unseen natural distortions. To address this, we integrate a novel Depth-Guided cross-attention and refinement (Depth-CAR) mechanism, which distills scene depth and spatial features into a structure-aware representation for improved NR-IQA. This brings in the knowledge of object saliency and relative contrast of the scene for more discriminative feature learning. Additionally, we introduce the idea of TCB (Transformer-CNN Bridge) to fuse high-level global contextual dependencies from a transformer backbone with local spatial features captured by a set of hierarchical CNN (convolutional neural network) layers. We implement TCB and Depth-CAR as multimodal attention-based projection functions to select the most informative features, which also improve training time and inference efficiency. Experimental results demonstrate that our proposed DGIQA model achieves state-of-the-art (SOTA) performance on both synthetic and authentic benchmark datasets. More importantly, DGIQA outperforms SOTA models on cross-dataset evaluations as well as in assessing natural image distortions such as low-light effects, hazy conditions, and lens flares.
Authors: Nokimul Hasan Arif, Shadman Rabby, Md Hefzul Hossain Papon, Sabbir Ahmed
Abstract: Visual hallucinations in Large Language Models (LLMs), where the model generates responses that are inconsistent with the visual input, pose a significant challenge to their reliability, particularly in contexts where precise and trustworthy outputs are critical. Current research largely emphasizes post-hoc correction or model-specific fine-tuning strategies, with limited exploration of preprocessing techniques to address hallucination issues at the input stage. This study presents a novel ensemble-based preprocessing framework that adaptively selects the most appropriate filtering approach -- noise reduced (NR), edge enhanced (EE), or unaltered input (org) based on the type of question posed, resulting into reduced hallucination without requiring any modifications to the underlying model architecture or training pipeline. Evaluated on the `HaloQuest' dataset -- a benchmark designed to test multimodal reasoning on visually complex inputs, our method achieves a 44.3% reduction in hallucination rates, as measured by Natural Language Inference (NLI) scores using SelfCheckGPT. This demonstrates that intelligent input conditioning alone can significantly enhance factual grounding in LLM responses. The findings highlight the importance of adaptive preprocessing techniques in mitigating hallucinations, paving the way for more reliable multimodal systems capable of addressing real-world challenges.
Authors: Sangwon Jung, Alex Oesterling, Claudio Mayrink Verdun, Sajani Vithana, Taesup Moon, Flavio P. Calmon
Abstract: Text-to-image (T2I) generative models can create vivid, realistic images from textual descriptions. As these models proliferate, they expose new concerns about their ability to represent diverse demographic groups, propagate stereotypes, and efface minority populations. Despite growing attention to the "safe" and "responsible" design of artificial intelligence (AI), there is no established methodology to systematically measure and control representational harms in image generation. This paper introduces a novel framework to measure the representation of intersectional groups in images generated by T2I models by applying the Multi-Group Proportional Representation (MPR) metric. MPR evaluates the worst-case deviation of representation statistics across given population groups in images produced by a generative model, allowing for flexible and context-specific measurements based on user requirements. We also develop an algorithm to optimize T2I models for this metric. Through experiments, we demonstrate that MPR can effectively measure representation statistics across multiple intersectional groups and, when used as a training objective, can guide models toward a more balanced generation across demographic groups while maintaining generation quality.
Authors: Chenbin Pan, Wenbin He, Zhengzhong Tu, Liu Ren
Abstract: The recent explosive interest in the reasoning capabilities of large language models, such as DeepSeek-R1, has demonstrated remarkable success through reinforcement learning-based fine-tuning frameworks, exemplified by methods like Group Relative Policy Optimization (GRPO). However, such reasoning abilities remain underexplored and notably absent in vision foundation models, including representation models like the DINO series. In this work, we propose \textbf{DINO-R1}, the first such attempt to incentivize visual in-context reasoning capabilities of vision foundation models using reinforcement learning. Specifically, DINO-R1 introduces \textbf{Group Relative Query Optimization (GRQO)}, a novel reinforcement-style training strategy explicitly designed for query-based representation models, which computes query-level rewards based on group-normalized alignment quality. We also apply KL-regularization to stabilize the objectness distribution to reduce the training instability. This joint optimization enables dense and expressive supervision across queries while mitigating overfitting and distributional drift. Building upon Grounding-DINO, we train a series of DINO-R1 family models that integrate a visual prompt encoder and a visual-guided query selection mechanism. Extensive experiments on COCO, LVIS, and ODinW demonstrate that DINO-R1 significantly outperforms supervised fine-tuning baselines, achieving strong generalization in both open-vocabulary and closed-set visual prompting scenarios.
Authors: Numair Nadeem, Muhammad Hamza Asad, Saeed Anwar, Abdul Bais
Abstract: Semantic segmentation of crops and weeds is crucial for site-specific farm management; however, most existing methods depend on labor intensive pixel-level annotations. A further challenge arises when models trained on one field (source domain) fail to generalize to new fields (target domain) due to domain shifts, such as variations in lighting, camera setups, soil composition, and crop growth stages. Unsupervised Domain Adaptation (UDA) addresses this by enabling adaptation without target-domain labels, but current UDA methods struggle with occlusions and visual blending between crops and weeds, leading to misclassifications in real-world conditions. To overcome these limitations, we introduce MaskAdapt, a novel approach that enhances segmentation accuracy through multimodal contextual learning by integrating RGB images with features derived from depth data. By computing depth gradients from depth maps, our method captures spatial transitions that help resolve texture ambiguities. These gradients, through a cross-attention mechanism, refines RGB feature representations, resulting in sharper boundary delineation. In addition, we propose a geometry-aware masking strategy that applies horizontal, vertical, and stochastic masks during training. This encourages the model to focus on the broader spatial context for robust visual recognition. Evaluations on real agricultural datasets demonstrate that MaskAdapt consistently outperforms existing State-of-the-Art (SOTA) UDA methods, achieving improved segmentation mean Intersection over Union (mIOU) across diverse field conditions.
Authors: Huan Ning, Zhenlong Li, Manzhu Yu, Wenpeng Yin
Abstract: Built environment auditing refers to the systematic documentation and assessment of urban and rural spaces' physical, social, and environmental characteristics, such as walkability, road conditions, and traffic lights. It is used to collect data for the evaluation of how built environments impact human behavior, health, mobility, and overall urban functionality. Traditionally, built environment audits were conducted using field surveys and manual observations, which were time-consuming and costly. The emerging street view imagery, e.g., Google Street View, has become a widely used data source for conducting built environment audits remotely. Deep learning and computer vision techniques can extract and classify objects from street images to enhance auditing productivity. Before meaningful analysis, the detected objects need to be geospatially mapped for accurate documentation. However, the mapping methods and tools based on street images are underexplored, and there are no universal frameworks or solutions yet, imposing difficulties in auditing the street objects. In this study, we introduced an open source street view mapping framework, providing three pipelines to map and measure: 1) width measurement for ground objects, such as roads; 2) 3D localization for objects with a known dimension (e.g., doors and stop signs); and 3) diameter measurements (e.g., street trees). These pipelines can help researchers, urban planners, and other professionals automatically measure and map target objects, promoting built environment auditing productivity and accuracy. Three case studies, including road width measurement, stop sign localization, and street tree diameter measurement, are provided in this paper to showcase pipeline usage.
Authors: Zeeshan Khan, Shizhe Chen, Cordelia Schmid
Abstract: Generating images from text involving complex and novel object arrangements remains a significant challenge for current text-to-image (T2I) models. Although prior layout-based methods improve object arrangements using spatial constraints with 2D layouts, they often struggle to capture 3D positioning and sacrifice quality and coherence. In this work, we introduce ComposeAnything, a novel framework for improving compositional image generation without retraining existing T2I models. Our approach first leverages the chain-of-thought reasoning abilities of LLMs to produce 2.5D semantic layouts from text, consisting of 2D object bounding boxes enriched with depth information and detailed captions. Based on this layout, we generate a spatial and depth aware coarse composite of objects that captures the intended composition, serving as a strong and interpretable prior that replaces stochastic noise initialization in diffusion-based T2I models. This prior guides the denoising process through object prior reinforcement and spatial-controlled denoising, enabling seamless generation of compositional objects and coherent backgrounds, while allowing refinement of inaccurate priors. ComposeAnything outperforms state-of-the-art methods on the T2I-CompBench and NSR-1K benchmarks for prompts with 2D/3D spatial arrangements, high object counts, and surreal compositions. Human evaluations further demonstrate that our model generates high-quality images with compositions that faithfully reflect the text.
Authors: Peiran Xu, Yadong Mu
Abstract: In this work, we focus on the task of weakly supervised affordance grounding, where a model is trained to identify affordance regions on objects using human-object interaction images and egocentric object images without dense labels. Previous works are mostly built upon class activation maps, which are effective for semantic segmentation but may not be suitable for locating actions and functions. Leveraging recent advanced foundation models, we develop a supervised training pipeline based on pseudo labels. The pseudo labels are generated from an off-the-shelf part segmentation model, guided by a mapping from affordance to part names. Furthermore, we introduce three key enhancements to the baseline model: a label refining stage, a fine-grained feature alignment process, and a lightweight reasoning module. These techniques harness the semantic knowledge of static objects embedded in off-the-shelf foundation models to improve affordance learning, effectively bridging the gap between objects and actions. Extensive experiments demonstrate that the performance of the proposed model has achieved a breakthrough improvement over existing methods. Our codes are available at https://github.com/woyut/WSAG-PLSP .
Authors: Alina Devkota, Annahita Amireskandari, Joel Palko, Shyam Thakkar, Donald Adjeroh, Xiajun Jiang, Binod Bhattarai, Prashnna K. Gyawali
Abstract: Gastrointestinal (GI) endoscopy is essential in identifying GI tract abnormalities in order to detect diseases in their early stages and improve patient outcomes. Although deep learning has shown success in supporting GI diagnostics and decision-making, these models require curated datasets with labels that are expensive to acquire. Foundation models offer a promising solution by learning general-purpose representations, which can be finetuned for specific tasks, overcoming data scarcity. Developing foundation models for medical imaging holds significant potential, but the sensitive and protected nature of medical data presents unique challenges. Foundation model training typically requires extensive datasets, and while hospitals generate large volumes of data, privacy restrictions prevent direct data sharing, making foundation model training infeasible in most scenarios. In this work, we propose a FL framework for training foundation models for gastroendoscopy imaging, enabling data to remain within local hospital environments while contributing to a shared model. We explore several established FL algorithms, assessing their suitability for training foundation models without relying on task-specific labels, conducting experiments in both homogeneous and heterogeneous settings. We evaluate the trained foundation model on three critical downstream tasks--classification, detection, and segmentation--and demonstrate that it achieves improved performance across all tasks, highlighting the effectiveness of our approach in a federated, privacy-preserving setting.
Authors: Ai Jian, Weijie Qiu, Xiaokun Wang, Peiyu Wang, Yunzhuo Hao, Jiangbo Pei, Yichen Wei, Yi Peng, Xuchen Song
Abstract: Vision-Language Models (VLMs) have demonstrated remarkable progress in multimodal understanding, yet their capabilities for scientific reasoning remains inadequately assessed. Current multimodal benchmarks predominantly evaluate generic image comprehension or text-driven reasoning, lacking authentic scientific contexts that require domain-specific knowledge integration with visual evidence analysis. To fill this gap, we present CSVQA, a diagnostic multimodal benchmark specifically designed for evaluating scientific reasoning through domain-grounded visual question answering.Our benchmark features 1,378 carefully constructed question-answer pairs spanning diverse STEM disciplines, each demanding domain knowledge, integration of visual evidence, and higher-order reasoning. Compared to prior multimodal benchmarks, CSVQA places greater emphasis on real-world scientific content and complex reasoning.We additionally propose a rigorous evaluation protocol to systematically assess whether model predictions are substantiated by valid intermediate reasoning steps based on curated explanations. Our comprehensive evaluation of 15 VLMs on this benchmark reveals notable performance disparities, as even the top-ranked proprietary model attains only 49.6\% accuracy.This empirical evidence underscores the pressing need for advancing scientific reasoning capabilities in VLMs. Our CSVQA is released at https://huggingface.co/datasets/Skywork/CSVQA.
Authors: Yichen Xie, Runsheng Xu, Tong He, Jyh-Jing Hwang, Katie Luo, Jingwei Ji, Hubert Lin, Letian Chen, Yiren Lu, Zhaoqi Leng, Dragomir Anguelov, Mingxing Tan
Abstract: The latest advancements in multi-modal large language models (MLLMs) have spurred a strong renewed interest in end-to-end motion planning approaches for autonomous driving. Many end-to-end approaches rely on human annotations to learn intermediate perception and prediction tasks, while purely self-supervised approaches--which directly learn from sensor inputs to generate planning trajectories without human annotations often underperform the state of the art. We observe a key gap in the input representation space: end-to-end approaches built on MLLMs are often pretrained with reasoning tasks in 2D image space rather than the native 3D space in which autonomous vehicles plan. To this end, we propose S4-Driver, a scalable self-supervised motion planning algorithm with spatio-temporal visual representation, based on the popular PaLI multimodal large language model. S4-Driver uses a novel sparse volume strategy to seamlessly transform the strong visual representation of MLLMs from perspective view to 3D space without the need to finetune the vision encoder. This representation aggregates multi-view and multi-frame visual inputs and enables better prediction of planning trajectories in 3D space. To validate our method, we run experiments on both nuScenes and Waymo Open Motion Dataset (with in-house camera data). Results show that S4-Driver performs favorably against existing supervised multi-task approaches while requiring no human annotations. It also demonstrates great scalability when pretrained on large volumes of unannotated driving logs.
Authors: Jiashuai Liu, Yingjia Shang, Yingkang Zhan, Di Zhang, Yi Niu, Dong Wei, Xian Wu, Zeyu Gao, Chen Li, Yefeng Zheng
Abstract: With the widespread adoption of pathology foundation models in both research and clinical decision support systems, exploring their security has become a critical concern. However, despite their growing impact, the vulnerability of these models to adversarial attacks remains largely unexplored. In this work, we present the first systematic investigation into the security of pathology foundation models for whole slide image~(WSI) analysis against adversarial attacks. Specifically, we introduce the principle of \textit{local perturbation with global impact} and propose a label-free attack framework that operates without requiring access to downstream task labels. Under this attack framework, we revise four classical white-box attack methods and redefine the perturbation budget based on the characteristics of WSI. We conduct comprehensive experiments on three representative pathology foundation models across five datasets and six downstream tasks. Despite modifying only 0.1\% of patches per slide with imperceptible noise, our attack leads to downstream accuracy degradation that can reach up to 20\% in the worst cases. Furthermore, we analyze key factors that influence attack success, explore the relationship between patch-level vulnerability and semantic content, and conduct a preliminary investigation into potential defence strategies. These findings lay the groundwork for future research on the adversarial robustness and reliable deployment of pathology foundation models. Our code is publicly available at: https://github.com/Jiashuai-Liu-hmos/Attack-WSI-pathology-foundation-models.
URLs: https://github.com/Jiashuai-Liu-hmos/Attack-WSI-pathology-foundation-models.
Authors: Chenyou Fan, Fangzheng Yan, Chenjia Bai, Jiepeng Wang, Chi Zhang, Zhen Wang, Xuelong Li
Abstract: Learning a generalizable bimanual manipulation policy is extremely challenging for embodied agents due to the large action space and the need for coordinated arm movements. Existing approaches rely on Vision-Language-Action (VLA) models to acquire bimanual policies. However, transferring knowledge from single-arm datasets or pre-trained VLA models often fails to generalize effectively, primarily due to the scarcity of bimanual data and the fundamental differences between single-arm and bimanual manipulation. In this paper, we propose a novel bimanual foundation policy by fine-tuning the leading text-to-video models to predict robot trajectories and training a lightweight diffusion policy for action generation. Given the lack of embodied knowledge in text-to-video models, we introduce a two-stage paradigm that fine-tunes independent text-to-flow and flow-to-video models derived from a pre-trained text-to-video model. Specifically, optical flow serves as an intermediate variable, providing a concise representation of subtle movements between images. The text-to-flow model predicts optical flow to concretize the intent of language instructions, and the flow-to-video model leverages this flow for fine-grained video prediction. Our method mitigates the ambiguity of language in single-stage text-to-video prediction and significantly reduces the robot-data requirement by avoiding direct use of low-level actions. In experiments, we collect high-quality manipulation data for real dual-arm robot, and the results of simulation and real-world experiments demonstrate the effectiveness of our method.
Authors: Bo Fang, Wenhao Wu, Qiangqiang Wu, Yuxin Song, Antoni B. Chan
Abstract: Employing Multimodal Large Language Models (MLLMs) for long video understanding remains a challenging problem due to the dilemma between the substantial number of video frames (i.e., visual tokens) versus the limited context length of language models. Traditional uniform sampling often leads to selection of irrelevant content, while post-training MLLMs on thousands of frames imposes a substantial computational burden. In this paper, we propose threading keyframes with narratives (Nar-KFC), a plug-and-play module to facilitate effective and efficient long video perception. Nar-KFC generally involves two collaborative steps. First, we formulate the keyframe selection process as an integer quadratic programming problem, jointly optimizing query-relevance and frame-diversity. To avoid its computational complexity, a customized greedy search strategy is designed as an efficient alternative. Second, to mitigate the temporal discontinuity caused by sparse keyframe sampling, we further introduce interleaved textual narratives generated from non-keyframes using off-the-shelf captioners. These narratives are inserted between keyframes based on their true temporal order, forming a coherent and compact representation. Nar-KFC thus serves as a temporal- and content-aware compression strategy that complements visual and textual modalities. Experimental results on multiple long-video benchmarks demonstrate that Nar-KFC significantly improves the performance of popular MLLMs. Code will be made publicly available.
Authors: Isaac Aguirre, Ivan Sipiran
Abstract: We present a simple yet effective training-free approach for zero-shot 3D symmetry detection that leverages visual features from foundation vision models such as DINOv2. Our method extracts features from rendered views of 3D objects and backprojects them onto the original geometry. We demonstrate the symmetric invariance of these features and use them to identify reflection-symmetry planes through a proposed algorithm. Experiments on a subset of ShapeNet demonstrate that our approach outperforms both traditional geometric methods and learning-based approaches without requiring any training data. Our work demonstrates how foundation vision models can help in solving complex 3D geometric problems such as symmetry detection.
Authors: Junyu Chen, Shuwen Wei, Yihao Liu, Aaron Carass, Yong Du
Abstract: Recent advances in deep learning-based medical image registration have shown that training deep neural networks~(DNNs) does not necessarily require medical images. Previous work showed that DNNs trained on randomly generated images with carefully designed noise and contrast properties can still generalize well to unseen medical data. Building on this insight, we propose using registration between random images as a proxy task for pretraining a foundation model for image registration. Empirical results show that our pretraining strategy improves registration accuracy, reduces the amount of domain-specific data needed to achieve competitive performance, and accelerates convergence during downstream training, thereby enhancing computational efficiency.
Authors: Tianhong Zhou, Yin Xu, Yingtao Zhu, Chuxi Xiao, Haiyang Bian, Lei Wei, Xuegong Zhang
Abstract: Vision-language models (VLMs) exhibit strong zero-shot generalization on natural images and show early promise in interpretable medical image analysis. However, existing benchmarks do not systematically evaluate whether these models truly reason like human clinicians or merely imitate superficial patterns. To address this gap, we propose DrVD-Bench, the first multimodal benchmark for clinical visual reasoning. DrVD-Bench consists of three modules: Visual Evidence Comprehension, Reasoning Trajectory Assessment, and Report Generation Evaluation, comprising a total of 7,789 image-question pairs. Our benchmark covers 20 task types, 17 diagnostic categories, and five imaging modalities-CT, MRI, ultrasound, radiography, and pathology. DrVD-Bench is explicitly structured to reflect the clinical reasoning workflow from modality recognition to lesion identification and diagnosis. We benchmark 19 VLMs, including general-purpose and medical-specific, open-source and proprietary models, and observe that performance drops sharply as reasoning complexity increases. While some models begin to exhibit traces of human-like reasoning, they often still rely on shortcut correlations rather than grounded visual understanding. DrVD-Bench offers a rigorous and structured evaluation framework to guide the development of clinically trustworthy VLMs.
Authors: Zhuobai Dong, Junchao Yi, Ziyuan Zheng, Haochen Han, Xiangxi Zheng, Alex Jinpeng Wang, Fangming Liu, Linjie Li
Abstract: Understanding the physical world - governed by laws of motion, spatial relations, and causality - poses a fundamental challenge for multimodal large language models (MLLMs). While recent advances such as OpenAI o3 and GPT-4o demonstrate impressive perceptual and reasoning capabilities, our investigation reveals these models struggle profoundly with visual physical reasoning, failing to grasp basic physical laws, spatial interactions, and causal effects in complex scenes. More importantly, they often fail to follow coherent reasoning chains grounded in visual evidence, especially when multiple steps are needed to arrive at the correct answer. To rigorously evaluate this capability, we introduce MVPBench, a curated benchmark designed to rigorously evaluate visual physical reasoning through the lens of visual chain-of-thought (CoT). Each example features interleaved multi-image inputs and demands not only the correct final answer but also a coherent, step-by-step reasoning path grounded in evolving visual cues. This setup mirrors how humans reason through real-world physical processes over time. To ensure fine-grained evaluation, we introduce a graph-based CoT consistency metric that verifies whether the reasoning path of model adheres to valid physical logic. Additionally, we minimize shortcut exploitation from text priors, encouraging models to rely on visual understanding. Experimental results reveal a concerning trend: even cutting-edge MLLMs exhibit poor visual reasoning accuracy and weak image-text alignment in physical domains. Surprisingly, RL-based post-training alignment - commonly believed to improve visual reasoning performance - often harms spatial reasoning, suggesting a need to rethink current fine-tuning practices.
Authors: Gang Wu, Junjun Jiang, Kui Jiang, Xianming Liu
Abstract: Unified image restoration models for diverse and mixed degradations often suffer from unstable optimization dynamics and inter-task conflicts. This paper introduces Self-Improved Privilege Learning (SIPL), a novel paradigm that overcomes these limitations by innovatively extending the utility of privileged information (PI) beyond training into the inference stage. Unlike conventional Privilege Learning, where ground-truth-derived guidance is typically discarded after training, SIPL empowers the model to leverage its own preliminary outputs as pseudo-privileged signals for iterative self-refinement at test time. Central to SIPL is Proxy Fusion, a lightweight module incorporating a learnable Privileged Dictionary. During training, this dictionary distills essential high-frequency and structural priors from privileged feature representations. Critically, at inference, the same learned dictionary then interacts with features derived from the model's initial restoration, facilitating a self-correction loop. SIPL can be seamlessly integrated into various backbone architectures, offering substantial performance improvements with minimal computational overhead. Extensive experiments demonstrate that SIPL significantly advances the state-of-the-art on diverse all-in-one image restoration benchmarks. For instance, when integrated with the PromptIR model, SIPL achieves remarkable PSNR improvements of +4.58 dB on composite degradation tasks and +1.28 dB on diverse five-task benchmarks, underscoring its effectiveness and broad applicability. Codes are available at our project page https://github.com/Aitical/SIPL.
Authors: Zheng Tan, Weizhen Wang, Andrea L. Bertozzi, Ernest K. Ryu
Abstract: Diffusion models (DMs) have demonstrated remarkable performance in high-fidelity image and video generation. Because high-quality generations with DMs typically require a large number of function evaluations (NFEs), resulting in slow sampling, there has been extensive research successfully reducing the NFE to a small range (<10) while maintaining acceptable image quality. However, many practical applications, such as those involving Stable Diffusion 3.5, FLUX, and SANA, commonly operate in the mid-NFE regime (20-50 NFE) to achieve superior results, and, despite the practical relevance, research on the effective sampling within this mid-NFE regime remains underexplored. In this work, we propose a novel, training-free, and structure-independent DM ODE solver called the Stabilized Taylor Orthogonal Runge--Kutta (STORK) method, based on a class of stiff ODE solvers with a Taylor expansion adaptation. Unlike prior work such as DPM-Solver, which is dependent on the semi-linear structure of the DM ODE, STORK is applicable to any DM sampling, including noise-based and flow matching-based models. Within the 20-50 NFE range, STORK achieves improved generation quality, as measured by FID scores, across unconditional pixel-level generation and conditional latent-space generation tasks using models like Stable Diffusion 3.5 and SANA. Code is available at https://github.com/ZT220501/STORK.
Authors: Redwan Sony, Parisa Farmanifard, Hamzeh Alzwairy, Nitish Shukla, Arun Ross
Abstract: The advent of foundation models, particularly Vision-Language Models (VLMs) and Multi-modal Large Language Models (MLLMs), has redefined the frontiers of artificial intelligence, enabling remarkable generalization across diverse tasks with minimal or no supervision. Yet, their potential in biometric recognition and analysis remains relatively underexplored. In this work, we introduce a comprehensive benchmark that evaluates the zero-shot and few-shot performance of state-of-the-art publicly available VLMs and MLLMs across six biometric tasks spanning the face and iris modalities: face verification, soft biometric attribute prediction (gender and race), iris recognition, presentation attack detection (PAD), and face manipulation detection (morphs and deepfakes). A total of 41 VLMs were used in this evaluation. Experiments show that embeddings from these foundation models can be used for diverse biometric tasks with varying degrees of success. For example, in the case of face verification, a True Match Rate (TMR) of 96.77 percent was obtained at a False Match Rate (FMR) of 1 percent on the Labeled Face in the Wild (LFW) dataset, without any fine-tuning. In the case of iris recognition, the TMR at 1 percent FMR on the IITD-R-Full dataset was 97.55 percent without any fine-tuning. Further, we show that applying a simple classifier head to these embeddings can help perform DeepFake detection for faces, Presentation Attack Detection (PAD) for irides, and extract soft biometric attributes like gender and ethnicity from faces with reasonably high accuracy. This work reiterates the potential of pretrained models in achieving the long-term vision of Artificial General Intelligence.
Authors: Prasanna Reddy Pulakurthi, Majid Rabbani, Jamison Heard, Sohail Dianat, Celso M. de Melo, Raghuveer Rao
Abstract: This work investigates Source-Free Domain Adaptation (SFDA), where a model adapts to a target domain without access to source data. A new augmentation technique, Shuffle PatchMix (SPM), and a novel reweighting strategy are introduced to enhance performance. SPM shuffles and blends image patches to generate diverse and challenging augmentations, while the reweighting strategy prioritizes reliable pseudo-labels to mitigate label noise. These techniques are particularly effective on smaller datasets like PACS, where overfitting and pseudo-label noise pose greater risks. State-of-the-art results are achieved on three major benchmarks: PACS, VisDA-C, and DomainNet-126. Notably, on PACS, improvements of 7.3% (79.4% to 86.7%) and 7.2% are observed in single-target and multi-target settings, respectively, while gains of 2.8% and 0.7% are attained on DomainNet-126 and VisDA-C. This combination of advanced augmentation and robust pseudo-label reweighting establishes a new benchmark for SFDA. The code is available at: https://github.com/PrasannaPulakurthi/SPM
Authors: Fangyikang Wang, Hubery Yin, Lei Qian, Yinan Li, Shaobin Zhuang, Huminhao Zhu, Yilin Zhang, Yanlong Tang, Chao Zhang, Hanbin Zhao, Hui Qian, Chen Li
Abstract: The diffusion models (DMs) have demonstrated the remarkable capability of generating images via learning the noised score function of data distribution. Current DM sampling techniques typically rely on first-order Langevin dynamics at each noise level, with efforts concentrated on refining inter-level denoising strategies. While leveraging additional second-order Hessian geometry to enhance the sampling quality of Langevin is a common practice in Markov chain Monte Carlo (MCMC), the naive attempts to utilize Hessian geometry in high-dimensional DMs lead to quadratic-complexity computational costs, rendering them non-scalable. In this work, we introduce a novel Levenberg-Marquardt-Langevin (LML) method that approximates the diffusion Hessian geometry in a training-free manner, drawing inspiration from the celebrated Levenberg-Marquardt optimization algorithm. Our approach introduces two key innovations: (1) A low-rank approximation of the diffusion Hessian, leveraging the DMs' inherent structure and circumventing explicit quadratic-complexity computations; (2) A damping mechanism to stabilize the approximated Hessian. This LML approximated Hessian geometry enables the diffusion sampling to execute more accurate steps and improve the image generation quality. We further conduct a theoretical analysis to substantiate the approximation error bound of low-rank approximation and the convergence property of the damping mechanism. Extensive experiments across multiple pretrained DMs validate that the LML method significantly improves image generation quality, with negligible computational overhead.
Authors: Haibo Jin, Peiyan Zhang, Man Luo, Haohan Wang
Abstract: Large Language Models (LLMs) have shown remarkable progress across domains, yet their ability to perform inductive reasoning - inferring latent rules from sparse examples - remains limited. It is often assumed that chain-of-thought (CoT) prompting, as used in Large Reasoning Models (LRMs), enhances such reasoning. We investigate this assumption with creating four controlled, diagnostic game-based tasks - chess, Texas Hold'em, dice games, and blackjack - with hidden human-defined rules. We find that CoT reasoning can degrade inductive performance, with LRMs often underperforming their non-reasoning counterparts. To explain this, we present a theoretical framework that reveals how reasoning steps can amplify error through three failure modes: incorrect sub-task decomposition, incorrect sub-task solving, and incorrect final answer summarization. Based on our theoretical and empirical analysis, we introduce structured interventions that adapt CoT generation according to our identified failure types. These interventions improve inductive accuracy without retraining. Our findings suggest that effective (CoT) reasoning depends not only on taking more steps but also on ensuring those steps are well-structured.
Authors: Ying Yang, Jie Zhang, Xiao Lv, Di Lin, Tao Xiang, Qing Guo
Abstract: While adversarial attacks on vision-and-language pretraining (VLP) models have been explored, generating natural adversarial samples crafted through realistic and semantically meaningful perturbations remains an open challenge. Existing methods, primarily designed for classification tasks, struggle when adapted to VLP models due to their restricted optimization spaces, leading to ineffective attacks or unnatural artifacts. To address this, we propose \textbf{LightD}, a novel framework that generates natural adversarial samples for VLP models via semantically guided relighting. Specifically, LightD leverages ChatGPT to propose context-aware initial lighting parameters and integrates a pretrained relighting model (IC-light) to enable diverse lighting adjustments. LightD expands the optimization space while ensuring perturbations align with scene semantics. Additionally, gradient-based optimization is applied to the reference lighting image to further enhance attack effectiveness while maintaining visual naturalness. The effectiveness and superiority of the proposed LightD have been demonstrated across various VLP models in tasks such as image captioning and visual question answering.
Authors: Haibo Jin, Peiyan Zhang, Peiran Wang, Man Luo, Haohan Wang
Abstract: Large foundation models (LFMs) are susceptible to two distinct vulnerabilities: hallucinations and jailbreak attacks. While typically studied in isolation, we observe that defenses targeting one often affect the other, hinting at a deeper connection. We propose a unified theoretical framework that models jailbreaks as token-level optimization and hallucinations as attention-level optimization. Within this framework, we establish two key propositions: (1) \textit{Similar Loss Convergence} - the loss functions for both vulnerabilities converge similarly when optimizing for target-specific outputs; and (2) \textit{Gradient Consistency in Attention Redistribution} - both exhibit consistent gradient behavior driven by shared attention dynamics. We validate these propositions empirically on LLaVA-1.5 and MiniGPT-4, showing consistent optimization trends and aligned gradients. Leveraging this connection, we demonstrate that mitigation techniques for hallucinations can reduce jailbreak success rates, and vice versa. Our findings reveal a shared failure mode in LFMs and suggest that robustness strategies should jointly address both vulnerabilities.
Authors: Bowen Dong, Minheng Ni, Zitong Huang, Guanglei Yang, Wangmeng Zuo, Lei Zhang
Abstract: Multimodal hallucination in multimodal large language models (MLLMs) restricts the correctness of MLLMs. However, multimodal hallucinations are multi-sourced and arise from diverse causes. Existing benchmarks fail to adequately distinguish between perception-induced hallucinations and reasoning-induced hallucinations. This failure constitutes a significant issue and hinders the diagnosis of multimodal reasoning failures within MLLMs. To address this, we propose the {\dataset} benchmark, which isolates reasoning hallucinations by constructing questions where input images are correctly perceived by MLLMs yet reasoning errors persist. {\dataset} introduces multi-granular evaluation metrics: accuracy, factuality, and LLMs hallucination score for hallucination quantification. Our analysis reveals that (1) the model scale, data scale, and training stages significantly affect the degree of logical, fabrication, and factual hallucinations; (2) current MLLMs show no effective improvement on spatial hallucinations caused by misinterpreted spatial relationships, indicating their limited visual reasoning capabilities; and (3) question types correlate with distinct hallucination patterns, highlighting targeted challenges and potential mitigation strategies. To address these challenges, we propose {\method}, a method that combines curriculum reinforcement fine-tuning to encourage models to generate logic-consistent reasoning chains by stepwise reducing learning difficulty, and collaborative hint inference to reduce reasoning complexity. {\method} establishes a baseline on {\dataset}, and reduces the logical hallucinations in original base models.
Authors: Xin Kang, Zihan Zheng, Lei Chu, Yue Gao, Jiahao Li, Hao Pan, Xuejin Chen, Yan Lu
Abstract: We present LTM3D, a Latent Token space Modeling framework for conditional 3D shape generation that integrates the strengths of diffusion and auto-regressive (AR) models. While diffusion-based methods effectively model continuous latent spaces and AR models excel at capturing inter-token dependencies, combining these paradigms for 3D shape generation remains a challenge. To address this, LTM3D features a Conditional Distribution Modeling backbone, leveraging a masked autoencoder and a diffusion model to enhance token dependency learning. Additionally, we introduce Prefix Learning, which aligns condition tokens with shape latent tokens during generation, improving flexibility across modalities. We further propose a Latent Token Reconstruction module with Reconstruction-Guided Sampling to reduce uncertainty and enhance structural fidelity in generated shapes. Our approach operates in token space, enabling support for multiple 3D representations, including signed distance fields, point clouds, meshes, and 3D Gaussian Splatting. Extensive experiments on image- and text-conditioned shape generation tasks demonstrate that LTM3D outperforms existing methods in prompt fidelity and structural accuracy while offering a generalizable framework for multi-modal, multi-representation 3D generation.
Authors: Minchul Kim, Anil Jain, Xiaoming Liu
Abstract: Over the past 50 years, automated face recognition has evolved from rudimentary, handcrafted systems into sophisticated deep learning models that rival and often surpass human performance. This paper chronicles the history and technological progression of FR, from early geometric and statistical methods to modern deep neural architectures leveraging massive real and AI-generated datasets. We examine key innovations that have shaped the field, including developments in dataset, loss function, neural network design and feature fusion. We also analyze how the scale and diversity of training data influence model generalization, drawing connections between dataset growth and benchmark improvements. Recent advances have achieved remarkable milestones: state-of-the-art face verification systems now report False Negative Identification Rates of 0.13% against a 12.4 million gallery in NIST FRVT evaluations for 1:N visa-to-border matching. While recent advances have enabled remarkable accuracy in high- and low-quality face scenarios, numerous challenges persist. While remarkable progress has been achieved, several open research problems remain. We outline critical challenges and promising directions for future face recognition research, including scalability, multi-modal fusion, synthetic identity generation, and explainable systems.
Authors: Qingyao Tian, Huai Liao, Xinyan Huang, Bingyu Yang, Hongbin Liu
Abstract: Vision-based 6-DOF bronchoscopy localization offers a promising solution for accurate and cost-effective interventional guidance. However, existing methods struggle with 1) limited generalization across patient cases due to scarce labeled data, and 2) poor robustness under visual degradation, as bronchoscopy procedures frequently involve artifacts such as occlusions and motion blur that impair visual information. To address these challenges, we propose PANSv2, a generalizable and robust bronchoscopy localization framework. Motivated by PANS that leverages multiple visual cues for pose likelihood measurement, PANSv2 integrates depth estimation, landmark detection, and centerline constraints into a unified pose optimization framework that evaluates pose probability and solves for the optimal bronchoscope pose. To further enhance generalization capabilities, we leverage the endoscopic foundation model EndoOmni for depth estimation and the video foundation model EndoMamba for landmark detection, incorporating both spatial and temporal analyses. Pretrained on diverse endoscopic datasets, these models provide stable and transferable visual representations, enabling reliable performance across varied bronchoscopy scenarios. Additionally, to improve robustness to visual degradation, we introduce an automatic re-initialization module that detects tracking failures and re-establishes pose using landmark detections once clear views are available. Experimental results on bronchoscopy dataset encompassing 10 patient cases show that PANSv2 achieves the highest tracking success rate, with an 18.1% improvement in SR-5 (percentage of absolute trajectory error under 5 mm) compared to existing methods, showing potential towards real clinical usage.
Authors: Ishaan Rawal, Suryansh Kumar
Abstract: Text-conditioned diffusion models have emerged as powerful tools for high-quality video generation. However, enabling Interactive Video Generation (IVG), where users control motion elements such as object trajectory, remains challenging. Recent training-free approaches introduce attention masking to guide trajectory, but this often degrades perceptual quality. We identify two key failure modes in these methods, both of which we interpret as domain shift problems, and propose solutions inspired by domain adaptation. First, we attribute the perceptual degradation to internal covariate shift induced by attention masking, as pretrained models are not trained to handle masked attention. To address this, we propose mask normalization, a pre-normalization layer designed to mitigate this shift via distribution matching. Second, we address initialization gap, where the randomly sampled initial noise does not align with IVG conditioning, by introducing a temporal intrinsic diffusion prior that enforces spatio-temporal consistency at each denoising step. Extensive qualitative and quantitative evaluations demonstrate that mask normalization and temporal intrinsic denoising improve both perceptual quality and trajectory control over the existing state-of-the-art IVG techniques.
Authors: Sahithya Ravi, Gabriel Sarch, Vibhav Vineet, Andrew D. Wilson, Balasaravanan Thoravi Kumaravel
Abstract: An embodied AI assistant operating on egocentric video must integrate spatial cues across time - for instance, determining where an object A, glimpsed a few moments ago lies relative to an object B encountered later. We introduce Disjoint-3DQA , a generative QA benchmark that evaluates this ability of VLMs by posing questions about object pairs that are not co-visible in the same frame. We evaluated seven state-of-the-art VLMs and found that models lag behind human performance by 28%, with steeper declines in accuracy (60% to 30 %) as the temporal gap widens. Our analysis further reveals that providing trajectories or bird's-eye-view projections to VLMs results in only marginal improvements, whereas providing oracle 3D coordinates leads to a substantial 20% performance increase. This highlights a core bottleneck of multi-frame VLMs in constructing and maintaining 3D scene representations over time from visual signals. Disjoint-3DQA therefore sets a clear, measurable challenge for long-horizon spatial reasoning and aims to catalyze future research at the intersection of vision, language, and embodied AI.
Authors: Zirui Shang, Xinxiao Wu, Shuo Yang
Abstract: Language-driven action localization in videos requires not only semantic alignment between language query and video segment, but also prediction of action boundaries. However, the language query primarily describes the main content of an action and usually lacks specific details of action start and end boundaries, which increases the subjectivity of manual boundary annotation and leads to boundary uncertainty in training data. In this paper, on one hand, we propose to expand the original query by generating textual descriptions of the action start and end boundaries through LLMs, which can provide more detailed boundary cues for localization and thus reduce the impact of boundary uncertainty. On the other hand, to enhance the tolerance to boundary uncertainty during training, we propose to model probability scores of action boundaries by calculating the semantic similarities between frames and the expanded query as well as the temporal distances between frames and the annotated boundary frames. They can provide more consistent boundary supervision, thus improving the stability of training. Our method is model-agnostic and can be seamlessly and easily integrated into any existing models of language-driven action localization in an off-the-shelf manner. Experimental results on several datasets demonstrate the effectiveness of our method.
Authors: Ege \"Ozsoy, Arda Mamur, Felix Tristram, Chantal Pellegrini, Magdalena Wysocki, Benjamin Busam, Nassir Navab
Abstract: Operating rooms (ORs) demand precise coordination among surgeons, nurses, and equipment in a fast-paced, occlusion-heavy environment, necessitating advanced perception models to enhance safety and efficiency. Existing datasets either provide partial egocentric views or sparse exocentric multi-view context, but do not explore the comprehensive combination of both. We introduce EgoExOR, the first OR dataset and accompanying benchmark to fuse first-person and third-person perspectives. Spanning 94 minutes (84,553 frames at 15 FPS) of two emulated spine procedures, Ultrasound-Guided Needle Insertion and Minimally Invasive Spine Surgery, EgoExOR integrates egocentric data (RGB, gaze, hand tracking, audio) from wearable glasses, exocentric RGB and depth from RGB-D cameras, and ultrasound imagery. Its detailed scene graph annotations, covering 36 entities and 22 relations (568,235 triplets), enable robust modeling of clinical interactions, supporting tasks like action recognition and human-centric perception. We evaluate the surgical scene graph generation performance of two adapted state-of-the-art models and offer a new baseline that explicitly leverages EgoExOR's multimodal and multi-perspective signals. This new dataset and benchmark set a new foundation for OR perception, offering a rich, multimodal resource for next-generation clinical perception.
Authors: Enshang Zhang, Zhicheng Zhang, Takashi Hanakawa
Abstract: Reconstructing visual stimuli from EEG signals is a crucial step in realizing brain-computer interfaces. In this paper, we propose a transformer-based EEG signal encoder integrating the Discrete Wavelet Transform (DWT) and the gating mechanism. Guided by the feature alignment and category-aware fusion losses, this encoder is used to extract features related to visual stimuli from EEG signals. Subsequently, with the aid of a pre-trained diffusion model, these features are reconstructed into visual stimuli. To verify the effectiveness of the model, we conducted EEG-to-image generation and classification tasks using the THINGS-EEG dataset. To address the limitations of quantitative analysis at the semantic level, we combined WordNet-based classification and semantic similarity metrics to propose a novel semantic-based score, emphasizing the ability of our model to transfer neural activities into visual representations. Experimental results show that our model significantly improves semantic alignment and classification accuracy, which achieves a maximum single-subject accuracy of 43\%, outperforming other state-of-the-art methods. The source code and supplementary material is available at https://github.com/zes0v0inn/DWT_EEG_Reconstruction/tree/main.
URLs: https://github.com/zes0v0inn/DWT_EEG_Reconstruction/tree/main.
Authors: Jiayan Li, Jun Li, Zhourui Zhang, Jianhua Xu
Abstract: In knowledge distillation (KD), logit distillation (LD) aims to transfer class-level knowledge from a more powerful teacher network to a small student model via accurate teacher-student alignment at the logits level. Since high-confidence object classes usually dominate the distillation process, low-probability classes which also contain discriminating information are downplayed in conventional methods, leading to insufficient knowledge transfer. To address this issue, we propose a simple yet effective LD method termed Progressive Class-level Distillation (PCD). In contrast to existing methods which perform all-class ensemble distillation, our PCD approach performs stage-wise distillation for step-by-step knowledge transfer. More specifically, we perform ranking on teacher-student logits difference for identifying distillation priority from scratch, and subsequently divide the entire LD process into multiple stages. Next, bidirectional stage-wise distillation incorporating fine-to-coarse progressive learning and reverse coarse-to-fine refinement is conducted, allowing comprehensive knowledge transfer via sufficient logits alignment within separate class groups in different distillation stages. Extension experiments on public benchmarking datasets demonstrate the superiority of our method compared to state-of-the-arts for both classification and detection tasks.
Authors: Jinlu Zhang, Yixin Chen, Zan Wang, Jie Yang, Yizhou Wang, Siyuan Huang
Abstract: Recent advances in 3D human-aware generation have made significant progress. However, existing methods still struggle with generating novel Human Object Interaction (HOI) from text, particularly for open-set objects. We identify three main challenges of this task: precise human-object relation reasoning, affordance parsing for any object, and detailed human interaction pose synthesis aligning description and object geometry. In this work, we propose a novel zero-shot 3D HOI generation framework without training on specific datasets, leveraging the knowledge from large-scale pre-trained models. Specifically, the human-object relations are inferred from large language models (LLMs) to initialize object properties and guide the optimization process. Then we utilize a pre-trained 2D image diffusion model to parse unseen objects and extract contact points, avoiding the limitations imposed by existing 3D asset knowledge. The initial human pose is generated by sampling multiple hypotheses through multi-view SDS based on the input text and object geometry. Finally, we introduce a detailed optimization to generate fine-grained, precise, and natural interaction, enforcing realistic 3D contact between the 3D object and the involved body parts, including hands in grasping. This is achieved by distilling human-level feedback from LLMs to capture detailed human-object relations from the text instruction. Extensive experiments validate the effectiveness of our approach compared to prior works, particularly in terms of the fine-grained nature of interactions and the ability to handle open-set 3D objects.
Authors: Jingjing Liu, Jiashun Jin, Xianchao Xiu, Jianhua Zhang, Wanquan Liu
Abstract: Remote sensing image (RSI) denoising is an important topic in the field of remote sensing. Despite the impressive denoising performance of RSI denoising methods, most current deep learning-based approaches function as black boxes and lack integration with physical information models, leading to limited interpretability. Additionally, many methods may struggle with insufficient attention to non-local self-similarity in RSI and require tedious tuning of regularization parameters to achieve optimal performance, particularly in conventional iterative optimization approaches. In this paper, we first propose a novel RSI denoising method named sparse tensor-aided representation network (STAR-Net), which leverages a low-rank prior to effectively capture the non-local self-similarity within RSI. Furthermore, we extend STAR-Net to a sparse variant called STAR-Net-S to deal with the interference caused by non-Gaussian noise in original RSI for the purpose of improving robustness. Different from conventional iterative optimization, we develop an alternating direction method of multipliers (ADMM)-guided deep unrolling network, in which all regularization parameters can be automatically learned, thus inheriting the advantages of both model-based and deep learning-based approaches and successfully addressing the above-mentioned shortcomings. Comprehensive experiments on synthetic and real-world datasets demonstrate that STAR-Net and STAR-Net-S outperform state-of-the-art RSI denoising methods.
Authors: Yingsen Zeng, Zepeng Huang, Yujie Zhong, Chengjian Feng, Jie Hu, Lin Ma, Yang Liu
Abstract: Despite advances in general video understanding, Video Large Language Models (Video-LLMs) face challenges in precise temporal localization due to discrete time representations and limited temporally aware datasets. Existing methods for temporal expression either conflate time with text-based numerical values, add a series of dedicated temporal tokens, or regress time using specialized temporal grounding heads. To address these issues, we introduce DisTime, a lightweight framework designed to enhance temporal comprehension in Video-LLMs. DisTime employs a learnable token to create a continuous temporal embedding space and incorporates a Distribution-based Time Decoder that generates temporal probability distributions, effectively mitigating boundary ambiguities and maintaining temporal continuity. Additionally, the Distribution-based Time Encoder re-encodes timestamps to provide time markers for Video-LLMs. To overcome temporal granularity limitations in existing datasets, we propose an automated annotation paradigm that combines the captioning capabilities of Video-LLMs with the localization expertise of dedicated temporal models. This leads to the creation of InternVid-TG, a substantial dataset with 1.25M temporally grounded events across 179k videos, surpassing ActivityNet-Caption by 55 times. Extensive experiments demonstrate that DisTime achieves state-of-the-art performance across benchmarks in three time-sensitive tasks while maintaining competitive performance in Video QA tasks. Code and data are released at https://github.com/josephzpng/DisTime.
Authors: Uzair Khan, Franco Fummi, Luigi Capogrosso
Abstract: In the era of intelligent manufacturing, anomaly detection has become essential for maintaining quality control on modern production lines. However, while many existing models show promising performance, they are often too large, computationally demanding, and impractical to deploy on resource-constrained embedded devices that can be easily installed on the production lines of Small and Medium Enterprises (SMEs). To bridge this gap, we present KairosAD, a novel supervised approach that uses the power of the Mobile Segment Anything Model (MobileSAM) for image-based anomaly detection. KairosAD has been evaluated on the two well-known industrial anomaly detection datasets, i.e., MVTec-AD and ViSA. The results show that KairosAD requires 78% fewer parameters and boasts a 4x faster inference time compared to the leading state-of-the-art model, while maintaining comparable AUROC performance. We deployed KairosAD on two embedded devices, the NVIDIA Jetson NX, and the NVIDIA Jetson AGX. Finally, KairosAD was successfully installed and tested on the real production line of the Industrial Computer Engineering Laboratory (ICE Lab) at the University of Verona. The code is available at https://github.com/intelligolabs/KairosAD.
Authors: Gilles Quentin Hacheme, Girmaw Abebe Tadesse, Caleb Robinson, Akram Zaytar, Rahul Dodhia, Juan M. Lavista Ferres
Abstract: Classifying geospatial imagery remains a major bottleneck for applications such as disaster response and land-use monitoring-particularly in regions where annotated data is scarce or unavailable. Existing tools (e.g., RS-CLIP) that claim zero-shot classification capabilities for satellite imagery nonetheless rely on task-specific pretraining and adaptation to reach competitive performance. We introduce GeoVision Labeler (GVL), a strictly zero-shot classification framework: a vision Large Language Model (vLLM) generates rich, human-readable image descriptions, which are then mapped to user-defined classes by a conventional Large Language Model (LLM). This modular, and interpretable pipeline enables flexible image classification for a large range of use cases. We evaluated GVL across three benchmarks-SpaceNet v7, UC Merced, and RESISC45. It achieves up to 93.2% zero-shot accuracy on the binary Buildings vs. No Buildings task on SpaceNet v7. For complex multi-class classification tasks (UC Merced, RESISC45), we implemented a recursive LLM-driven clustering to form meta-classes at successive depths, followed by hierarchical classification-first resolving coarse groups, then finer distinctions-to deliver competitive zero-shot performance. GVL is open-sourced at https://github.com/microsoft/geo-vision-labeler to catalyze adoption in real-world geospatial workflows.
Authors: Fanhang Man, Xiaoyue Chen, Huandong Wang, Baining Zhao, Han Li, Xinlei Chen, Yong Li
Abstract: Understanding what emotions images evoke in their viewers is a foundational goal in human-centric visual computing. While recent advances in vision-language models (VLMs) have shown promise for visual emotion analysis (VEA), several key challenges remain unresolved. Emotional cues in images are often abstract, overlapping, and entangled, making them difficult to model and interpret. Moreover, VLMs struggle to align these complex visual patterns with emotional semantics due to limited supervision and sparse emotional grounding. Finally, existing approaches lack structured affective knowledge to resolve ambiguity and ensure consistent emotional reasoning across diverse visual domains. To address these limitations, we propose \textbf{K-EVER\textsuperscript{2}}, a knowledge-enhanced framework for emotion reasoning and retrieval. Our approach introduces a semantically structured formulation of visual emotion cues and integrates external affective knowledge through multimodal alignment. Without relying on handcrafted labels or direct emotion supervision, K-EVER\textsuperscript{2} achieves robust and interpretable emotion predictions across heterogeneous image types. We validate our framework on three representative benchmarks, Emotion6, EmoSet, and M-Disaster, covering social media imagery, human-centric scenes, and disaster contexts. K-EVER\textsuperscript{2} consistently outperforms strong CNN and VLM baselines, achieving up to a \textbf{19\% accuracy gain} for specific emotions and a \textbf{12.3\% average accuracy gain} across all emotion categories. Our results demonstrate a scalable and generalizable solution for advancing emotional understanding of visual content.
Authors: Ziyi Wang, Zhi Gao, Boxuan Yu, Zirui Dai, Yuxiang Song, Qingyuan Lu, Jin Chen, Xinxiao Wu
Abstract: Video understanding has made remarkable progress in recent years, largely driven by advances in deep models and the availability of large-scale annotated datasets. However, existing works typically ignore the inherent domain shifts encountered in real-world video applications, leaving domain generalization (DG) in video understanding underexplored. Hence, we propose Video Understanding Domain Generalization (VUDG), a novel dataset designed specifically for evaluating the DG performance in video understanding. VUDG contains videos from 11 distinct domains that cover three types of domain shifts, and maintains semantic similarity across different domains to ensure fair and meaningful evaluation. We propose a multi-expert progressive annotation framework to annotate each video with both multiple-choice and open-ended question-answer pairs. Extensive experiments on 9 representative large video-language models (LVLMs) and several traditional video question answering methods show that most models (including state-of-the-art LVLMs) suffer performance degradation under domain shifts. These results highlight the challenges posed by VUDG and the difference in the robustness of current models to data distribution shifts. We believe VUDG provides a valuable resource for prompting future research in domain generalization video understanding.
Authors: Roger Ferrod, C\'assio F. Dantas, Luigi Di Caro, Dino Ienco
Abstract: Multi-modal RGB and Depth (RGBD) data are predominant in many domains such as robotics, autonomous driving and remote sensing. The combination of these multi-modal data enhances environmental perception by providing 3D spatial context, which is absent in standard RGB images. Although RGBD multi-modal data can be available to train computer vision models, accessing all sensor modalities during the inference stage may be infeasible due to sensor failures or resource constraints, leading to a mismatch between data modalities available during training and inference. Traditional Cross-Modal Knowledge Distillation (CMKD) frameworks, developed to address this task, are typically based on a teacher/student paradigm, where a multi-modal teacher distills knowledge into a single-modality student model. However, these approaches face challenges in teacher architecture choices and distillation process selection, thus limiting their adoption in real-world scenarios. To overcome these issues, we introduce CroDiNo-KD (Cross-Modal Disentanglement: a New Outlook on Knowledge Distillation), a novel cross-modal knowledge distillation framework for RGBD semantic segmentation. Our approach simultaneously learns single-modality RGB and Depth models by exploiting disentanglement representation, contrastive learning and decoupled data augmentation with the aim to structure the internal manifolds of neural network models through interaction and collaboration. We evaluated CroDiNo-KD on three RGBD datasets across diverse domains, considering recent CMKD frameworks as competitors. Our findings illustrate the quality of CroDiNo-KD, and they suggest reconsidering the conventional teacher/student paradigm to distill information from multi-modal data to single-modality neural networks.
Authors: Md Intisar Chowdhury, Kittinun Aukkapinyo, Hiroshi Fujimura, Joo Ann Woo, Wasu Wasusatein, Fadoua Ghourabi
Abstract: In this paper, we propose a Grid-based Local and Global Area Transcription (Grid-LoGAT) system for Video Question Answering (VideoQA). The system operates in two phases. First, extracting text transcripts from video frames using a Vision-Language Model (VLM). Next, processing questions using these transcripts to generate answers through a Large Language Model (LLM). This design ensures image privacy by deploying the VLM on edge devices and the LLM in the cloud. To improve transcript quality, we propose grid-based visual prompting, which extracts intricate local details from each grid cell and integrates them with global information. Evaluation results show that Grid-LoGAT, using the open-source VLM (LLaVA-1.6-7B) and LLM (Llama-3.1-8B), outperforms state-of-the-art methods with similar baseline models on NExT-QA and STAR-QA datasets with an accuracy of 65.9% and 50.11% respectively. Additionally, our method surpasses the non-grid version by 24 points on localization-based questions we created using NExT-QA.
Authors: Yichi Zhang, Gongwei Chen, Jun Zhu, Jia Wan
Abstract: Visual Grounding is a task that aims to localize a target region in an image based on a free-form natural language description. With the rise of Transformer architectures, there is an increasing need for larger datasets to boost performance. However, the high cost of manual annotation poses a challenge, hindering the scale of data and the ability of large models to enhance their effectiveness. Previous pseudo label generation methods heavily rely on human-labeled captions of the original dataset, limiting scalability and diversity. To address this, we propose D2AF, a robust annotation framework for visual grounding using only input images. This approach overcomes dataset size limitations and enriches both the quantity and diversity of referring expressions. Our approach leverages multimodal large models and object detection models. By implementing dual-driven annotation strategies, we effectively generate detailed region-text pairs using both closed-set and open-set approaches. We further conduct an in-depth analysis of data quantity and data distribution. Our findings demonstrate that increasing data volume enhances model performance. However, the degree of improvement depends on how well the pseudo labels broaden the original data distribution. Based on these insights, we propose a consistency and distribution aware filtering method to further improve data quality by effectively removing erroneous and redundant data. This approach effectively eliminates noisy data, leading to improved performance. Experiments on three visual grounding tasks demonstrate that our method significantly improves the performance of existing models and achieves state-of-the-art results.
Authors: Maciej Wielgosz, Simon Berg, Heikki Korpunen, Stephan Hoffmann
Abstract: This paper presents a deep learning-based framework for classifying forestry operations from dashcam video footage. Focusing on four key work elements - crane-out, cutting-and-to-processing, driving, and processing - the approach employs a 3D ResNet-50 architecture implemented with PyTorchVideo. Trained on a manually annotated dataset of field recordings, the model achieves strong performance, with a validation F1 score of 0.88 and precision of 0.90. These results underscore the effectiveness of spatiotemporal convolutional networks for capturing both motion patterns and appearance in real-world forestry environments. The system integrates standard preprocessing and augmentation techniques to improve generalization, but overfitting is evident, highlighting the need for more training data and better class balance. Despite these challenges, the method demonstrates clear potential for reducing the manual workload associated with traditional time studies, offering a scalable solution for operational monitoring and efficiency analysis in forestry. This work contributes to the growing application of AI in natural resource management and sets the foundation for future systems capable of real-time activity recognition in forest machinery. Planned improvements include dataset expansion, enhanced regularization, and deployment trials on embedded systems for in-field use.
Authors: Zheng Wang
Abstract: Fine-grained bird image classification (FBIC) is not only of great significance for ecological monitoring and species identification, but also holds broad research value in the fields of image recognition and fine-grained visual modeling. Compared with general image classification tasks, FBIC poses more formidable challenges: 1) the differences in species size and imaging distance result in the varying sizes of birds presented in the images; 2) complex natural habitats often introduce strong background interference; 3) and highly flexible poses such as flying, perching, or foraging result in substantial intra-class variability. These factors collectively make it difficult for traditional methods to stably extract discriminative features, thereby limiting the generalizability and interpretability of models in real-world applications. To address these challenges, this paper proposes a fine-grained bird classification framework based on strip-aware spatial perception, which aims to capture long-range spatial dependencies across entire rows or columns in bird images, thereby enhancing the model's robustness and interpretability. The proposed method incorporates two novel modules: extensional perception aggregator (EPA) and channel semantic weaving (CSW). Specifically, EPA integrates local texture details with global structural cues by aggregating information across horizontal and vertical spatial directions. CSW further refines the semantic representations by adaptively fusing long-range and short-range information along the channel dimension. Built upon a ResNet-50 backbone, the model enables jump-wise connection of extended structural features across the spatial domain. Experimental results on the CUB-200-2011 dataset demonstrate that our framework achieves significant performance improvements while maintaining architectural efficiency.
Authors: Liangyang Ouyang, Yuki Sakai, Ryosuke Furuta, Hisataka Nozawa, Hikoro Matsui, Yoichi Sato
Abstract: This paper addresses the task of assessing PICU team's leadership skills by developing an automated analysis framework based on egocentric vision. We identify key behavioral cues, including fixation object, eye contact, and conversation patterns, as essential indicators of leadership assessment. In order to capture these multimodal signals, we employ Aria Glasses to record egocentric video, audio, gaze, and head movement data. We collect one-hour videos of four simulated sessions involving doctors with different roles and levels. To automate data processing, we propose a method leveraging REMoDNaV, SAM, YOLO, and ChatGPT for fixation object detection, eye contact detection, and conversation classification. In the experiments, significant correlations are observed between leadership skills and behavioral metrics, i.e., the output of our proposed methods, such as fixation time, transition patterns, and direct orders in speech. These results indicate that our proposed data collection and analysis framework can effectively solve skill assessment for training PICU teams.
Authors: Xianheng Ma, Hongchen Tan, Xiuping Liu, Yi Zhang, Huasheng Wang, Jiang Liu, Ying Chen, Hantao Liu
Abstract: In this paper, we leverage the advantages of event cameras to resist harsh lighting conditions, reduce background interference, achieve high time resolution, and protect facial information to study the long-sequence event-based person re-identification (Re-ID) task. To this end, we propose a simple and efficient long-sequence event Re-ID model, namely the Spike-guided Spatiotemporal Semantic Coupling and Expansion Network (S3CE-Net). To better handle asynchronous event data, we build S3CE-Net based on spiking neural networks (SNNs). The S3CE-Net incorporates the Spike-guided Spatial-temporal Attention Mechanism (SSAM) and the Spatiotemporal Feature Sampling Strategy (STFS). The SSAM is designed to carry out semantic interaction and association in both spatial and temporal dimensions, leveraging the capabilities of SNNs. The STFS involves sampling spatial feature subsequences and temporal feature subsequences from the spatiotemporal dimensions, driving the Re-ID model to perceive broader and more robust effective semantics. Notably, the STFS introduces no additional parameters and is only utilized during the training stage. Therefore, S3CE-Net is a low-parameter and high-efficiency model for long-sequence event-based person Re-ID. Extensive experiments have verified that our S3CE-Net achieves outstanding performance on many mainstream long-sequence event-based person Re-ID datasets. Code is available at:https://github.com/Mhsunshine/SC3E_Net.
Authors: Mika Feng, Koichi Ito, Takafumi Aoki, Tetsushi Ohki, Masakatsu Nishigaki
Abstract: Face recognition systems are designed to be robust against changes in head pose, illumination, and blurring during image capture. If a malicious person presents a face photo of the registered user, they may bypass the authentication process illegally. Such spoofing attacks need to be detected before face recognition. In this paper, we propose a spoofing attack detection method based on Vision Transformer (ViT) to detect minute differences between live and spoofed face images. The proposed method utilizes the intermediate features of ViT, which have a good balance between local and global features that are important for spoofing attack detection, for calculating loss in training and score in inference. The proposed method also introduces two data augmentation methods: face anti-spoofing data augmentation and patch-wise data augmentation, to improve the accuracy of spoofing attack detection. We demonstrate the effectiveness of the proposed method through experiments using the OULU-NPU and SiW datasets.
Authors: Kanokphan Lertniphonphan, Feng Chen, Junda Xu, Fengbu Lan, Jun Xie, Tao Zhang, Zhepeng Wang
Abstract: This report presents our team's PCIE_Interaction solution for the Ego4D Social Interaction Challenge at CVPR 2025, addressing both Looking At Me (LAM) and Talking To Me (TTM) tasks. The challenge requires accurate detection of social interactions between subjects and the camera wearer, with LAM relying exclusively on face crop sequences and TTM combining speaker face crops with synchronized audio segments. In the LAM track, we employ face quality enhancement and ensemble methods. For the TTM task, we extend visual interaction analysis by fusing audio and visual cues, weighted by a visual quality score. Our approach achieved 0.81 and 0.71 mean average precision (mAP) on the LAM and TTM challenges leader board. Code is available at https://github.com/KanokphanL/PCIE_Ego4D_Social_Interaction
URLs: https://github.com/KanokphanL/PCIE_Ego4D_Social_Interaction
Authors: Hanting Wang, Tao Jin, Wang Lin, Shulei Wang, Hai Huang, Shengpeng Ji, Zhou Zhao
Abstract: Bridge models in image restoration construct a diffusion process from degraded to clear images. However, existing methods typically require training a bridge model from scratch for each specific type of degradation, resulting in high computational costs and limited performance. This work aims to efficiently leverage pretrained generative priors within existing image restoration bridges to eliminate this requirement. The main challenge is that standard generative models are typically designed for a diffusion process that starts from pure noise, while restoration tasks begin with a low-quality image, resulting in a mismatch in the state distributions between the two processes. To address this challenge, we propose a transition equation that bridges two diffusion processes with the same endpoint distribution. Based on this, we introduce the IRBridge framework, which enables the direct utilization of generative models within image restoration bridges, offering a more flexible and adaptable approach to image restoration. Extensive experiments on six image restoration tasks demonstrate that IRBridge efficiently integrates generative priors, resulting in improved robustness and generalization performance. Code will be available at GitHub.
Authors: Feng Chen, Kanokphan Lertniphonphan, Qiancheng Yan, Xiaohui Fan, Jun Xie, Tao Zhang, Zhepeng Wang
Abstract: This report introduces our team's (PCIE_EgoPose) solutions for the EgoExo4D Pose and Proficiency Estimation Challenges at CVPR2025. Focused on the intricate task of estimating 21 3D hand joints from RGB egocentric videos, which are complicated by subtle movements and frequent occlusions, we developed the Hand Pose Vision Transformer (HP-ViT+). This architecture synergizes a Vision Transformer and a CNN backbone, using weighted fusion to refine the hand pose predictions. For the EgoExo4D Body Pose Challenge, we adopted a multimodal spatio-temporal feature integration strategy to address the complexities of body pose estimation across dynamic contexts. Our methods achieved remarkable performance: 8.31 PA-MPJPE in the Hand Pose Challenge and 11.25 MPJPE in the Body Pose Challenge, securing championship titles in both competitions. We extended our pose estimation solutions to the Proficiency Estimation task, applying core technologies such as transformer-based architectures. This extension enabled us to achieve a top-1 accuracy of 0.53, a SOTA result, in the Demonstrator Proficiency Estimation competition.
Authors: Runnan Lu, Yuxuan Zhang, Jailing Liu, Haifa Wang, Yiren Song
Abstract: Generating accurate multilingual text with diffusion models has long been desired but remains challenging. Recent methods have made progress in rendering text in a single language, but rendering arbitrary languages is still an unexplored area. This paper introduces EasyText, a text rendering framework based on DiT (Diffusion Transformer), which connects denoising latents with multilingual character tokens encoded as character tokens. We propose character positioning encoding and position encoding interpolation techniques to achieve controllable and precise text rendering. Additionally, we construct a large-scale synthetic text image dataset with 1 million multilingual image-text annotations as well as a high-quality dataset of 20K annotated images, which are used for pretraining and fine-tuning respectively. Extensive experiments and evaluations demonstrate the effectiveness and advancement of our approach in multilingual text rendering, visual quality, and layout-aware text integration.
Authors: Bozhong Zheng, Jinye Gan, Xiaohao Xu, Wenqiao Li, Xiaonan Huang, Na Ni, Yingna Wu
Abstract: 3D point cloud anomaly detection is essential for robust vision systems but is challenged by pose variations and complex geometric anomalies. Existing patch-based methods often suffer from geometric fidelity issues due to discrete voxelization or projection-based representations, limiting fine-grained anomaly localization. We introduce Pose-Aware Signed Distance Field (PASDF), a novel framework that integrates 3D anomaly detection and repair by learning a continuous, pose-invariant shape representation. PASDF leverages a Pose Alignment Module for canonicalization and a SDF Network to dynamically incorporate pose, enabling implicit learning of high-fidelity anomaly repair templates from the continuous SDF. This facilitates precise pixel-level anomaly localization through an Anomaly-Aware Scoring Module. Crucially, the continuous 3D representation in PASDF extends beyond detection, facilitating in-situ anomaly repair. Experiments on Real3D-AD and Anomaly-ShapeNet demonstrate state-of-the-art performance, achieving high object-level AUROC scores of 80.2% and 90.0%, respectively. These results highlight the effectiveness of continuous geometric representations in advancing 3D anomaly detection and facilitating practical anomaly region repair. The code is available at https://github.com/ZZZBBBZZZ/PASDF to support further research.
Authors: Chunxu Liu, Chi Xie, Xiaxu Chen, Wei Li, Feng Zhu, Rui Zhao, Limin Wang
Abstract: Text-to-Image Retrieval (T2IR) is a highly valuable task that aims to match a given textual query to images in a gallery. Existing benchmarks primarily focus on textual queries describing overall image semantics or foreground salient objects, possibly overlooking inconspicuous small objects, especially in complex environments. Such small object retrieval is crucial, as in real-world applications, the targets of interest are not always prominent in the image. Thus, we introduce SORCE (Small Object Retrieval in Complex Environments), a new subfield of T2IR, focusing on retrieving small objects in complex images with textual queries. We propose a new benchmark, SORCE-1K, consisting of images with complex environments and textual queries describing less conspicuous small objects with minimal contextual cues from other salient objects. Preliminary analysis on SORCE-1K finds that existing T2IR methods struggle to capture small objects and encode all the semantics into a single embedding, leading to poor retrieval performance on SORCE-1K. Therefore, we propose to represent each image with multiple distinctive embeddings. We leverage Multimodal Large Language Models (MLLMs) to extract multiple embeddings for each image instructed by a set of Regional Prompts (ReP). Experimental results show that our multi-embedding approach through MLLM and ReP significantly outperforms existing T2IR methods on SORCE-1K. Our experiments validate the effectiveness of SORCE-1K for benchmarking SORCE performances, highlighting the potential of multi-embedding representation and text-customized MLLM features for addressing this task.
Authors: Heejo Kong, Sung-Jin Kim, Gunho Jung, Seong-Whan Lee
Abstract: Conventional semi-supervised learning (SSL) ideally assumes that labeled and unlabeled data share an identical class distribution, however in practice, this assumption is easily violated, as unlabeled data often includes unknown class data, i.e., outliers. The outliers are treated as noise, considerably degrading the performance of SSL models. To address this drawback, we propose a novel framework, Diversify and Conquer (DAC), to enhance SSL robustness in the context of open-set semi-supervised learning. In particular, we note that existing open-set SSL methods rely on prediction discrepancies between inliers and outliers from a single model trained on labeled data. This approach can be easily failed when the labeled data is insufficient, leading to performance degradation that is worse than naive SSL that do not account for outliers. In contrast, our approach exploits prediction disagreements among multiple models that are differently biased towards the unlabeled distribution. By leveraging the discrepancies arising from training on unlabeled data, our method enables robust outlier detection even when the labeled data is underspecified. Our key contribution is constructing a collection of differently biased models through a single training process. By encouraging divergent heads to be differently biased towards outliers while making consistent predictions for inliers, we exploit the disagreement among these heads as a measure to identify unknown concepts. Our code is available at https://github.com/heejokong/DivCon.
Authors: Yingjia Xu, Jinlin Wu, Zhen Chen, Daming Gao, Yang Yang, Zhen Lei, Min Cao
Abstract: Text-based person retrieval aims to identify a target individual from a gallery of images based on a natural language description. It presents a significant challenge due to the complexity of real-world scenes and the ambiguity of appearance-related descriptions. Existing methods primarily emphasize appearance-based cross-modal retrieval, often neglecting the contextual information embedded within the scene, which can offer valuable complementary insights for retrieval. To address this, we introduce SCENEPERSON-13W, a large-scale dataset featuring over 100,000 scenes with rich annotations covering both pedestrian appearance and environmental cues. Based on this, we propose SA-Person, a two-stage retrieval framework. In the first stage, it performs discriminative appearance grounding by aligning textual cues with pedestrian-specific regions. In the second stage, it introduces SceneRanker, a training-free, scene-aware re-ranking method leveraging multimodal large language models to jointly reason over pedestrian appearance and the global scene context. Experiments on SCENEPERSON-13W validate the effectiveness of our framework in challenging scene-level retrieval scenarios. The code and dataset will be made publicly available.
Authors: Cheng Zeng, Xiatian Qi, Chi Chen, Kai Sun, Wangle Zhang, Yuxuan Liu, Yan Meng, Bisheng Yang
Abstract: Transformers have been seldom employed in point cloud roof plane instance segmentation, which is the focus of this study, and existing superpoint Transformers suffer from limited performance due to the use of low-quality superpoints. To address this challenge, we establish two criteria that high-quality superpoints for Transformers should satisfy and introduce a corresponding two-stage superpoint generation process. The superpoints generated by our method not only have accurate boundaries, but also exhibit consistent geometric sizes and shapes, both of which greatly benefit the feature learning of superpoint Transformers. To compensate for the limitations of deep learning features when the training set size is limited, we incorporate multidimensional handcrafted features into the model. Additionally, we design a decoder that combines a Kolmogorov-Arnold Network with a Transformer module to improve instance prediction and mask extraction. Finally, our network's predictions are refined using traditional algorithm-based postprocessing. For evaluation, we annotated a real-world dataset and corrected annotation errors in the existing RoofN3D dataset. Experimental results show that our method achieves state-of-the-art performance on our dataset, as well as both the original and reannotated RoofN3D datasets. Moreover, our model is not sensitive to plane boundary annotations during training, significantly reducing the annotation burden. Through comprehensive experiments, we also identified key factors influencing roof plane segmentation performance: in addition to roof types, variations in point cloud density, density uniformity, and 3D point precision have a considerable impact. These findings underscore the importance of incorporating data augmentation strategies that account for point cloud quality to enhance model robustness under diverse and challenging conditions.
Authors: Yuting Zhang, Hao Lu, Qingyong Hu, Yin Wang, Kaishen Yuan, Xin Liu, Kaishun Wu
Abstract: Periodic or quasi-periodic phenomena reveal intrinsic characteristics in various natural processes, such as weather patterns, movement behaviors, traffic flows, and biological signals. Given that these phenomena span multiple modalities, the capabilities of Multimodal Large Language Models (MLLMs) offer promising potential to effectively capture and understand their complex nature. However, current MLLMs struggle with periodic tasks due to limitations in: 1) lack of temporal modelling and 2) conflict between short and long periods. This paper introduces Period-LLM, a multimodal large language model designed to enhance the performance of periodic tasks across various modalities, and constructs a benchmark of various difficulty for evaluating the cross-modal periodic capabilities of large models. Specially, We adopt an "Easy to Hard Generalization" paradigm, starting with relatively simple text-based tasks and progressing to more complex visual and multimodal tasks, ensuring that the model gradually builds robust periodic reasoning capabilities. Additionally, we propose a "Resisting Logical Oblivion" optimization strategy to maintain periodic reasoning abilities during semantic alignment. Extensive experiments demonstrate the superiority of the proposed Period-LLM over existing MLLMs in periodic tasks. The code is available at https://github.com/keke-nice/Period-LLM.
Authors: Jing Huang, Yongkang Zhao, Yuhan Li, Zhitao Dai, Cheng Chen, Qiying Lai
Abstract: The U-shaped encoder-decoder architecture with skip connections has become a prevailing paradigm in medical image segmentation due to its simplicity and effectiveness. While many recent works aim to improve this framework by designing more powerful encoders and decoders, employing advanced convolutional neural networks (CNNs) for local feature extraction, Transformers or state space models (SSMs) such as Mamba for global context modeling, or hybrid combinations of both, these methods often struggle to fully utilize pretrained vision backbones (e.g., ResNet, ViT, VMamba) due to structural mismatches. To bridge this gap, we introduce ACM-UNet, a general-purpose segmentation framework that retains a simple UNet-like design while effectively incorporating pretrained CNNs and Mamba models through a lightweight adapter mechanism. This adapter resolves architectural incompatibilities and enables the model to harness the complementary strengths of CNNs and SSMs-namely, fine-grained local detail extraction and long-range dependency modeling. Additionally, we propose a hierarchical multi-scale wavelet transform module in the decoder to enhance feature fusion and reconstruction fidelity. Extensive experiments on the Synapse and ACDC benchmarks demonstrate that ACM-UNet achieves state-of-the-art performance while remaining computationally efficient. Notably, it reaches 85.12% Dice Score and 13.89mm HD95 on the Synapse dataset with 17.93G FLOPs, showcasing its effectiveness and scalability. Code is available at: https://github.com/zyklcode/ACM-UNet.
Authors: Anasse Boutayeb, Iyad Lahsen-cherif, Ahmed El Khadimi
Abstract: Object detection has recently seen an interesting trend in terms of the most innovative research work, this task being of particular importance in the field of remote sensing, given the consistency of these images in terms of geographical coverage and the objects present. Furthermore, Deep Learning (DL) models, in particular those based on Transformers, are especially relevant for visual computing tasks in general, and target detection in particular. Thus, the present work proposes an application of Deformable-DETR model, a specific architecture using deformable attention mechanisms, on remote sensing images in two different modes, especially optical and Synthetic Aperture Radar (SAR). To achieve this objective, two datasets are used, one optical, which is Pleiades Aircraft dataset, and the other SAR, in particular SAR Ship Detection Dataset (SSDD). The results of a 10-fold stratified validation showed that the proposed model performed particularly well, obtaining an F1 score of 95.12% for the optical dataset and 94.54% for SSDD, while comparing these results with several models detections, especially those based on CNNs and transformers, as well as those specifically designed to detect different object classes in remote sensing images.
Authors: Ximing Xing, Yandong Guan, Jing Zhang, Dong Xu, Qian Yu
Abstract: Generating high-quality Scalable Vector Graphics (SVGs) is challenging for Large Language Models (LLMs), as it requires advanced reasoning for structural validity, semantic faithfulness, and visual coherence -- capabilities in which current LLMs often fall short. In this work, we introduce Reason-SVG, a novel framework designed to enhance LLM reasoning for SVG generation. Reason-SVG pioneers the "Drawing-with-Thought" (DwT) paradigm, in which models generate both SVG code and explicit design rationales, mimicking the human creative process. Reason-SVG adopts a two-stage training strategy: First, Supervised Fine-Tuning (SFT) trains the LLM on the DwT paradigm to activate foundational reasoning abilities. Second, Reinforcement Learning (RL), utilizing Group Relative Policy Optimization (GRPO), empowers the model to generate both DwT and SVGs rationales through refined, reward-driven reasoning. To facilitate reasoning-driven SVG generation, we design a Hybrid Reward function that evaluates the presence and utility of DwT reasoning, along with structural validity, semantic alignment, and visual quality. We also introduce the SVGX-DwT-10k dataset, a high-quality corpus of 10,000 SVG-DwT pairs, where each SVG code is generated based on explicit DwT reasoning. By integrating DwT, SFT, and Hybrid Reward-guided RL, Reason-SVG significantly improves LLM performance in generating accurate and visually compelling SVGs, potentially fostering "Aha moments" in design.
Authors: Yinqi Li, Jiahe Zhao, Hong Chang, Ruibing Hou, Shiguang Shan, Xilin Chen
Abstract: Contrastive Language-Image Pre-training (CLIP) has become a foundation model and has been applied to various vision and multimodal tasks. However, recent works indicate that CLIP falls short in distinguishing detailed differences in images and shows suboptimal performance on dense-prediction and vision-centric multimodal tasks. Therefore, this work focuses on improving existing CLIP models, aiming to capture as many visual details in images as possible. We find that a specific type of generative models, unCLIP, provides a suitable framework for achieving our goal. Specifically, unCLIP trains an image generator conditioned on the CLIP image embedding. In other words, it inverts the CLIP image encoder. Compared to discriminative models like CLIP, generative models are better at capturing image details because they are trained to learn the data distribution of images. Additionally, the conditional input space of unCLIP aligns with CLIP's original image-text embedding space. Therefore, we propose to invert unCLIP (dubbed un$^2$CLIP) to improve the CLIP model. In this way, the improved image encoder can gain unCLIP's visual detail capturing ability while preserving its alignment with the original text encoder simultaneously. We evaluate our improved CLIP across various tasks to which CLIP has been applied, including the challenging MMVP-VLM benchmark, the dense-prediction open-vocabulary segmentation task, and multimodal large language model tasks. Experiments show that un$^2$CLIP significantly improves the original CLIP and previous CLIP improvement methods. Code and models will be available at https://github.com/LiYinqi/un2CLIP.
Authors: Yuqi Zhang, Yuchun Miao, Zuchao Li, Liang Ding
Abstract: We introduce AMIA, a lightweight, inference-only defense for Large Vision-Language Models (LVLMs) that (1) Automatically Masks a small set of text-irrelevant image patches to disrupt adversarial perturbations, and (2) conducts joint Intention Analysis to uncover and mitigate hidden harmful intents before response generation. Without any retraining, AMIA improves defense success rates across diverse LVLMs and jailbreak benchmarks from an average of 52.4% to 81.7%, preserves general utility with only a 2% average accuracy drop, and incurs only modest inference overhead. Ablation confirms both masking and intention analysis are essential for a robust safety-utility trade-off.
Authors: Yang-Tian Sun, Xin Yu, Zehuan Huang, Yi-Hua Huang, Yuan-Chen Guo, Ziyi Yang, Yan-Pei Cao, Xiaojuan Qi
Abstract: Recently, methods leveraging diffusion model priors to assist monocular geometric estimation (e.g., depth and normal) have gained significant attention due to their strong generalization ability. However, most existing works focus on estimating geometric properties within the camera coordinate system of individual video frames, neglecting the inherent ability of diffusion models to determine inter-frame correspondence. In this work, we demonstrate that, through appropriate design and fine-tuning, the intrinsic consistency of video generation models can be effectively harnessed for consistent geometric estimation. Specifically, we 1) select geometric attributes in the global coordinate system that share the same correspondence with video frames as the prediction targets, 2) introduce a novel and efficient conditioning method by reusing positional encodings, and 3) enhance performance through joint training on multiple geometric attributes that share the same correspondence. Our results achieve superior performance in predicting global geometric attributes in videos and can be directly applied to reconstruction tasks. Even when trained solely on static video data, our approach exhibits the potential to generalize to dynamic video scenes.
Authors: Simone Cammarasana, Giuseppe Patan\`e
Abstract: The paper introduces the weighted convolution, a novel approach to the convolution for signals defined on regular grids (e.g., 2D images) through the application of an optimal density function to scale the contribution of neighbouring pixels based on their distance from the central pixel. This choice differs from the traditional uniform convolution, which treats all neighbouring pixels equally. Our weighted convolution can be applied to convolutional neural network problems to improve the approximation accuracy. Given a convolutional network, we define a framework to compute the optimal density function through a minimisation model. The framework separates the optimisation of the convolutional kernel weights (using stochastic gradient descent) from the optimisation of the density function (using DIRECT-L). Experimental results on a learning model for an image-to-image task (e.g., image denoising) show that the weighted convolution significantly reduces the loss (up to 53% improvement) and increases the test accuracy compared to standard convolution. While this method increases execution time by 11%, it is robust across several hyperparameters of the learning model. Future work will apply the weighted convolution to real-case 2D and 3D image convolutional learning problems.
Authors: Pedram Ghamisi, Weikang Yu, Xiaokang Zhang, Aldino Rizaldy, Jian Wang, Chufeng Zhou, Richard Gloaguen, Gustau Camps-Valls
Abstract: Foundation Models (FMs) are large-scale, pre-trained AI systems that have revolutionized natural language processing and computer vision, and are now advancing geospatial analysis and Earth Observation (EO). They promise improved generalization across tasks, scalability, and efficient adaptation with minimal labeled data. However, despite the rapid proliferation of geospatial FMs, their real-world utility and alignment with global sustainability goals remain underexplored. We introduce SustainFM, a comprehensive benchmarking framework grounded in the 17 Sustainable Development Goals with extremely diverse tasks ranging from asset wealth prediction to environmental hazard detection. This study provides a rigorous, interdisciplinary assessment of geospatial FMs and offers critical insights into their role in attaining sustainability goals. Our findings show: (1) While not universally superior, FMs often outperform traditional approaches across diverse tasks and datasets. (2) Evaluating FMs should go beyond accuracy to include transferability, generalization, and energy efficiency as key criteria for their responsible use. (3) FMs enable scalable, SDG-grounded solutions, offering broad utility for tackling complex sustainability challenges. Critically, we advocate for a paradigm shift from model-centric development to impact-driven deployment, and emphasize metrics such as energy efficiency, robustness to domain shifts, and ethical considerations.
Authors: Xin He, Xumeng Han, Longhui Wei, Lingxi Xie, Qi Tian
Abstract: Multimodal large language models (MLLMs) require a nuanced interpretation of complex image information, typically leveraging a vision encoder to perceive various visual scenarios. However, relying solely on a single vision encoder to handle diverse task domains proves difficult and inevitably leads to conflicts. Recent work enhances data perception by directly integrating multiple domain-specific vision encoders, yet this structure adds complexity and limits the potential for joint optimization. In this paper, we introduce Mixpert, an efficient mixture-of-vision-experts architecture that inherits the joint learning advantages from a single vision encoder while being restructured into a multi-expert paradigm for task-specific fine-tuning across different visual tasks. Additionally, we design a dynamic routing mechanism that allocates input images to the most suitable visual expert. Mixpert effectively alleviates domain conflicts encountered by a single vision encoder in multi-task learning with minimal additional computational cost, making it more efficient than multiple encoders. Furthermore, Mixpert integrates seamlessly into any MLLM, with experimental results demonstrating substantial performance gains across various tasks.
Authors: Simone Cammarasana, Giuseppe Patan\`e
Abstract: We introduce a novel weighted convolution operator that enhances traditional convolutional neural networks (CNNs) by integrating a spatial density function into the convolution operator. This extension enables the network to differentially weight neighbouring pixels based on their relative position to the reference pixel, improving spatial characterisation and feature extraction. The proposed operator maintains the same number of trainable parameters and is fully compatible with existing CNN architectures. Although developed for 2D image data, the framework is generalisable to signals on regular grids of arbitrary dimensions, such as 3D volumetric data or 1D time series. We propose an efficient implementation of the weighted convolution by pre-computing the density function and achieving execution times comparable to standard convolution layers. We evaluate our method on two deep learning tasks: image classification using the CIFAR-100 dataset [KH+09] and image denoising using the DIV2K dataset [AT17]. Experimental results with state-of-the-art classification (e.g., VGG [SZ15], ResNet [HZRS16]) and denoising (e.g., DnCNN [ZZC+17], NAFNet [CCZS22]) methods show that the weighted convolution improves performance with respect to standard convolution across different quantitative metrics. For example, VGG achieves an accuracy of 66.94% with weighted convolution versus 56.89% with standard convolution on the classification problem, while DnCNN improves the PSNR value from 20.17 to 22.63 on the denoising problem. All models were trained on the CINECA Leonardo cluster to reduce the execution time and improve the tuning of the density function values. The PyTorch implementation of the weighted convolution is publicly available at: https://github.com/cammarasana123/weightedConvolution2.0.
URLs: https://github.com/cammarasana123/weightedConvolution2.0.
Authors: Qinghe Ma, Jian Zhang, Lei Qi, Qian Yu, Yinghuan Shi, Yang Gao
Abstract: Both limited annotation and domain shift are prevalent challenges in medical image segmentation. Traditional semi-supervised segmentation and unsupervised domain adaptation methods address one of these issues separately. However, the coexistence of limited annotation and domain shift is quite common, which motivates us to introduce a novel and challenging scenario: Mixed Domain Semi-supervised medical image Segmentation (MiDSS), where limited labeled data from a single domain and a large amount of unlabeled data from multiple domains. To tackle this issue, we propose the UST-RUN framework, which fully leverages intermediate domain information to facilitate knowledge transfer. We employ Unified Copy-paste (UCP) to construct intermediate domains, and propose a Symmetric GuiDance training strategy (SymGD) to supervise unlabeled data by merging pseudo-labels from intermediate samples. Subsequently, we introduce a Training Process aware Random Amplitude MixUp (TP-RAM) to progressively incorporate style-transition components into intermediate samples. To generate more diverse intermediate samples, we further select reliable samples with high-quality pseudo-labels, which are then mixed with other unlabeled data. Additionally, we generate sophisticated intermediate samples with high-quality pseudo-labels for unreliable samples, ensuring effective knowledge transfer for them. Extensive experiments on four public datasets demonstrate the superiority of UST-RUN. Notably, UST-RUN achieves a 12.94% improvement in Dice score on the Prostate dataset. Our code is available at https://github.com/MQinghe/UST-RUN
Authors: Omer Nacar, Yasser Al-Habashi, Serry Sibaee, Adel Ammar, Wadii Boulila
Abstract: Arabic Optical Character Recognition (OCR) is essential for converting vast amounts of Arabic print media into digital formats. However, training modern OCR models, especially powerful vision-language models, is hampered by the lack of large, diverse, and well-structured datasets that mimic real-world book layouts. Existing Arabic OCR datasets often focus on isolated words or lines or are limited in scale, typographic variety, or structural complexity found in books. To address this significant gap, we introduce SARD (Large-Scale Synthetic Arabic OCR Dataset). SARD is a massive, synthetically generated dataset specifically designed to simulate book-style documents. It comprises 843,622 document images containing 690 million words, rendered across ten distinct Arabic fonts to ensure broad typographic coverage. Unlike datasets derived from scanned documents, SARD is free from real-world noise and distortions, offering a clean and controlled environment for model training. Its synthetic nature provides unparalleled scalability and allows for precise control over layout and content variation. We detail the dataset's composition and generation process and provide benchmark results for several OCR models, including traditional and deep learning approaches, highlighting the challenges and opportunities presented by this dataset. SARD serves as a valuable resource for developing and evaluating robust OCR and vision-language models capable of processing diverse Arabic book-style texts.
Authors: Panagiotis Rigas, Panagiotis Drivas, Charalambos Tzamos, Ioannis Chamodrakas, George Ioannakis, Leonidas J. Guibas, Ioannis Z. Emiris
Abstract: We introduce GARLIC (GAussian Representation LearnIng for spaCe partitioning), a novel indexing structure based on \(N\)-dimensional Gaussians for efficiently learning high-dimensional vector spaces. Our approach is inspired from Gaussian splatting techniques, typically used in 3D rendering, which we adapt for high-dimensional search and classification. We optimize Gaussian parameters using information-theoretic objectives that balance coverage, assignment confidence, and structural and semantic consistency. A key contribution is to progressively refine the representation through split and clone operations, handling hundreds of dimensions, thus handling varying data densities. GARLIC offers the fast building times of traditional space partitioning methods (e.g., under \(\sim5\) min build time for SIFT1M) while achieving \(\sim50\%\) Recall10@10 in low-candidate regimes. Experimental results on standard benchmarks demonstrate our method's consistency in (a) \(k\)-NN retrieval, outperforming methods, such as Faiss-IVF, in fast-recall by using about half their probes for the same Recall10@10 in Fashion-MNIST, and (b) in classification tasks, beating by \(\sim15\%\) accuracy other majority voting methods. Further, we show strong generalization capabilities, maintaining high accuracy even with downsampled training data: using just \(1\%\) of the training data returns \(\sim 45\%\) Recall@1, thus making GARLIC quite powerful for applications requiring both speed and accuracy.
Authors: Duo Zheng, Shijia Huang, Yanyang Li, Liwei Wang
Abstract: Previous research has investigated the application of Multimodal Large Language Models (MLLMs) in understanding 3D scenes by interpreting them as videos. These approaches generally depend on comprehensive 3D data inputs, such as point clouds or reconstructed Bird's-Eye View (BEV) maps. In our research, we advance this field by enhancing the capability of MLLMs to understand and reason in 3D spaces directly from video data, without the need for additional 3D input. We propose a novel and efficient method, the Video-3D Geometry Large Language Model (VG LLM). Our approach employs a 3D visual geometry encoder that extracts 3D prior information from video sequences. This information is integrated with visual tokens and fed into the MLLM. Extensive experiments have shown that our method has achieved substantial improvements in various tasks related to 3D scene understanding and spatial reasoning, all directly learned from video sources. Impressively, our 4B model, which does not rely on explicit 3D data inputs, achieves competitive results compared to existing state-of-the-art methods, and even surpasses the Gemini-1.5-Pro in the VSI-Bench evaluations.
Authors: Xuzhi Wang, Wei Feng, Lingdong Kong, Liang Wan
Abstract: LiDAR semantic segmentation plays a vital role in autonomous driving. Existing voxel-based methods for LiDAR semantic segmentation apply uniform partition to the 3D LiDAR point cloud to form a structured representation based on cartesian/cylindrical coordinates. Although these methods show impressive performance, the drawback of existing voxel-based methods remains in two aspects: (1) it requires a large enough input voxel resolution, which brings a large amount of computation cost and memory consumption. (2) it does not well handle the unbalanced point distribution of LiDAR point cloud. In this paper, we propose a non-uniform cylindrical partition network named NUC-Net to tackle the above challenges. Specifically, we propose the Arithmetic Progression of Interval (API) method to non-uniformly partition the radial axis and generate the voxel representation which is representative and efficient. Moreover, we propose a non-uniform multi-scale aggregation method to improve contextual information. Our method achieves state-of-the-art performance on SemanticKITTI and nuScenes datasets with much faster speed and much less training time. And our method can be a general component for LiDAR semantic segmentation, which significantly improves both the accuracy and efficiency of the uniform counterpart by $4 \times$ training faster and $2 \times$ GPU memory reduction and $3 \times$ inference speedup. We further provide theoretical analysis towards understanding why NUC is effective and how point distribution affects performance. Code is available at \href{https://github.com/alanWXZ/NUC-Net}{https://github.com/alanWXZ/NUC-Net}.
URLs: https://github.com/alanWXZ/NUC-Net, https://github.com/alanWXZ/NUC-Net
Authors: Marios Glytsos, Panagiotis P. Filntisis, George Retsinas, Petros Maragos
Abstract: Accurate 6D object pose estimation is essential for robotic grasping and manipulation, particularly in agriculture, where fruits and vegetables exhibit high intra-class variability in shape, size, and texture. The vast majority of existing methods rely on instance-specific CAD models or require depth sensors to resolve geometric ambiguities, making them impractical for real-world agricultural applications. In this work, we introduce PLANTPose, a novel framework for category-level 6D pose estimation that operates purely on RGB input. PLANTPose predicts both the 6D pose and deformation parameters relative to a base mesh, allowing a single category-level CAD model to adapt to unseen instances. This enables accurate pose estimation across varying shapes without relying on instance-specific data. To enhance realism and improve generalization, we also leverage Stable Diffusion to refine synthetic training images with realistic texturing, mimicking variations due to ripeness and environmental factors and bridging the domain gap between synthetic data and the real world. Our evaluations on a challenging benchmark that includes bananas of various shapes, sizes, and ripeness status demonstrate the effectiveness of our framework in handling large intraclass variations while maintaining accurate 6D pose predictions, significantly outperforming the state-of-the-art RGB-based approach MegaPose.
Authors: Zahid Hassan Tushar, Adeleke Ademakinwa, Jianwu Wang, Zhibo Zhang, Sanjay Purushotham
Abstract: Cloud Optical Thickness (COT) is a critical cloud property influencing Earth's climate, weather, and radiation budget. Satellite radiance measurements enable global COT retrieval, but challenges like 3D cloud effects, viewing angles, and atmospheric interference must be addressed to ensure accurate estimation. Traditionally, the Independent Pixel Approximation (IPA) method, which treats individual pixels independently, has been used for COT estimation. However, IPA introduces significant bias due to its simplified assumptions. Recently, deep learning-based models have shown improved performance over IPA but lack robustness, as they are sensitive to variations in radiance intensity, distortions, and cloud shadows. These models also introduce substantial errors in COT estimation under different solar and viewing zenith angles. To address these challenges, we propose a novel angle-invariant, attention-based deep model called Cloud-Attention-Net with Angle Coding (CAAC). Our model leverages attention mechanisms and angle embeddings to account for satellite viewing geometry and 3D radiative transfer effects, enabling more accurate retrieval of COT. Additionally, our multi-angle training strategy ensures angle invariance. Through comprehensive experiments, we demonstrate that CAAC significantly outperforms existing state-of-the-art deep learning models, reducing cloud property retrieval errors by at least a factor of nine.
Authors: Chengzhi Wu, Qianliang Huang, Kun Jin, Julius Pfrommer, J\"urgen Beyerer
Abstract: Contrastive learning is an essential method in self-supervised learning. It primarily employs a multi-branch strategy to compare latent representations obtained from different branches and train the encoder. In the case of multi-modal input, diverse modalities of the same object are fed into distinct branches. When using single-modal data, the same input undergoes various augmentations before being fed into different branches. However, all existing contrastive learning frameworks have so far only performed contrastive operations on the learned features at the final loss end, with no information exchange between different branches prior to this stage. In this paper, for point cloud unsupervised learning without the use of extra training data, we propose a Contrastive Cross-branch Attention-based framework for Point cloud data (termed PoCCA), to learn rich 3D point cloud representations. By introducing sub-branches, PoCCA allows information exchange between different branches before the loss end. Experimental results demonstrate that in the case of using no extra training data, the representations learned with our self-supervised model achieve state-of-the-art performances when used for downstream tasks on point clouds.
Authors: Huu-Thien Tran, Thanh-Dat Truong, Khoa Luu
Abstract: Large vision-language models have become widely adopted to advance in various domains. However, developing a trustworthy system with minimal interpretable characteristics of large-scale models presents a significant challenge. One of the most prevalent terms associated with the fallacy functions caused by these systems is hallucination, where the language model generates a response that does not correspond to the visual content. To mitigate this problem, several approaches have been developed, and one prominent direction is to ameliorate the decoding process. In this paper, we propose a new Bijective Maximum Likelihood Learning (BIMA) approach to hallucination mitigation using normalizing flow theories. The proposed BIMA method can efficiently mitigate the hallucination problem in prevailing vision-language models, resulting in significant improvements. Notably, BIMA achieves the average F1 score of 85.06% on POPE benchmark and remarkably reduce CHAIRS and CHAIRI by 7.6% and 2.6%, respectively. To the best of our knowledge, this is one of the first studies that contemplates the bijection means to reduce hallucination induced by large vision-language models.
Authors: Jiahe Chen, Jiahe Ying, Shen Wang, Jianwei Zheng
Abstract: Confronting the critical challenge of insufficiently annotated samples in medical domain, semi-supervised medical image segmentation (SSMIS) emerges as a promising solution. Specifically, most methodologies following the Mean Teacher (MT) or Dual Students (DS) architecture have achieved commendable results. However, to date, these approaches face a performance bottleneck due to two inherent limitations, \textit{e.g.}, the over-coupling problem within MT structure owing to the employment of exponential moving average (EMA) mechanism, as well as the severe cognitive bias between two students of DS structure, both of which potentially lead to reduced efficacy, or even model collapse eventually. To mitigate these issues, a Decoupled Competitive Framework (DCF) is elaborated in this work, which utilizes a straightforward competition mechanism for the update of EMA, effectively decoupling students and teachers in a dynamical manner. In addition, the seamless exchange of invaluable and precise insights is facilitated among students, guaranteeing a better learning paradigm. The DCF introduced undergoes rigorous validation on three publicly accessible datasets, which encompass both 2D and 3D datasets. The results demonstrate the superiority of our method over previous cutting-edge competitors. Code will be available at https://github.com/JiaheChen2002/DCF.
Authors: Chengzhi Wu, Hao Fu, Jan-Philipp Kaiser, Erik Tabuchi Barczak, Julius Pfrommer, Gisela Lanza, Michael Heizmann, J\"urgen Beyerer
Abstract: The accurate estimation of 6D pose remains a challenging task within the computer vision domain, even when utilizing 3D point cloud data. Conversely, in the manufacturing domain, instances arise where leveraging prior knowledge can yield advancements in this endeavor. This study focuses on the disassembly of starter motors to augment the engineering of product life cycles. A pivotal objective in this context involves the identification and 6D pose estimation of bolts affixed to the motors, facilitating automated disassembly within the manufacturing workflow. Complicating matters, the presence of occlusions and the limitations of single-view data acquisition, notably when motors are placed in a clamping system, obscure certain portions and render some bolts imperceptible. Consequently, the development of a comprehensive pipeline capable of acquiring complete bolt information is imperative to avoid oversight in bolt detection. In this paper, employing the task of bolt detection within the scope of our project as a pertinent use case, we introduce a meticulously devised pipeline. This multi-stage pipeline effectively captures the 6D information with regard to all bolts on the motor, thereby showcasing the effective utilization of prior knowledge in handling this challenging task. The proposed methodology not only contributes to the field of 6D pose estimation but also underscores the viability of integrating domain-specific insights to tackle complex problems in manufacturing and automation.
Authors: Evangelos Sariyanidi, Lisa Yankowitz, Robert T. Schultz, John D. Herrington, Birkan Tunc, Jeffrey Cohn
Abstract: The Facial Action Coding System (FACS) has been used by numerous studies to investigate the links between facial behavior and mental health. The laborious and costly process of FACS coding has motivated the development of machine learning frameworks for Action Unit (AU) detection. Despite intense efforts spanning three decades, the detection accuracy for many AUs is considered to be below the threshold needed for behavioral research. Also, many AUs are excluded altogether, making it impossible to fulfill the ultimate goal of FACS-the representation of any facial expression in its entirety. This paper considers an alternative approach. Instead of creating automated tools that mimic FACS experts, we propose to use a new coding system that mimics the key properties of FACS. Specifically, we construct a data-driven coding system called the Facial Basis, which contains units that correspond to localized and interpretable 3D facial movements, and overcomes three structural limitations of automated FACS coding. First, the proposed method is completely unsupervised, bypassing costly, laborious and variable manual annotation. Second, Facial Basis reconstructs all observable movement, rather than relying on a limited repertoire of recognizable movements (as in automated FACS). Finally, the Facial Basis units are additive, whereas AUs may fail detection when they appear in a non-additive combination. The proposed method outperforms the most frequently used AU detector in predicting autism diagnosis from in-person and remote conversations, highlighting the importance of encoding facial behavior comprehensively. To our knowledge, Facial Basis is the first alternative to FACS for deconstructing facial expressions in videos into localized movements. We provide an open source implementation of the method at github.com/sariyanidi/FacialBasis.
Authors: Simone Alberto Peirone, Francesca Pistilli, Antonio Alliegro, Tatiana Tommasi, Giuseppe Averta
Abstract: Our comprehension of video streams depicting human activities is naturally multifaceted: in just a few moments, we can grasp what is happening, identify the relevance and interactions of objects in the scene, and forecast what will happen soon, everything all at once. To endow autonomous systems with such holistic perception, learning how to correlate concepts, abstract knowledge across diverse tasks, and leverage tasks synergies when learning novel skills is essential. In this paper, we introduce Hier-EgoPack, a unified framework able to create a collection of task perspectives that can be carried across downstream tasks and used as a potential source of additional insights, as a backpack of skills that a robot can carry around and use when needed.
Authors: Julio Silva-Rodr\'iguez, Ismail Ben Ayed, Jose Dolz
Abstract: Vision-language models pre-trained at large scale have shown unprecedented adaptability and generalization to downstream tasks. Although its discriminative potential has been widely explored, its reliability and uncertainty are still overlooked. In this work, we investigate the capabilities of CLIP models under the split conformal prediction paradigm, which provides theoretical guarantees to black-box models based on a small, labeled calibration set. In contrast to the main body of literature on conformal predictors in vision classifiers, foundation models exhibit a particular characteristic: they are pre-trained on a one-time basis on an inaccessible source domain, different from the transferred task. This domain drift negatively affects the efficiency of the conformal sets and poses additional challenges. To alleviate this issue, we propose Conf-OT, a transfer learning setting that operates transductive over the combined calibration and query sets. Solving an optimal transport problem, the proposed method bridges the domain gap between pre-training and adaptation without requiring additional data splits but still maintaining coverage guarantees. We comprehensively explore this conformal prediction strategy on a broad span of 15 datasets and three non-conformity scores. Conf-OT provides consistent relative improvements of up to 20% on set efficiency while being 15 times faster than popular transductive approaches.
Authors: Raman Jha, Adithya Lenka, Mani Ramanagopal, Aswin Sankaranarayanan, Kaushik Mitra
Abstract: In nighttime conditions, high noise levels and bright illumination sources degrade image quality, making low-light image enhancement challenging. Thermal images provide complementary information, offering richer textures and structural details. We propose RT-X Net, a cross-attention network that fuses RGB and thermal images for nighttime image enhancement. We leverage self-attention networks for feature extraction and a cross-attention mechanism for fusion to effectively integrate information from both modalities. To support research in this domain, we introduce the Visible-Thermal Image Enhancement Evaluation (V-TIEE) dataset, comprising 50 co-located visible and thermal images captured under diverse nighttime conditions. Extensive evaluations on the publicly available LLVIP dataset and our V-TIEE dataset demonstrate that RT-X Net outperforms state-of-the-art methods in low-light image enhancement. The code and the V-TIEE can be found here https://github.com/jhakrraman/rt-xnet.
Authors: Jisheng Dang, Jingze Wu, Teng Wang, Xuanhui Lin, Nannan Zhu, Hongbo Chen, Wei-Shi Zheng, Meng Wang, Tat-Seng Chua
Abstract: Recent advancements in reinforcement learning, particularly through Group Relative Policy Optimization (GRPO), have significantly improved multimodal large language models for complex reasoning tasks. However, two critical limitations persist: 1) they often produce unfocused, verbose reasoning chains that obscure salient spatiotemporal cues and 2) binary rewarding fails to account for partially correct answers, resulting in high reward variance and inefficient learning. In this paper, we propose TW-GRPO, a novel framework that enhances visual reasoning with focused thinking and dense reward granularity. Specifically, we employs a token weighting mechanism that prioritizes tokens with high informational density (estimated by intra-group variance), suppressing redundant tokens like generic reasoning prefixes. Furthermore, we reformulate RL training by shifting from single-choice to multi-choice QA tasks, where soft rewards enable finer-grained gradient estimation by distinguishing partial correctness. Additionally, we propose question-answer inversion, a data augmentation strategy to generate diverse multi-choice samples from existing benchmarks. Experiments demonstrate state-of-the-art performance on several video reasoning and general understanding benchmarks. Notably, TW-GRPO achieves 50.4\% accuracy on CLEVRER (18.8\% improvement over Video-R1) and 65.8\% on MMVU. Our codes are available at \href{https://github.com/longmalongma/TW-GRPO}{https://github.com/longmalongma/TW-GRPO}.
URLs: https://github.com/longmalongma/TW-GRPO, https://github.com/longmalongma/TW-GRPO
Authors: Jiaxu Zhang, Xianfang Zeng, Xin Chen, Wei Zuo, Gang Yu, Guosheng Lin, Zhigang Tu
Abstract: This paper presents DreamDance, a novel character art animation framework capable of producing stable, consistent character and scene motion conditioned on precise camera trajectories. To achieve this, we re-formulate the animation task as two inpainting-based steps: Camera-aware Scene Inpainting and Pose-aware Video Inpainting. The first step leverages a pre-trained image inpainting model to generate multi-view scene images from the reference art and optimizes a stable large-scale Gaussian field, which enables coarse background video rendering with camera trajectories. However, the rendered video is rough and only conveys scene motion. To resolve this, the second step trains a pose-aware video inpainting model that injects the dynamic character into the scene video while enhancing background quality. Specifically, this model is a DiT-based video generation model with a gating strategy that adaptively integrates the character's appearance and pose information into the base background video. Through extensive experiments, we demonstrate the effectiveness and generalizability of DreamDance, producing high-quality and consistent character animations with remarkable camera dynamics.
Authors: Jiazhong Cen, Xudong Zhou, Jiemin Fang, Changsong Wen, Lingxi Xie, Xiaopeng Zhang, Wei Shen, Qi Tian
Abstract: Recent advancements in 3D Gaussian Splatting (3D-GS) enable high-quality 3D scene reconstruction from RGB images. Many studies extend this paradigm for language-driven open-vocabulary scene understanding. However, most of them simply project 2D semantic features onto 3D Gaussians and overlook a fundamental gap between 2D and 3D understanding: a 3D object may exhibit various semantics from different viewpoints--a phenomenon we term view-dependent semantics. To address this challenge, we propose LaGa (Language Gaussians), which establishes cross-view semantic connections by decomposing the 3D scene into objects. Then, it constructs view-aggregated semantic representations by clustering semantic descriptors and reweighting them based on multi-view semantics. Extensive experiments demonstrate that LaGa effectively captures key information from view-dependent semantics, enabling a more comprehensive understanding of 3D scenes. Notably, under the same settings, LaGa achieves a significant improvement of +18.7% mIoU over the previous SOTA on the LERF-OVS dataset. Our code is available at: https://github.com/SJTU-DeepVisionLab/LaGa.
Authors: Yucheng Zhou, Jiahao Yuan, Qianning Wang
Abstract: Recent advancements in text-to-image (T2I) generation have enabled models to produce high-quality images from textual descriptions. However, these models often struggle with complex instructions involving multiple objects, attributes, and spatial relationships. Existing benchmarks for evaluating T2I models primarily focus on general text-image alignment and fail to capture the nuanced requirements of complex, multi-faceted prompts. Given this gap, we introduce LongBench-T2I, a comprehensive benchmark specifically designed to evaluate T2I models under complex instructions. LongBench-T2I consists of 500 intricately designed prompts spanning nine diverse visual evaluation dimensions, enabling a thorough assessment of a model's ability to follow complex instructions. Beyond benchmarking, we propose an agent framework (Plan2Gen) that facilitates complex instruction-driven image generation without requiring additional model training. This framework integrates seamlessly with existing T2I models, using large language models to interpret and decompose complex prompts, thereby guiding the generation process more effectively. As existing evaluation metrics, such as CLIPScore, fail to adequately capture the nuances of complex instructions, we introduce an evaluation toolkit that automates the quality assessment of generated images using a set of multi-dimensional metrics. The data and code are released at https://github.com/yczhou001/LongBench-T2I.
Authors: Xinliu Zhong, Leo Hwa Liang, Angela S. Koh, Yeo Si Yong
Abstract: Traditional diagnostic methods like colonoscopy are invasive yet critical tools necessary for accurately diagnosing colorectal cancer (CRC). Detection of CRC at early stages is crucial for increasing patient survival rates. However, colonoscopy is dependent on obtaining adequate and high-quality endoscopic images. Prolonged invasive procedures are inherently risky for patients, while suboptimal or insufficient images hamper diagnostic accuracy. These images, typically derived from video frames, often exhibit similar patterns, posing challenges in discrimination. To overcome these challenges, we propose a novel Deep Learning network built on a Few-Shot Learning architecture, which includes a tailored feature extractor, task interpolation, relational embedding, and a bi-level routing attention mechanism. The Few-Shot Learning paradigm enables our model to rapidly adapt to unseen fine-grained endoscopic image patterns, and the task interpolation augments the insufficient images artificially from varied instrument viewpoints. Our relational embedding approach discerns critical intra-image features and captures inter-image transitions between consecutive endoscopic frames, overcoming the limitations of Convolutional Neural Networks (CNNs). The integration of a light-weight attention mechanism ensures a concentrated analysis of pertinent image regions. By training on diverse datasets, the model's generalizability and robustness are notably improved for handling endoscopic images. Evaluated on Kvasir dataset, our model demonstrated superior performance, achieving an accuracy of 90.1\%, precision of 0.845, recall of 0.942, and an F1 score of 0.891. This surpasses current state-of-the-art methods, presenting a promising solution to the challenges of invasive colonoscopy by optimizing CRC detection through advanced image analysis.
Authors: Jiangpeng He, Zhihao Duan, Fengqing Zhu
Abstract: Class-Incremental Learning (CIL) aims to learn new classes sequentially while retaining the knowledge of previously learned classes. Recently, pre-trained models (PTMs) combined with parameter-efficient fine-tuning (PEFT) have shown remarkable performance in rehearsal-free CIL without requiring exemplars from previous tasks. However, existing adapter-based methods, which incorporate lightweight learnable modules into PTMs for CIL, create new adapters for each new task, leading to both parameter redundancy and failure to leverage shared knowledge across tasks. In this work, we propose ContinuaL Low-Rank Adaptation (CL-LoRA), which introduces a novel dual-adapter architecture combining \textbf{task-shared adapters} to learn cross-task knowledge and \textbf{task-specific adapters} to capture unique features of each new task. Specifically, the shared adapters utilize random orthogonal matrices and leverage knowledge distillation with gradient reassignment to preserve essential shared knowledge. In addition, we introduce learnable block-wise weights for task-specific adapters, which mitigate inter-task interference while maintaining the model's plasticity. We demonstrate CL-LoRA consistently achieves promising performance under multiple benchmarks with reduced training and inference computation, establishing a more efficient and scalable paradigm for continual learning with pre-trained models.
Authors: Marta L\'opez-Rauhut, Hongyu Zhou, Mathieu Aubry, Loic Landrieu
Abstract: Historical maps offer an invaluable perspective into territory evolution across past centuries--long before satellite or remote sensing technologies existed. Deep learning methods have shown promising results in segmenting historical maps, but publicly available datasets typically focus on a single map type or period, require extensive and costly annotations, and are not suited for nationwide, long-term analyses. In this paper, we introduce a new dataset of historical maps tailored for analyzing large-scale, long-term land use and land cover evolution with limited annotations. Spanning metropolitan France (548,305 km^2), our dataset contains three map collections from the 18th, 19th, and 20th centuries. We provide both comprehensive modern labels and 22,878 km^2 of manually annotated historical labels for the 18th and 19th century maps. Our dataset illustrates the complexity of the segmentation task, featuring stylistic inconsistencies, interpretive ambiguities, and significant landscape changes (e.g., marshlands disappearing in favor of forests). We assess the difficulty of these challenges by benchmarking three approaches: a fully-supervised model trained with historical labels, and two weakly-supervised models that rely only on modern annotations. The latter either use the modern labels directly or first perform image-to-image translation to address the stylistic gap between historical and contemporary maps. Finally, we discuss how these methods can support long-term environment monitoring, offering insights into centuries of landscape transformation. Our official project repository is publicly available at https://github.com/Archiel19/FRAx4.git.
Authors: Yinglian Zhu, Haiyang Yu, Qizao Wang, Wei Lu, Xiangyang Xue, Bin Li
Abstract: Chinese Character Recognition (CCR) is a fundamental technology for intelligent document processing. Unlike Latin characters, Chinese characters exhibit unique spatial structures and compositional rules, allowing for the use of fine-grained semantic information in representation. However, existing approaches are usually based on auto-regressive as well as edit distance post-process and typically rely on a single-level character representation. In this paper, we propose a Hierarchical Multi-Granularity Image-Text Aligning (Hi-GITA) framework based on a contrastive paradigm. To leverage the abundant fine-grained semantic information of Chinese characters, we propose multi-granularity encoders on both image and text sides. Specifically, the Image Multi-Granularity Encoder extracts hierarchical image representations from character images, capturing semantic cues from localized strokes to holistic structures. The Text Multi-Granularity Encoder extracts stroke and radical sequence representations at different levels of granularity. To better capture the relationships between strokes and radicals, we introduce Multi-Granularity Fusion Modules on the image and text sides, respectively. Furthermore, to effectively bridge the two modalities, we further introduce a Fine-Grained Decoupled Image-Text Contrastive loss, which aligns image and text representations across multiple granularities. Extensive experiments demonstrate that our proposed Hi-GITA significantly outperforms existing zero-shot CCR methods. For instance, it brings about 20% accuracy improvement in handwritten character and radical zero-shot settings. Code and models will be released soon.
Authors: Brandon Man, Ghadi Nehme, Md Ferdous Alam, Faez Ahmed
Abstract: Computer-Aided Design (CAD) is a time-consuming and complex process, requiring precise, long-horizon user interactions with intricate 3D interfaces. While recent advances in AI-driven user interface (UI) agents show promise, most existing datasets and methods focus on short, low-complexity tasks in mobile or web applications, failing to capture the demands of professional engineering tools. In this work, we introduce VideoCAD, the first attempt at engineering UI interaction learning for precision tasks. Specifically, VideoCAD is a large-scale synthetic dataset consisting of over 41K annotated video recordings of CAD operations, generated using an automated framework for collecting high-fidelity UI action data from human-made CAD designs. Compared to existing datasets, VideoCAD offers an order of magnitude higher complexity in UI interaction learning for real-world engineering tasks, having up to a 20x longer time horizon than other datasets. We show two important downstream applications of VideoCAD: learning UI interactions from professional precision 3D CAD tools and a visual question-answering (VQA) benchmark designed to evaluate multimodal large language models' (LLM) spatial reasoning and video understanding abilities. To learn the UI interactions, we propose VideoCADFormer - a state-of-the-art model in learning CAD interactions directly from video, which outperforms multiple behavior cloning baselines. Both VideoCADFormer and the VQA benchmark derived from VideoCAD reveal key challenges in the current state of video-based UI understanding, including the need for precise action grounding, multi-modal and spatial reasoning, and long-horizon dependencies.
Authors: Yuwen Tan, Yuan Qing, Boqing Gong
Abstract: This paper reveals that many state-of-the-art large language models (LLMs) lack hierarchical knowledge about our visual world, unaware of even well-established biology taxonomies. This shortcoming makes LLMs a bottleneck for vision LLMs' hierarchical visual understanding (e.g., recognizing Anemone Fish but not Vertebrate). We arrive at these findings using about one million four-choice visual question answering (VQA) tasks constructed from six taxonomies and four image datasets. Interestingly, finetuning a vision LLM using our VQA tasks reaffirms LLMs' bottleneck effect to some extent because the VQA tasks improve the LLM's hierarchical consistency more than the vision LLM's. We conjecture that one cannot make vision LLMs understand visual concepts fully hierarchical until LLMs possess corresponding taxonomy knowledge.
Authors: Charig Yang, Samiul Alam, Shakhrul Iman Siam, Michael J. Proulx, Lambert Mathias, Kiran Somasundaram, Luis Pesqueira, James Fort, Sheroze Sheriffdeen, Omkar Parkhi, Carl Ren, Mi Zhang, Yuning Chai, Richard Newcombe, Hyo Jin Kim
Abstract: To enable egocentric contextual AI in always-on smart glasses, it is crucial to be able to keep a record of the user's interactions with the world, including during reading. In this paper, we introduce a new task of reading recognition to determine when the user is reading. We first introduce the first-of-its-kind large-scale multimodal Reading in the Wild dataset, containing 100 hours of reading and non-reading videos in diverse and realistic scenarios. We then identify three modalities (egocentric RGB, eye gaze, head pose) that can be used to solve the task, and present a flexible transformer model that performs the task using these modalities, either individually or combined. We show that these modalities are relevant and complementary to the task, and investigate how to efficiently and effectively encode each modality. Additionally, we show the usefulness of this dataset towards classifying types of reading, extending current reading understanding studies conducted in constrained settings to larger scale, diversity and realism. Code, model, and data will be public.
Authors: Cailin Zhuang, Ailin Huang, Wei Cheng, Jingwei Wu, Yaoqi Hu, Jiaqi Liao, Zhewei Huang, Hongyuan Wang, Xinyao Liao, Weiwei Cai, Hengyuan Xu, Xuanyang Zhang, Xianfang Zeng, Gang Yu, Chi Zhang
Abstract: Story visualization, which aims to generate a sequence of visually coherent images aligning with a given narrative and reference images, has seen significant progress with recent advancements in generative models. To further enhance the performance of story visualization frameworks in real-world scenarios, we introduce a comprehensive evaluation benchmark, ViStoryBench. We collect a diverse dataset encompassing various story types and artistic styles, ensuring models are evaluated across multiple dimensions such as different plots (e.g., comedy, horror) and visual aesthetics (e.g., anime, 3D renderings). ViStoryBench is carefully curated to balance narrative structures and visual elements, featuring stories with single and multiple protagonists to test models' ability to maintain character consistency. Additionally, it includes complex plots and intricate world-building to challenge models in generating accurate visuals. To ensure comprehensive comparisons, our benchmark incorporates a wide range of evaluation metrics assessing critical aspects. This structured and multifaceted framework enables researchers to thoroughly identify both the strengths and weaknesses of different models, fostering targeted improvements.
Authors: Xinqi Xiong, Prakrut Patel, Qingyuan Fan, Amisha Wadhwa, Sarathy Selvam, Xiao Guo, Luchao Qi, Xiaoming Liu, Roni Sengupta
Abstract: The rapid advancement of talking-head deepfake generation fueled by advanced generative models has elevated the realism of synthetic videos to a level that poses substantial risks in domains such as media, politics, and finance. However, current benchmarks for deepfake talking-head detection fail to reflect this progress, relying on outdated generators and offering limited insight into model robustness and generalization. We introduce TalkingHeadBench, a comprehensive multi-model multi-generator benchmark and curated dataset designed to evaluate the performance of state-of-the-art detectors on the most advanced generators. Our dataset includes deepfakes synthesized by leading academic and commercial models and features carefully constructed protocols to assess generalization under distribution shifts in identity and generator characteristics. We benchmark a diverse set of existing detection methods, including CNNs, vision transformers, and temporal models, and analyze their robustness and generalization capabilities. In addition, we provide error analysis using Grad-CAM visualizations to expose common failure modes and detector biases. TalkingHeadBench is hosted on https://huggingface.co/datasets/luchaoqi/TalkingHeadBench with open access to all data splits and protocols. Our benchmark aims to accelerate research towards more robust and generalizable detection models in the face of rapidly evolving generative techniques.
URLs: https://huggingface.co/datasets/luchaoqi/TalkingHeadBench
Authors: Ujjwal Upadhyay, Mukul Ranjan, Zhiqiang Shen, Mohamed Elhoseiny
Abstract: Recent advances in vision-language models (VLMs) have made impressive strides in understanding spatio-temporal relationships in videos. However, when spatial information is obscured, these models struggle to capture purely temporal patterns. We introduce $\textbf{SpookyBench}$, a benchmark where information is encoded solely in temporal sequences of noise-like frames, mirroring natural phenomena from biological signaling to covert communication. Interestingly, while humans can recognize shapes, text, and patterns in these sequences with over 98% accuracy, state-of-the-art VLMs achieve 0% accuracy. This performance gap highlights a critical limitation: an over-reliance on frame-level spatial features and an inability to extract meaning from temporal cues. Furthermore, when trained in data sets with low spatial signal-to-noise ratios (SNR), temporal understanding of models degrades more rapidly than human perception, especially in tasks requiring fine-grained temporal reasoning. Overcoming this limitation will require novel architectures or training paradigms that decouple spatial dependencies from temporal processing. Our systematic analysis shows that this issue persists across model scales and architectures. We release SpookyBench to catalyze research in temporal pattern recognition and bridge the gap between human and machine video understanding. Dataset and code has been made available on our project website: https://timeblindness.github.io/.
Authors: Ce Zhang, Yan-Bo Lin, Ziyang Wang, Mohit Bansal, Gedas Bertasius
Abstract: Recent advances in test-time optimization have led to remarkable reasoning capabilities in Large Language Models (LLMs), enabling them to solve highly complex problems in math and coding. However, the reasoning capabilities of multimodal LLMs (MLLMs) still significantly lag, especially for complex video-language tasks. To address this issue, we present SiLVR, a Simple Language-based Video Reasoning framework that decomposes complex video understanding into two stages. In the first stage, SiLVR transforms raw video into language-based representations using multisensory inputs, such as short clip captions and audio/speech subtitles. In the second stage, language descriptions are fed into a powerful reasoning LLM to solve complex video-language understanding tasks. To handle long-context multisensory inputs, we use an adaptive token reduction scheme, which dynamically determines the temporal granularity with which to sample the tokens. Our simple, modular, and training-free video reasoning framework achieves the best-reported results on Video-MME (long), Video-MMMU (comprehension), Video-MMLU, CGBench, and EgoLife. Furthermore, our empirical study focused on video reasoning capabilities shows that, despite not being explicitly trained on video, strong reasoning LLMs can effectively aggregate multisensory input information from video, speech, and audio for complex temporal, causal, long-context, and knowledge acquisition reasoning tasks in video. Code is available at https://github.com/CeeZh/SILVR.
Authors: Zehan Wang, Jiayang Xu, Ziang Zhang, Tianyu Pan, Chao Du, Hengshuang Zhao, Zhou Zhao
Abstract: Humans can intuitively compose and arrange scenes in the 3D space for photography. However, can advanced AI image generators plan scenes with similar 3D spatial awareness when creating images from text or image prompts? We present GenSpace, a novel benchmark and evaluation pipeline to comprehensively assess the spatial awareness of current image generation models. Furthermore, standard evaluations using general Vision-Language Models (VLMs) frequently fail to capture the detailed spatial errors. To handle this challenge, we propose a specialized evaluation pipeline and metric, which reconstructs 3D scene geometry using multiple visual foundation models and provides a more accurate and human-aligned metric of spatial faithfulness. Our findings show that while AI models create visually appealing images and can follow general instructions, they struggle with specific 3D details like object placement, relationships, and measurements. We summarize three core limitations in the spatial perception of current state-of-the-art image generation models: 1) Object Perspective Understanding, 2) Egocentric-Allocentric Transformation and 3) Metric Measurement Adherence, highlighting possible directions for improving spatial intelligence in image generation.
Authors: Yiqing Liang, Jielin Qiu, Wenhao Ding, Zuxin Liu, James Tompkin, Mengdi Xu, Mengzhou Xia, Zhengzhong Tu, Laixi Shi, Jiacheng Zhu
Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has recently emerged as a powerful paradigm for post-training large language models (LLMs), achieving state-of-the-art performance on tasks with structured, verifiable answers. Applying RLVR to Multimodal LLMs (MLLMs) presents significant opportunities but is complicated by the broader, heterogeneous nature of vision-language tasks that demand nuanced visual, logical, and spatial capabilities. As such, training MLLMs using RLVR on multiple datasets could be beneficial but creates challenges with conflicting objectives from interaction among diverse datasets, highlighting the need for optimal dataset mixture strategies to improve generalization and reasoning. We introduce a systematic post-training framework for Multimodal LLM RLVR, featuring a rigorous data mixture problem formulation and benchmark implementation. Specifically, (1) We developed a multimodal RLVR framework for multi-dataset post-training by curating a dataset that contains different verifiable vision-language problems and enabling multi-domain online RL learning with different verifiable rewards; (2) We proposed a data mixture strategy that learns to predict the RL fine-tuning outcome from the data mixture distribution, and consequently optimizes the best mixture. Comprehensive experiments showcase that multi-domain RLVR training, when combined with mixture prediction strategies, can significantly boost MLLM general reasoning capacities. Our best mixture improves the post-trained model's accuracy on out-of-distribution benchmarks by an average of 5.24% compared to the same model post-trained with uniform data mixture, and by a total of 20.74% compared to the pre-finetuning baseline.
Authors: Zilin Xiao, Jaywon Koo, Siru Ouyang, Jefferson Hernandez, Yu Meng, Vicente Ordonez
Abstract: Recent advancements in reinforcement learning with verifiable rewards have pushed the boundaries of the visual reasoning capabilities in large vision-language models (LVLMs). However, training LVLMs with reinforcement fine-tuning (RFT) is computationally expensive, posing a significant challenge to scaling model size. In this work, we propose ProxyThinker, an inference-time technique that enables large models to inherit the visual reasoning capabilities from small, slow-thinking visual reasoners without any training. By subtracting the output distributions of base models from those of RFT reasoners, ProxyThinker modifies the decoding dynamics and successfully elicits the slow-thinking reasoning demonstrated by the emerged sophisticated behaviors such as self-verification and self-correction. ProxyThinker consistently boosts performance on challenging visual benchmarks on spatial, mathematical, and multi-disciplinary reasoning, enabling untuned base models to compete with the performance of their full-scale RFT counterparts. Furthermore, our implementation efficiently coordinates multiple language models with parallelism techniques and achieves up to 38 $\times$ faster inference compared to previous decoding-time methods, paving the way for the practical deployment of ProxyThinker. Code is available at https://github.com/MrZilinXiao/ProxyThinker.
Authors: Bojia Zi, Weixuan Peng, Xianbiao Qi, Jianan Wang, Shihao Zhao, Rong Xiao, Kam-Fai Wong
Abstract: Recent advances in video diffusion models have driven rapid progress in video editing techniques. However, video object removal, a critical subtask of video editing, remains challenging due to issues such as hallucinated objects and visual artifacts. Furthermore, existing methods often rely on computationally expensive sampling procedures and classifier-free guidance (CFG), resulting in slow inference. To address these limitations, we propose MiniMax-Remover, a novel two-stage video object removal approach. Motivated by the observation that text condition is not best suited for this task, we simplify the pretrained video generation model by removing textual input and cross-attention layers, resulting in a more lightweight and efficient model architecture in the first stage. In the second stage, we distilled our remover on successful videos produced by the stage-1 model and curated by human annotators, using a minimax optimization strategy to further improve editing quality and inference speed. Specifically, the inner maximization identifies adversarial input noise ("bad noise") that makes failure removals, while the outer minimization step trains the model to generate high-quality removal results even under such challenging conditions. As a result, our method achieves a state-of-the-art video object removal results with as few as 6 sampling steps and doesn't rely on CFG, significantly improving inference efficiency. Extensive experiments demonstrate the effectiveness and superiority of MiniMax-Remover compared to existing methods. Codes and Videos are available at: https://minimax-remover.github.io.
Authors: Yu Zhang, Yunqi Li, Yifan Yang, Rui Wang, Yuqing Yang, Dai Qi, Jianmin Bao, Dongdong Chen, Chong Luo, Lili Qiu
Abstract: Although chain-of-thought reasoning and reinforcement learning (RL) have driven breakthroughs in NLP, their integration into generative vision models remains underexplored. We introduce ReasonGen-R1, a two-stage framework that first imbues an autoregressive image generator with explicit text-based "thinking" skills via supervised fine-tuning on a newly generated reasoning dataset of written rationales, and then refines its outputs using Group Relative Policy Optimization. To enable the model to reason through text before generating images, We automatically generate and release a corpus of model crafted rationales paired with visual prompts, enabling controlled planning of object layouts, styles, and scene compositions. Our GRPO algorithm uses reward signals from a pretrained vision language model to assess overall visual quality, optimizing the policy in each update. Evaluations on GenEval, DPG, and the T2I benchmark demonstrate that ReasonGen-R1 consistently outperforms strong baselines and prior state-of-the-art models. More: aka.ms/reasongen.
Authors: Tajamul Ashraf, Amal Saqib, Hanan Ghani, Muhra AlMahri, Yuhao Li, Noor Ahsan, Umair Nawaz, Jean Lahoud, Hisham Cholakkal, Mubarak Shah, Philip Torr, Fahad Shahbaz Khan, Rao Muhammad Anwer, Salman Khan
Abstract: Deep reasoning is fundamental for solving complex tasks, especially in vision-centric scenarios that demand sequential, multimodal understanding. However, existing benchmarks typically evaluate agents with fully synthetic, single-turn queries, limited visual modalities, and lack a framework to assess reasoning quality over multiple steps as required in real-world settings. To address this, we introduce Agent-X, a large-scale benchmark for evaluating vision-centric agents multi-step and deep reasoning capabilities in real-world, multimodal settings. Agent- X features 828 agentic tasks with authentic visual contexts, including images, multi-image comparisons, videos, and instructional text. These tasks span six major agentic environments: general visual reasoning, web browsing, security and surveillance, autonomous driving, sports, and math reasoning. Our benchmark requires agents to integrate tool use with explicit, stepwise decision-making in these diverse settings. In addition, we propose a fine-grained, step-level evaluation framework that assesses the correctness and logical coherence of each reasoning step and the effectiveness of tool usage throughout the task. Our results reveal that even the best-performing models, including GPT, Gemini, and Qwen families, struggle to solve multi-step vision tasks, achieving less than 50% full-chain success. These findings highlight key bottlenecks in current LMM reasoning and tool-use capabilities and identify future research directions in vision-centric agentic reasoning models. Our data and code are publicly available at https://github.com/mbzuai-oryx/Agent-X
Authors: Yangyi Huang, Ye Yuan, Xueting Li, Jan Kautz, Umar Iqbal
Abstract: Existing methods for image-to-3D avatar generation struggle to produce highly detailed, animation-ready avatars suitable for real-world applications. We introduce AdaHuman, a novel framework that generates high-fidelity animatable 3D avatars from a single in-the-wild image. AdaHuman incorporates two key innovations: (1) A pose-conditioned 3D joint diffusion model that synthesizes consistent multi-view images in arbitrary poses alongside corresponding 3D Gaussian Splats (3DGS) reconstruction at each diffusion step; (2) A compositional 3DGS refinement module that enhances the details of local body parts through image-to-image refinement and seamlessly integrates them using a novel crop-aware camera ray map, producing a cohesive detailed 3D avatar. These components allow AdaHuman to generate highly realistic standardized A-pose avatars with minimal self-occlusion, enabling rigging and animation with any input motion. Extensive evaluation on public benchmarks and in-the-wild images demonstrates that AdaHuman significantly outperforms state-of-the-art methods in both avatar reconstruction and reposing. Code and models will be publicly available for research purposes.
Authors: Jian Lan, Yifei Fu, Udo Schlegel, Gengyuan Zhang, Tanveer Hannan, Haokun Chen, Thomas Seidl
Abstract: Social bias is a critical issue in large vision-language models (VLMs), where fairness- and ethics-related problems harm certain groups of people in society. It is unknown to what extent VLMs yield social bias in generative responses. In this study, we focus on evaluating and mitigating social bias on both the model's response and probability distribution. To do so, we first evaluate four state-of-the-art VLMs on PAIRS and SocialCounterfactuals datasets with the multiple-choice selection task. Surprisingly, we find that models suffer from generating gender-biased or race-biased responses. We also observe that models are prone to stating their responses are fair, but indeed having mis-calibrated confidence levels towards particular social groups. While investigating why VLMs are unfair in this study, we observe that VLMs' hidden layers exhibit substantial fluctuations in fairness levels. Meanwhile, residuals in each layer show mixed effects on fairness, with some contributing positively while some lead to increased bias. Based on these findings, we propose a post-hoc method for the inference stage to mitigate social bias, which is training-free and model-agnostic. We achieve this by ablating bias-associated residuals while amplifying fairness-associated residuals on model hidden layers during inference. We demonstrate that our post-hoc method outperforms the competing training strategies, helping VLMs have fairer responses and more reliable confidence levels.
Authors: Anam Hashmi, Julia Dietlmeier, Kathleen M. Curran, Noel E. O'Connor
Abstract: Attention is a fundamental component of the human visual recognition system. The inclusion of attention in a convolutional neural network amplifies relevant visual features and suppresses the less important ones. Integrating attention mechanisms into convolutional neural networks enhances model performance and interpretability. Spatial and channel attention mechanisms have shown significant advantages across many downstream tasks in medical imaging. While existing attention modules have proven to be effective, their design often lacks a robust theoretical underpinning. In this study, we address this gap by proposing a non-linear attention architecture for cardiac MRI reconstruction and hypothesize that insights from ecological principles can guide the development of effective and efficient attention mechanisms. Specifically, we investigate a non-linear ecological difference equation that describes single-species population growth to devise a parameter-free attention module surpassing current state-of-the-art parameter-free methods.
Authors: Tianyuan Zhang, Sai Bi, Yicong Hong, Kai Zhang, Fujun Luan, Songlin Yang, Kalyan Sunkavalli, William T. Freeman, Hao Tan
Abstract: Test-Time Training (TTT) models context dependencies by adapting part of the model's weights (referred to as fast weights) during inference. This fast weight, akin to recurrent states in RNNs, stores temporary memories of past tokens in the current sequence. Existing TTT methods struggled to show effectiveness in handling long-context data, due to their inefficiency on modern GPUs. The TTT layers in many of these approaches operate with extremely low FLOPs utilization (often <5%) because they deliberately apply small online minibatch sizes (e.g., updating fast weights every 16 or 64 tokens). Moreover, a small minibatch implies fine-grained block-wise causal dependencies in the data, unsuitable for data beyond 1D ordered sequences, like sets or N-dimensional grids such as images or videos. In contrast, we pursue the opposite direction by using an extremely large chunk update, ranging from 2K to 1M tokens across tasks of varying modalities, which we refer to as Large Chunk Test-Time Training (LaCT). It improves hardware utilization by orders of magnitude, and more importantly, facilitates scaling of nonlinear state size (up to 40% of model parameters), hence substantially improving state capacity, all without requiring cumbersome and error-prone kernel implementations. It also allows easy integration of sophisticated optimizers, e.g. Muon for online updates. We validate our approach across diverse modalities and tasks, including novel view synthesis with image set, language models, and auto-regressive video diffusion. Our approach can scale up to 14B-parameter AR video diffusion model on sequences up to 56K tokens. In our longest sequence experiment, we perform novel view synthesis with 1 million context length. We hope this work will inspire and accelerate new research in the field of long-context modeling and test-time training. Website: https://tianyuanzhang.com/projects/ttt-done-right
Authors: Charles Bricout, Samira Ebrahimi Kahou, Sylvain Bouix
Abstract: Motion-related artifacts are inevitable in Magnetic Resonance Imaging (MRI) and can bias automated neuroanatomical metrics such as cortical thickness. Manual review cannot objectively quantify motion in anatomical scans, and existing automated approaches often require specialized hardware or rely on unbalanced noisy training data. Here, we train a 3D convolutional neural network to estimate motion severity using only synthetically corrupted volumes. We validate our method with one held-out site from our training cohort and with 14 fully independent datasets, including one with manual ratings, achieving a representative $R^2 = 0.65$ versus manual labels and significant thickness-motion correlations in 12/15 datasets. Furthermore, our predicted motion correlates with subject age in line with prior studies. Our approach generalizes across scanner brands and protocols, enabling objective, scalable motion assessment in structural MRI studies without prospective motion correction.
Authors: An Vo, Khai-Nguyen Nguyen, Mohammad Reza Taesiri, Vy Tuong Dang, Anh Totti Nguyen, Daeyoung Kim
Abstract: Large language models (LLMs) memorize a vast amount of prior knowledge from the Internet that help them on downstream tasks but also may notoriously sway their outputs towards wrong or biased answers. In this work, we test how the knowledge about popular subjects hurt the accuracy of vision language models (VLMs) on standard, objective visual tasks of counting and identification. We find that state-of-the-art VLMs are strongly biased (e.g, unable to recognize a fourth stripe has been added to a 3-stripe Adidas logo) scoring an average of 17.05% accuracy in counting (e.g., counting stripes in an Adidas-like logo) across 7 diverse domains from animals, logos, chess, board games, optical illusions, to patterned grids. Insert text (e.g., "Adidas") describing the subject name into the counterfactual image further decreases VLM accuracy. The biases in VLMs are so strong that instructing them to double-check their results or rely exclusively on image details to answer improves counting accuracy by only +2 points, on average. Our work presents an interesting failure mode in VLMs and an automated framework for testing VLM biases. Code and data are available at: vlmsarebiased.github.io.
Authors: Ziming Zhao, ChengAo Shen, Hanghang Tong, Dongjin Song, Zhigang Deng, Qingsong Wen, Jingchao Ni
Abstract: Transformer-based models have gained increasing attention in time series research, driving interest in Large Language Models (LLMs) and foundation models for time series analysis. As the field moves toward multi-modality, Large Vision Models (LVMs) are emerging as a promising direction. In the past, the effectiveness of Transformer and LLMs in time series has been debated. When it comes to LVMs, a similar question arises: are LVMs truely useful for time series analysis? To address it, we design and conduct the first principled study involving 4 LVMs, 8 imaging methods, 18 datasets and 26 baselines across both high-level (classification) and low-level (forecasting) tasks, with extensive ablation analysis. Our findings indicate LVMs are indeed useful for time series classification but face challenges in forecasting. Although effective, the contemporary best LVM forecasters are limited to specific types of LVMs and imaging methods, exhibit a bias toward forecasting periods, and have limited ability to utilize long look-back windows. We hope our findings could serve as a cornerstone for future research on LVM- and multimodal-based solutions to different time series tasks.
Authors: L\'eo And\'eol, Luca Mossina, Adrien Mazoyer, S\'ebastien Gerchinovitz
Abstract: Recent advances in object detectors have led to their adoption for industrial uses. However, their deployment in critical applications is hindered by the inherent lack of reliability of neural networks and the complex structure of object detection models. To address these challenges, we turn to Conformal Prediction, a post-hoc procedure which offers statistical guarantees that are valid for any dataset size, without requiring prior knowledge on the model or data distribution. Our contribution is manifold: first, we formally define the problem of Conformal Object Detection (COD) and introduce a novel method, Sequential Conformal Risk Control (SeqCRC), that extends the statistical guarantees of Conformal Risk Control (CRC) to two sequential tasks with two parameters, as required in the COD setting. Then, we propose loss functions and prediction sets suited to applying CRC to different applications and certification requirements. Finally, we present a conformal toolkit, enabling replication and further exploration of our methods. Using this toolkit, we perform extensive experiments, yielding a benchmark that validates the investigated methods and emphasizes trade-offs and other practical consequences.
Authors: Zixun Huang, Cho-Ying Wu, Yuliang Guo, Xinyu Huang, Liu Ren
Abstract: 3D Gaussian Splatting (3DGS) marks a significant milestone in balancing the quality and efficiency of differentiable rendering. However, its high efficiency stems from an approximation of projecting 3D Gaussians onto the image plane as 2D Gaussians, which inherently limits rendering quality--particularly under large Field-of-View (FoV) camera inputs. While several recent works have extended 3DGS to mitigate these approximation errors, none have successfully achieved both exactness and high efficiency simultaneously. In this work, we introduce 3DGEER, an Exact and Efficient Volumetric Gaussian Rendering method. Starting from first principles, we derive a closed-form expression for the density integral along a ray traversing a 3D Gaussian distribution. This formulation enables precise forward rendering with arbitrary camera models and supports gradient-based optimization of 3D Gaussian parameters. To ensure both exactness and real-time performance, we propose an efficient method for computing a tight Particle Bounding Frustum (PBF) for each 3D Gaussian, enabling accurate and efficient ray-Gaussian association. We also introduce a novel Bipolar Equiangular Projection (BEAP) representation to accelerate ray association under generic camera models. BEAP further provides a more uniform ray sampling strategy to apply supervision, which empirically improves reconstruction quality. Experiments on multiple pinhole and fisheye datasets show that our method consistently outperforms prior methods, establishing a new state-of-the-art in real-time neural rendering.
Authors: Kamyar Barakati, Yu Liu, Hiroshi Funakubo, Sergei V. Kalinin
Abstract: Domain-wall dynamics in ferroelectric materials are strongly position-dependent since each polar interface is locked into a unique local microstructure. This necessitates spatially resolved studies of the wall-pinning using scanning-probe microscopy techniques. The pinning centers and preexisting domain walls are usually sparse within image plane, precluding the use of dense hyperspectral imaging modes and requiring time-consuming human experimentation. Here, a large area epitaxial PbTiO$_3$ film on cubic KTaO$_3$ were investigated to quantify the electric field driven dynamics of the polar-strain domain structures using ML-controlled automated Piezoresponse Force Microscopy. Analysis of 1500 switching events reveals that domain wall displacement depends not only on field parameters but also on the local ferroelectric-ferroelastic configuration. For example, twin boundaries in polydomains regions like a$_1^-$/$c^+$ $\parallel$ a$_2^-$/$c^-$ stay pinned up to a certain level of bias magnitude and change only marginally as the bias increases from 20V to 30V, whereas single variant boundaries like a$_2^+$/$c^+$ $\parallel$ a$_2^-$/$c^-$ stack are already activated at 20V. These statistics on the possible ferroelectric and ferroelastic wall orientations, together with the automated, high-throughput AFM workflow, can be distilled into a predictive map that links domain configurations to pulse parameters. This microstructure-specific rule set forms the foundation for designing ferroelectric memories.
Authors: Chen Huang, Skyler Seto, Hadi Pouransari, Mehrdad Farajtabar, Raviteja Vemulapalli, Fartash Faghri, Oncel Tuzel, Barry-John Theobald, Josh Susskind
Abstract: Vision foundation models pre-trained on massive data encode rich representations of real-world concepts, which can be adapted to downstream tasks by fine-tuning. However, fine-tuning foundation models on one task often leads to the issue of concept forgetting on other tasks. Recent methods of robust fine-tuning aim to mitigate forgetting of prior knowledge without affecting the fine-tuning performance. Knowledge is often preserved by matching the original and fine-tuned model weights or feature pairs. However, such point-wise matching can be too strong, without explicit awareness of the feature neighborhood structures that encode rich knowledge as well. We propose a novel regularization method Proxy-FDA that explicitly preserves the structural knowledge in feature space. Proxy-FDA performs Feature Distribution Alignment (using nearest neighbor graphs) between the pre-trained and fine-tuned feature spaces, and the alignment is further improved by informative proxies that are generated dynamically to increase data diversity. Experiments show that Proxy-FDA significantly reduces concept forgetting during fine-tuning, and we find a strong correlation between forgetting and a distributional distance metric (in comparison to L2 distance). We further demonstrate Proxy-FDA's benefits in various fine-tuning settings (end-to-end, few-shot and continual tuning) and across different tasks like image classification, captioning and VQA.
Authors: Ricardo Baptista, Andrew M. Stuart, Son Tran
Abstract: Multimodal contrastive learning is a methodology for linking different data modalities; the canonical example is linking image and text data. The methodology is typically framed as the identification of a set of encoders, one for each modality, that align representations within a common latent space. In this work, we focus on the bimodal setting and interpret contrastive learning as the optimization of (parameterized) encoders that define conditional probability distributions, for each modality conditioned on the other, consistent with the available data. This provides a framework for multimodal algorithms such as crossmodal retrieval, which identifies the mode of one of these conditional distributions, and crossmodal classification, which is similar to retrieval but includes a fine-tuning step to make it task specific. The framework we adopt also gives rise to crossmodal generative models. This probabilistic perspective suggests two natural generalizations of contrastive learning: the introduction of novel probabilistic loss functions, and the use of alternative metrics for measuring alignment in the common latent space. We study these generalizations of the classical approach in the multivariate Gaussian setting. In this context we view the latent space identification as a low-rank matrix approximation problem. This allows us to characterize the capabilities of loss functions and alignment metrics to approximate natural statistics, such as conditional means and covariances; doing so yields novel variants on contrastive learning algorithms for specific mode-seeking and for generative tasks. The framework we introduce is also studied through numerical experiments on multivariate Gaussians, the labeled MNIST dataset, and on a data assimilation application arising in oceanography.
Authors: Ya\c{s}ar Utku Al\c{c}alar, Mehmet Ak\c{c}akaya
Abstract: Physics-driven deep learning (PD-DL) models have proven to be a powerful approach for improved reconstruction of rapid MRI scans. In order to train these models in scenarios where fully-sampled reference data is unavailable, self-supervised learning has gained prominence. However, its application at high acceleration rates frequently introduces artifacts, compromising image fidelity. To mitigate this shortcoming, we propose a novel way to train PD-DL networks via carefully-designed perturbations. In particular, we enhance the k-space masking idea of conventional self-supervised learning with a novel consistency term that assesses the model's ability to accurately predict the added perturbations in a sparse domain, leading to more reliable and artifact-free reconstructions. The results obtained from the fastMRI knee and brain datasets show that the proposed training strategy effectively reduces aliasing artifacts and mitigates noise amplification at high acceleration rates, outperforming state-of-the-art self-supervised methods both visually and quantitatively.
Authors: Junyu Chen, Shuwen Wei, Joel Honkamaa, Pekka Marttinen, Hang Zhang, Min Liu, Yichao Zhou, Zuopeng Tan, Zhuoyuan Wang, Yi Wang, Hongchao Zhou, Shunbo Hu, Yi Zhang, Qian Tao, Lukas F\"orner, Thomas Wendler, Bailiang Jian, Benedikt Wiestler, Tim Hable, Jin Kim, Dan Ruan, Frederic Madesta, Thilo Sentker, Wiebke Heyer, Lianrui Zuo, Yuwei Dai, Jing Wu, Jerry L. Prince, Harrison Bai, Yong Du, Yihao Liu, Alessa Hering, Reuben Dorent, Lasse Hansen, Mattias P. Heinrich, Aaron Carass
Abstract: Medical image challenges have played a transformative role in advancing the field, catalyzing algorithmic innovation and establishing new performance standards across diverse clinical applications. Image registration, a foundational task in neuroimaging pipelines, has similarly benefited from the Learn2Reg initiative. Building on this foundation, we introduce the Large-scale Unsupervised Brain MRI Image Registration (LUMIR) challenge, a next-generation benchmark designed to assess and advance unsupervised brain MRI registration. Distinct from prior challenges that leveraged anatomical label maps for supervision, LUMIR removes this dependency by providing over 4,000 preprocessed T1-weighted brain MRIs for training without any label maps, encouraging biologically plausible deformation modeling through self-supervision. In addition to evaluating performance on 590 held-out test subjects, LUMIR introduces a rigorous suite of zero-shot generalization tasks, spanning out-of-domain imaging modalities (e.g., FLAIR, T2-weighted, T2*-weighted), disease populations (e.g., Alzheimer's disease), acquisition protocols (e.g., 9.4T MRI), and species (e.g., macaque brains). A total of 1,158 subjects and over 4,000 image pairs were included for evaluation. Performance was assessed using both segmentation-based metrics (Dice coefficient, 95th percentile Hausdorff distance) and landmark-based registration accuracy (target registration error). Across both in-domain and zero-shot tasks, deep learning-based methods consistently achieved state-of-the-art accuracy while producing anatomically plausible deformation fields. The top-performing deep learning-based models demonstrated diffeomorphic properties and inverse consistency, outperforming several leading optimization-based methods, and showing strong robustness to most domain shifts, the exception being a drop in performance on out-of-domain contrasts.
Authors: Shilin Xu, Yanwei Li, Rui Yang, Tao Zhang, Yueyi Sun, Wei Chow, Linfeng Li, Hang Song, Qi Xu, Yunhai Tong, Xiangtai Li, Hao Fei
Abstract: Recent works on large language models (LLMs) have successfully demonstrated the emergence of reasoning capabilities via reinforcement learning (RL). Although recent efforts leverage group relative policy optimization (GRPO) for MLLMs post-training, they constantly explore one specific aspect, such as grounding tasks, math problems, or chart analysis. There are no works that can leverage multi-source MLLM tasks for stable reinforcement learning. In this work, we present a unified perspective to solve this problem. We present Mixed-R1, a unified yet straightforward framework that contains a mixed reward function design (Mixed-Reward) and a mixed post-training dataset (Mixed-45K). We first design a data engine to select high-quality examples to build the Mixed-45K post-training dataset. Then, we present a Mixed-Reward design, which contains various reward functions for various MLLM tasks. In particular, it has four different reward functions: matching reward for binary answer or multiple-choice problems, chart reward for chart-aware datasets, IoU reward for grounding problems, and open-ended reward for long-form text responses such as caption datasets. To handle the various long-form text content, we propose a new open-ended reward named Bidirectional Max-Average Similarity (BMAS) by leveraging tokenizer embedding matching between the generated response and the ground truth. Extensive experiments show the effectiveness of our proposed method on various MLLMs, including Qwen2.5-VL and Intern-VL on various sizes. Our dataset and model are available at https://github.com/xushilin1/mixed-r1.
Authors: Yipan Wei, Yuchen Zou, Yapeng Li, Bo Du
Abstract: Federated Multi-Task Learning (FMTL) enables multiple clients performing heterogeneous tasks without exchanging their local data, offering broad potential for privacy preserving multi-task collaboration. However, most existing methods focus on building personalized models for each client and unable to support the aggregation of multiple heterogeneous tasks into a unified model. As a result, in real-world scenarios where task objectives, label spaces, and optimization paths vary significantly, conventional FMTL methods struggle to achieve effective joint training. To address this challenge, we propose FedDEA (Federated Decoupled Aggregation), an update-structure-aware aggregation method specifically designed for multi-task model integration. Our method dynamically identifies task-relevant dimensions based on the response strength of local updates and enhances their optimization effectiveness through rescaling. This mechanism effectively suppresses cross-task interference and enables task-level decoupled aggregation within a unified global model. FedDEA does not rely on task labels or architectural modifications, making it broadly applicable and deployment-friendly. Experimental results demonstrate that it can be easily integrated into various mainstream federated optimization algorithms and consistently delivers significant overall performance improvements on widely used NYUD-V2 and PASCAL-Context. These results validate the robustness and generalization capabilities of FedDEA under highly heterogeneous task settings.
Authors: Lan-Cuong Nguyen, Quan Nguyen-Tri, Bang Tran Khanh, Dung D. Le, Long Tran-Thanh, Khoat Than
Abstract: Few-shot image classification remains challenging due to the scarcity of labeled training examples. Augmenting them with synthetic data has emerged as a promising way to alleviate this issue, but models trained on synthetic samples often face performance degradation due to the inherent gap between real and synthetic distributions. To address this limitation, we develop a theoretical framework that quantifies the impact of such distribution discrepancies on supervised learning, specifically in the context of image classification. More importantly, our framework suggests practical ways to generate good synthetic samples and to train a predictor with high generalization ability. Building upon this framework, we propose a novel theoretical-based algorithm that integrates prototype learning to optimize both data partitioning and model training, effectively bridging the gap between real few-shot data and synthetic data. Extensive experiments results show that our approach demonstrates superior performance compared to state-of-the-art methods, outperforming them across multiple datasets.
Authors: Mingxu Zhang, Xiaoqi Li, Jiahui Xu, Kaichen Zhou, Hojin Bae, Yan Shen, Chuyan Xiong, Jiaming Liu, Hao Dong
Abstract: Recent advancements in 3D robotic manipulation have improved grasping of everyday objects, but transparent and specular materials remain challenging due to depth sensing limitations. While several 3D reconstruction and depth completion approaches address these challenges, they suffer from setup complexity or limited observation information utilization. To address this, leveraging the power of single view 3D object reconstruction approaches, we propose a training free framework SR3D that enables robotic grasping of transparent and specular objects from a single view observation. Specifically, given single view RGB and depth images, SR3D first uses the external visual models to generate 3D reconstructed object mesh based on RGB image. Then, the key idea is to determine the 3D object's pose and scale to accurately localize the reconstructed object back into its original depth corrupted 3D scene. Therefore, we propose view matching and keypoint matching mechanisms,which leverage both the 2D and 3D's inherent semantic and geometric information in the observation to determine the object's 3D state within the scene, thereby reconstructing an accurate 3D depth map for effective grasp detection. Experiments in both simulation and real world show the reconstruction effectiveness of SR3D.
Authors: Peng Qi, Wenxi Qu, Tianliang Yao, Haonan Ma, Dylan Wintle, Yinyi Lai, Giorgos Papanastasiou, Chengjia Wang
Abstract: Percutaneous Coronary Intervention (PCI) is a minimally invasive procedure that improves coronary blood flow and treats coronary artery disease. Although PCI typically requires 2D X-ray angiography (XRA) to guide catheter placement at real-time, computed tomography angiography (CTA) may substantially improve PCI by providing precise information of 3D vascular anatomy and status. To leverage real-time XRA and detailed 3D CTA anatomy for PCI, accurate multimodal image registration of XRA and CTA is required, to guide the procedure and avoid complications. This is a challenging process as it requires registration of images from different geometrical modalities (2D -> 3D and vice versa), with variations in contrast and noise levels. In this paper, we propose a novel multimodal coronary artery image registration method based on a swarm optimization algorithm, which effectively addresses challenges such as large deformations, low contrast, and noise across these imaging modalities. Our algorithm consists of two main modules: 1) preprocessing of XRA and CTA images separately, and 2) a registration module based on feature extraction using the Steger and Superpixel Particle Swarm Optimization algorithms. Our technique was evaluated on a pilot dataset of 28 pairs of XRA and CTA images from 10 patients who underwent PCI. The algorithm was compared with four state-of-the-art (SOTA) methods in terms of registration accuracy, robustness, and efficiency. Our method outperformed the selected SOTA baselines in all aspects. Experimental results demonstrate the significant effectiveness of our algorithm, surpassing the previous benchmarks and proposes a novel clinical approach that can potentially have merit for improving patient outcomes in coronary artery disease.
Authors: Wenlong Jiao, Binglong Li, Wei Shang, Ping Wang, Dongwei Ren
Abstract: Image deblurring plays a crucial role in enhancing visual clarity across various applications. Although most deep learning approaches primarily focus on sRGB images, which inherently lose critical information during the image signal processing pipeline, RAW images, being unprocessed and linear, possess superior restoration potential but remain underexplored. Deblurring RAW images presents unique challenges, particularly in handling frequency-dependent blur while maintaining computational efficiency. To address these issues, we propose Frequency Enhanced Network (FrENet), a framework specifically designed for RAW-to-RAW deblurring that operates directly in the frequency domain. We introduce a novel Adaptive Frequency Positional Modulation module, which dynamically adjusts frequency components according to their spectral positions, thereby enabling precise control over the deblurring process. Additionally, frequency domain skip connections are adopted to further preserve high-frequency details. Experimental results demonstrate that FrENet surpasses state-of-the-art deblurring methods in RAW image deblurring, achieving significantly better restoration quality while maintaining high efficiency in terms of reduced MACs. Furthermore, FrENet's adaptability enables it to be extended to sRGB images, where it delivers comparable or superior performance compared to methods specifically designed for sRGB data. The code will be available at https://github.com/WenlongJiao/FrENet .
Authors: Abdul-mojeed Olabisi Ilyas, Adeleke Maradesa, Jamal Banzi, Jianpan Huang, Henry K. F. Mak, Kannie W. Y. Chan
Abstract: Medical imaging is critical for diagnostics, but clinical adoption of advanced AI-driven imaging faces challenges due to patient variability, image artifacts, and limited model generalization. While deep learning has transformed image analysis, 3D medical imaging still suffers from data scarcity and inconsistencies due to acquisition protocols, scanner differences, and patient motion. Traditional augmentation uses a single pipeline for all transformations, disregarding the unique traits of each augmentation and struggling with large data volumes. To address these challenges, we propose a Multi-encoder Augmentation-Aware Learning (MEAL) framework that leverages four distinct augmentation variants processed through dedicated encoders. Three fusion strategies such as concatenation (CC), fusion layer (FL), and adaptive controller block (BD) are integrated to build multi-encoder models that combine augmentation-specific features before decoding. MEAL-BD uniquely preserves augmentation-aware representations, enabling robust, protocol-invariant feature learning. As demonstrated in a Computed Tomography (CT)-to-T1-weighted Magnetic Resonance Imaging (MRI) translation study, MEAL-BD consistently achieved the best performance on both unseen- and predefined-test data. On both geometric transformations (like rotations and flips) and non-augmented inputs, MEAL-BD outperformed other competing methods, achieving higher mean peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) scores. These results establish MEAL as a reliable framework for preserving structural fidelity and generalizing across clinically relevant variability. By reframing augmentation as a source of diverse, generalizable features, MEAL supports robust, protocol-invariant learning, advancing clinically reliable medical imaging solutions.
Authors: Amit Peleg, Naman Deep Singh, Matthias Hein
Abstract: Vision-language models like CLIP have demonstrated remarkable zero-shot capabilities in classification and retrieval. However, these models often struggle with compositional reasoning - the ability to understand the relationships between concepts. A recent benchmark, SugarCrepe++, reveals that previous works on improving compositionality have mainly improved lexical sensitivity but neglected semantic understanding. In addition, downstream retrieval performance often deteriorates, although one would expect that improving compositionality should enhance retrieval. In this work, we introduce CLIC (Compositionally-aware Learning in CLIP), a fine-tuning method based on a novel training technique combining multiple images and their associated captions. CLIC improves compositionality across architectures as well as differently pre-trained CLIP models, both in terms of lexical and semantic understanding, and achieves consistent gains in retrieval performance. This even applies to the recent CLIPS, which achieves SOTA retrieval performance. Nevertheless, the short fine-tuning with CLIC leads to an improvement in retrieval and to the best compositional CLIP model on SugarCrepe++. All our models and code are available at https://clic-compositional-clip.github.io
Authors: Md Shahriar Rahim Siddiqui, Moshe Eliasof, Eldad Haber
Abstract: Flow matching casts sample generation as learning a continuous-time velocity field that transports noise to data. Existing flow matching networks typically predict each point's velocity independently, considering only its location and time along its flow trajectory, and ignoring neighboring points. However, this pointwise approach may overlook correlations between points along the generation trajectory that could enhance velocity predictions, thereby improving downstream generation quality. To address this, we propose Graph Flow Matching (GFM), a lightweight enhancement that decomposes the learned velocity into a reaction term -- any standard flow matching network -- and a diffusion term that aggregates neighbor information via a graph neural module. This reaction-diffusion formulation retains the scalability of deep flow models while enriching velocity predictions with local context, all at minimal additional computational cost. Operating in the latent space of a pretrained variational autoencoder, GFM consistently improves Fr\'echet Inception Distance (FID) and recall across five image generation benchmarks (LSUN Church, LSUN Bedroom, FFHQ, AFHQ-Cat, and CelebA-HQ at $256\times256$), demonstrating its effectiveness as a modular enhancement to existing flow matching architectures.
Authors: Janek Gr\"ohl, Leonid Kunyansky, Jenni Poimala, Thomas R. Else, Francesca Di Cecio, Sarah E. Bohndiek, Ben T. Cox, Andreas Hauptmann
Abstract: Quantitative comparison of the quality of photoacoustic image reconstruction algorithms remains a major challenge. No-reference image quality measures are often inadequate, but full-reference measures require access to an ideal reference image. While the ground truth is known in simulations, it is unknown in vivo, or in phantom studies, as the reference depends on both the phantom properties and the imaging system. We tackle this problem by using numerical digital twins of tissue-mimicking phantoms and the imaging system to perform a quantitative calibration to reduce the simulation gap. The contributions of this paper are two-fold: First, we use this digital-twin framework to compare multiple state-of-the-art reconstruction algorithms. Second, among these is a Fourier transform-based reconstruction algorithm for circular detection geometries, which we test on experimental data for the first time. Our results demonstrate the usefulness of digital phantom twins by enabling assessment of the accuracy of the numerical forward model and enabling comparison of image reconstruction schemes with full-reference image quality assessment. We show that the Fourier transform-based algorithm yields results comparable to those of iterative time reversal, but at a lower computational cost. All data and code are publicly available on Zenodo: https://doi.org/10.5281/zenodo.15388429.
Authors: Ivan Pereira-S\'anchez, Julia Navarro, Ana Bel\'en Petro, Joan Duran
Abstract: This paper addresses the problem of reconstructing a high-resolution hyperspectral image from a low-resolution multispectral observation. While spatial super-resolution and spectral super-resolution have been extensively studied, joint spatio-spectral super-resolution remains relatively explored. We propose an end-to-end model-driven framework that explicitly decomposes the joint spatio-spectral super-resolution problem into spatial super-resolution, spectral super-resolution and fusion tasks. Each sub-task is addressed by unfolding a variational-based approach, where the operators involved in the proximal gradient iterative scheme are replaced with tailored learnable modules. In particular, we design an upsampling operator for spatial super-resolution based on classical back-projection algorithms, adapted to handle arbitrary scaling factors. Spectral reconstruction is performed using learnable cluster-based upsampling and downsampling operators. For image fusion, we integrate low-frequency estimation and high-frequency injection modules to combine the spatial and spectral information from spatial super-resolution and spectral super-resolution outputs. Additionally, we introduce an efficient nonlocal post-processing step that leverages image self-similarity by combining a multi-head attention mechanism with residual connections. Extensive evaluations on several datasets and sampling factors demonstrate the effectiveness of our approach. The source code will be available at https://github.com/TAMI-UIB/JSSUNet
Authors: Wenyuan Li, Guang Li, Keisuke Maeda, Takahiro Ogawa, Miki Haseyama
Abstract: To address the computational and storage challenges posed by large-scale datasets in deep learning, dataset distillation has been proposed to synthesize a compact dataset that replaces the original while maintaining comparable model performance. Unlike optimization-based approaches that require costly bi-level optimization, distribution matching (DM) methods improve efficiency by aligning the distributions of synthetic and original data, thereby eliminating nested optimization. DM achieves high computational efficiency and has emerged as a promising solution. However, existing DM methods, constrained to Euclidean space, treat data as independent and identically distributed points, overlooking complex geometric and hierarchical relationships. To overcome this limitation, we propose a novel hyperbolic dataset distillation method, termed HDD. Hyperbolic space, characterized by negative curvature and exponential volume growth with distance, naturally models hierarchical and tree-like structures. HDD embeds features extracted by a shallow network into the Lorentz hyperbolic space, where the discrepancy between synthetic and original data is measured by the hyperbolic (geodesic) distance between their centroids. By optimizing this distance, the hierarchical structure is explicitly integrated into the distillation process, guiding synthetic samples to gravitate towards the root-centric regions of the original data distribution while preserving their underlying geometric characteristics. Furthermore, we find that pruning in hyperbolic space requires only 20% of the distilled core set to retain model performance, while significantly improving training stability. Notably, HDD is seamlessly compatible with most existing DM methods, and extensive experiments on different datasets validate its effectiveness.
Authors: Maria Rafaela Gkeka, Bowen Sun, Evgenia Smirni, Christos D. Antonopoulos, Spyros Lalis, Nikolaos Bellas
Abstract: Continuous advancements in deep learning have led to significant progress in feature detection, resulting in enhanced accuracy in tasks like Simultaneous Localization and Mapping (SLAM). Nevertheless, the vulnerability of deep neural networks to adversarial attacks remains a challenge for their reliable deployment in applications, such as navigation of autonomous agents. Even though CNN-based SLAM algorithms are a growing area of research there is a notable absence of a comprehensive presentation and examination of adversarial attacks targeting CNN-based feature detectors, as part of a SLAM system. Our work introduces black-box adversarial perturbations applied to the RGB images fed into the GCN-SLAM algorithm. Our findings on the TUM dataset [30] reveal that even attacks of moderate scale can lead to tracking failure in as many as 76% of the frames. Moreover, our experiments highlight the catastrophic impact of attacking depth instead of RGB input images on the SLAM system.
Authors: Shengyuan Liu, Wenting Chen, Boyun Zheng, Wentao Pan, Xiang Li, Yixuan Yuan
Abstract: Tumor data synthesis offers a promising solution to the shortage of annotated medical datasets. However, current approaches either limit tumor diversity by using predefined masks or employ computationally expensive two-stage processes with multiple denoising steps, causing computational inefficiency. Additionally, these methods typically rely on binary masks that fail to capture the gradual transitions characteristic of tumor boundaries. We present TumorGen, a novel Boundary-Aware Tumor-Mask Synthesis with Rectified Flow Matching for efficient 3D tumor synthesis with three key components: a Boundary-Aware Pseudo Mask Generation module that replaces strict binary masks with flexible bounding boxes; a Spatial-Constraint Vector Field Estimator that simultaneously synthesizes tumor latents and masks using rectified flow matching to ensure computational efficiency; and a VAE-guided mask refiner that enhances boundary realism. TumorGen significantly improves computational efficiency by requiring fewer sampling steps while maintaining pathological accuracy through coarse and fine-grained spatial constraints. Experimental results demonstrate TumorGen's superior performance over existing tumor synthesis methods in both efficiency and realism, offering a valuable contribution to AI-driven cancer diagnostics.
Authors: Dennis Jacob, Chong Xiang, Prateek Mittal
Abstract: Deep learning techniques have enabled vast improvements in computer vision technologies. Nevertheless, these models are vulnerable to adversarial patch attacks which catastrophically impair performance. The physically realizable nature of these attacks calls for certifiable defenses, which feature provable guarantees on robustness. While certifiable defenses have been successfully applied to single-label classification, limited work has been done for multi-label classification. In this work, we present PatchDEMUX, a certifiably robust framework for multi-label classifiers against adversarial patches. Our approach is a generalizable method which can extend any existing certifiable defense for single-label classification; this is done by considering the multi-label classification task as a series of isolated binary classification problems to provably guarantee robustness. Furthermore, in the scenario where an attacker is limited to a single patch we propose an additional certification procedure that can provide tighter robustness bounds. Using the current state-of-the-art (SOTA) single-label certifiable defense PatchCleanser as a backbone, we find that PatchDEMUX can achieve non-trivial robustness on the MS-COCO and PASCAL VOC datasets while maintaining high clean performance
Authors: Xinliu Zhong, Ruiying Liu, Emily S. Nichols, Xuzhe Zhang, Andrew F. Laine, Emma G. Duerden, Yun Wang
Abstract: Accurate placental segmentation is essential for quantitative analysis of the placenta. However, this task is particularly challenging in T2*-weighted placental imaging due to: (1) weak and inconsistent boundary contrast across individual echoes; (2) the absence of manual ground truth annotations for all echo times; and (3) motion artifacts across echoes caused by fetal and maternal movement. In this work, we propose a contrast-augmented segmentation framework that leverages complementary information across multi-echo T2*-weighted MRI to learn robust, contrast-invariant representations. Our method integrates: (i) masked autoencoding (MAE) for self-supervised pretraining on unlabeled multi-echo slices; (ii) masked pseudo-labeling (MPL) for unsupervised domain adaptation across echo times; and (iii) global-local collaboration to align fine-grained features with global anatomical context. We further introduce a semantic matching loss to encourage representation consistency across echoes of the same subject. Experiments on a clinical multi-echo placental MRI dataset demonstrate that our approach generalizes effectively across echo times and outperforms both single-echo and naive fusion baselines. To our knowledge, this is the first work to systematically exploit multi-echo T2*-weighted MRI for placental segmentation.
Authors: Karim Abou-Moustafa
Abstract: We consider the problem of estimating a regularization parameter, or a shrinkage coefficient $\alpha \in (0,1)$ for Regularized Tyler's M-estimator (RTME). In particular, we propose to estimate an optimal shrinkage coefficient by setting $\alpha$ as the solution to a suitably chosen objective function; namely the leave-one-out cross-validated (LOOCV) log-likelihood loss. Since LOOCV is computationally prohibitive even for moderate sample size $n$, we propose a computationally efficient approximation for the LOOCV log-likelihood loss that eliminates the need for invoking the RTME procedure $n$ times for each sample left out during the LOOCV procedure. This approximation yields an $O(n)$ reduction in the running time complexity for the LOOCV procedure, which results in a significant speedup for computing the LOOCV estimate. We demonstrate the efficiency and accuracy of the proposed approach on synthetic high-dimensional data sampled from heavy-tailed elliptical distributions, as well as on real high-dimensional datasets for object recognition, face recognition, and handwritten digit's recognition. Our experiments show that the proposed approach is efficient and consistently more accurate than other methods in the literature for shrinkage coefficient estimation.
Authors: Eran Bamani Beeri, Eden Nissinman, Avishai Sintov
Abstract: Dynamic hand gestures play a pivotal role in assistive human-robot interaction (HRI), facilitating intuitive, non-verbal communication, particularly for individuals with mobility constraints or those operating robots remotely. Current gesture recognition methods are mostly limited to short-range interactions, reducing their utility in scenarios demanding robust assistive communication from afar. In this paper, we introduce a novel approach designed specifically for assistive robotics, enabling dynamic gesture recognition at extended distances of up to 30 meters, thereby significantly improving accessibility and quality of life. Our proposed Distance-aware Gesture Network (DiG-Net) effectively combines Depth-Conditioned Deformable Alignment (DADA) blocks with Spatio-Temporal Graph modules, enabling robust processing and classification of gesture sequences captured under challenging conditions, including significant physical attenuation, reduced resolution, and dynamic gesture variations commonly experienced in real-world assistive environments. We further introduce the Radiometric Spatio-Temporal Depth Attenuation Loss (RSTDAL), shown to enhance learning and strengthen model robustness across varying distances. Our model demonstrates significant performance improvement over state-of-the-art gesture recognition frameworks, achieving a recognition accuracy of 97.3% on a diverse dataset with challenging hyper-range gestures. By effectively interpreting gestures from considerable distances, DiG-Net significantly enhances the usability of assistive robots in home healthcare, industrial safety, and remote assistance scenarios, enabling seamless and intuitive interactions for users regardless of physical limitations
Authors: Zimu Liao, Jifeng Ding, Rong Fu, Siwei Cui, Ruixuan Gong, Li Wang, Boni Hu, Yi Wang, Hengjie Li, XIngcheng Zhang, Hui Wang
Abstract: 3D Gaussian Splatting (3DGS) renders pixels by rasterizing Gaussian primitives, where conditional alpha-blending dominates the time cost in the rendering pipeline. This paper proposes TC-GS, an algorithm-independent universal module that expands Tensor Core (TCU) applicability for 3DGS, leading to substantial speedups and seamless integration into existing 3DGS optimization frameworks. The key innovation lies in mapping alpha computation to matrix multiplication, fully utilizing otherwise idle TCUs in existing 3DGS implementations. TC-GS provides plug-and-play acceleration for existing top-tier acceleration algorithms tightly coupled with rendering pipeline designs, like Gaussian compression and redundancy elimination algorithms. Additionally, we introduce a global-to-local coordinate transformation to mitigate rounding errors from quadratic terms of pixel coordinates caused by Tensor Core half-precision computation. Extensive experiments demonstrate that our method maintains rendering quality while providing an additional 2.18x speedup over existing Gaussian acceleration algorithms, thus reaching up to a total 5.6x acceleration. The code is currently available at anonymous \href{https://github.com/TensorCore3DGS/3DGSTensorCore}
Authors: Aditya Retnanto (Asian Development Bank, Philippines), Son Le (Asian Development Bank, Philippines), Sebastian Mueller (Asian Development Bank, Philippines), Armin Leitner (GeoVille Information Systems and Data Processing GmbH, Austria), Konrad Schindler (ETH Z\"urich, Switzerland), Yohan Iddawela (Asian Development Bank, Philippines), Michael Riffler (GeoVille Information Systems and Data Processing GmbH, Austria)
Abstract: Super-resolution aims to increase the resolution of satellite images by reconstructing high-frequency details, which go beyond na\"ive upsampling. This has particular relevance for Earth observation missions like Sentinel-2, which offer frequent, regular coverage at no cost; but at coarse resolution. Its pixel footprint is too large to capture small features like houses, streets, or hedge rows. To address this, we present SEN4X, a hybrid super-resolution architecture that combines the advantages of single-image and multi-image techniques. It combines temporal oversampling from repeated Sentinel-2 acquisitions with a learned prior from high-resolution Pl\'eiades Neo data. In doing so, SEN4X upgrades Sentinel-2 imagery to 2.5 m ground sampling distance. We test the super-resolved images on urban land-cover classification in Hanoi, Vietnam. We find that they lead to a significant performance improvement over state-of-the-art super-resolution baselines.
Authors: Haozhan Tang, Tianyi Zhang, Matthew Johnson-Roberson, Weiming Zhi
Abstract: Robot manipulation, especially bimanual manipulation, often requires setting up multiple cameras on multiple robot manipulators. Before robot manipulators can generate motion or even build representations of their environments, the cameras rigidly mounted to the robot need to be calibrated. Camera calibration is a cumbersome process involving collecting a set of images, with each capturing a pre-determined marker. In this work, we introduce the Bi-Manual Joint Calibration and Representation Framework (Bi-JCR). Bi-JCR enables multiple robot manipulators, each with cameras mounted, to circumvent taking images of calibration markers. By leveraging 3D foundation models for dense, marker-free multi-view correspondence, Bi-JCR jointly estimates: (i) the extrinsic transformation from each camera to its end-effector, (ii) the inter-arm relative poses between manipulators, and (iii) a unified, scale-consistent 3D representation of the shared workspace, all from the same captured RGB image sets. The representation, jointly constructed from images captured by cameras on both manipulators, lives in a common coordinate frame and supports collision checking and semantic segmentation to facilitate downstream bimanual coordination tasks. We empirically evaluate the robustness of Bi-JCR on a variety of tabletop environments, and demonstrate its applicability on a variety of downstream tasks.
Authors: Li yunhan, Wu gengshen
Abstract: As large language models (LLMs) are increasingly used in legal applications, current evaluation benchmarks tend to focus mainly on factual accuracy while largely neglecting important linguistic quality aspects such as clarity, coherence, and terminology. To address this gap, we propose three steps: First, we develop a regression model to evaluate the quality of legal texts based on clarity, coherence, and terminology. Second, we create a specialized set of legal questions. Third, we analyze 49 LLMs using this evaluation framework. Our analysis identifies three key findings: First, model quality levels off at 14 billion parameters, with only a marginal improvement of $2.7\%$ noted at 72 billion parameters. Second, engineering choices such as quantization and context length have a negligible impact, as indicated by statistical significance thresholds above 0.016. Third, reasoning models consistently outperform base architectures. A significant outcome of our research is the release of a ranking list and Pareto analysis, which highlight the Qwen3 series as the optimal choice for cost-performance tradeoffs. This work not only establishes standardized evaluation protocols for legal LLMs but also uncovers fundamental limitations in current training data refinement approaches. Code and models are available at: https://github.com/lyxx3rd/LegalEval-Q.
Authors: Yaxin Luo, Zhaoyi Li, Jiacheng Liu, Jiacheng Cui, Xiaohan Zhao, Zhiqiang Shen
Abstract: CAPTCHAs have been a critical bottleneck for deploying web agents in real-world applications, often blocking them from completing end-to-end automation tasks. While modern multimodal LLM agents have demonstrated impressive performance in static perception tasks, their ability to handle interactive, multi-step reasoning challenges like CAPTCHAs is largely untested. To address this gap, we introduce Open CaptchaWorld, the first web-based benchmark and platform specifically designed to evaluate the visual reasoning and interaction capabilities of MLLM-powered agents through diverse and dynamic CAPTCHA puzzles. Our benchmark spans 20 modern CAPTCHA types, totaling 225 CAPTCHAs, annotated with a new metric we propose: CAPTCHA Reasoning Depth, which quantifies the number of cognitive and motor steps required to solve each puzzle. Experimental results show that humans consistently achieve near-perfect scores, state-of-the-art MLLM agents struggle significantly, with success rates at most 40.0% by Browser-Use Openai-o3, far below human-level performance, 93.3%. This highlights Open CaptchaWorld as a vital benchmark for diagnosing the limits of current multimodal agents and guiding the development of more robust multimodal reasoning systems. Code and Data are available at this https URL.
Authors: David Junhao Zhang, Jay Zhangjie Wu, Jia-Wei Liu, Rui Zhao, Lingmin Ran, Yuchao Gu, Difei Gao, Mike Zheng Shou
Abstract: Significant advancements have been achieved in the realm of large-scale pre-trained text-to-video Diffusion Models (VDMs). However, previous methods either rely solely on pixel-based VDMs, which come with high computational costs, or on latent-based VDMs, which often struggle with precise text-video alignment. In this paper, we are the first to propose a hybrid model, dubbed as Show-1, which marries pixel-based and latent-based VDMs for text-to-video generation. Our model first uses pixel-based VDMs to produce a low-resolution video of strong text-video correlation. After that, we propose a novel expert translation method that employs the latent-based VDMs to further upsample the low-resolution video to high resolution, which can also remove potential artifacts and corruptions from low-resolution videos. Compared to latent VDMs, Show-1 can produce high-quality videos of precise text-video alignment; Compared to pixel VDMs, Show-1 is much more efficient (GPU memory usage during inference is 15G vs 72G). Furthermore, our Show-1 model can be readily adapted for motion customization and video stylization applications through simple temporal attention layer finetuning. Our model achieves state-of-the-art performance on standard video generation benchmarks. Our code and model weights are publicly available at https://github.com/showlab/Show-1.
Authors: Andr\'es Villa, Juan Carlos Le\'on Alc\'azar, Alvaro Soto, Bernard Ghanem
Abstract: Large Vision and Language Models have enabled significant advances in fully supervised and zero-shot visual tasks. These large architectures serve as the baseline to what is currently known as Instruction Tuning Large Vision and Language models (IT-LVLMs). IT-LVLMs are general-purpose multi-modal assistants whose responses are modulated by natural language instructions and visual data. Despite this versatility, IT-LVLM effectiveness in fundamental computer vision problems remains unclear, primarily due to the absence of a standardized evaluation benchmark. This paper introduces a Multi-modal Evaluation Benchmark named MERLIM, a scalable test-bed to assess the capabilities of IT-LVLMs on fundamental computer vision tasks. MERLIM contains over 300K image-question pairs and has a strong focus on detecting cross-modal "hallucination" events in IT-LVLMs. Our results bring important insights on the performance of state-of-the-art IT-LVLMs including limitations at identifying fine-grained visual concepts, object hallucinations across tasks, and biases towards the language query. Our findings also suggest that these models have weak visual grounding, but manage to make adequate guesses from global visual patterns or language biases contained in the LLM component. We name this phenomenon of correct answers with no visual grounding as hidden hallucinations.
Authors: Zhangyang Qi, Ye Fang, Zeyi Sun, Xiaoyang Wu, Tong Wu, Jiaqi Wang, Dahua Lin, Hengshuang Zhao
Abstract: Multimodal Large Language Models (MLLMs) have excelled in 2D image-text comprehension and image generation, but their understanding of the 3D world is notably deficient, limiting progress in 3D language understanding and generation. To solve this problem, we introduce GPT4Point, an innovative groundbreaking point-language multimodal model designed specifically for unified 3D object understanding and generation within the MLLM framework. GPT4Point as a powerful 3D MLLM seamlessly can execute a variety of point-text reference tasks such as point-cloud captioning and Q&A. Additionally, GPT4Point is equipped with advanced capabilities for controllable 3D generation, it can get high-quality results through a low-quality point-text feature maintaining the geometric shapes and colors. To support the expansive needs of 3D object-text pairs, we develop Pyramid-XL, a point-language dataset annotation engine. It constructs a large-scale database over 1M objects of varied text granularity levels from the Objaverse-XL dataset, essential for training GPT4Point. A comprehensive benchmark has been proposed to evaluate 3D point-language understanding capabilities. In extensive evaluations, GPT4Point has demonstrated superior performance in understanding and generation.
Authors: Hanzheng Wang, Wei Li, Xiang-Gen Xia, Qian Du, Jing Tian
Abstract: Hyperspectral video (HSV) offers valuable spatial, spectral, and temporal information simultaneously, making it highly suitable for handling challenges such as background clutter and visual similarity in object tracking. However, existing methods primarily focus on band regrouping and rely on RGB trackers for feature extraction, resulting in limited exploration of spectral information and difficulties in achieving complementary representations of object features. In this paper, a spatial-spectral fusion network with spectral angle awareness (SST-Net) is proposed for hyperspectral (HS) object tracking. Firstly, to address the issue of insufficient spectral feature extraction in existing networks, a spatial-spectral feature backbone ($S^2$FB) is designed. With the spatial and spectral extraction branch, a joint representation of texture and spectrum is obtained. Secondly, a spectral attention fusion module (SAFM) is presented to capture the intra- and inter-modality correlation to obtain the fused features from the HS and RGB modalities. It can incorporate the visual information into the HS spectral context to form a robust representation. Thirdly, to ensure a more accurate response of the tracker to the object position, a spectral angle awareness module (SAAM) investigates the region-level spectral similarity between the template and search images during the prediction stage. Furthermore, we develop a novel spectral angle awareness loss (SAAL) to offer guidance for the SAAM based on similar regions. Finally, to obtain the robust tracking results, a weighted prediction method is considered to combine the HS and RGB predicted motions of objects to leverage the strengths of each modality. Extensive experiments on the HOTC dataset demonstrate the effectiveness of the proposed SSF-Net, compared with state-of-the-art trackers.
Authors: Junhao Cheng, Baiqiao Yin, Kaixin Cai, Minbin Huang, Hanhui Li, Yuxin He, Xi Lu, Yue Li, Yifei Li, Yuhao Cheng, Yiqiang Yan, Xiaodan Liang
Abstract: Recent advances in diffusion models can generate high-quality and stunning images from text. However, multi-turn image generation, which is of high demand in real-world scenarios, still faces challenges in maintaining semantic consistency between images and texts, as well as contextual consistency of the same subject across multiple interactive turns. To address this issue, we introduce TheaterGen, a training-free framework that integrates large language models (LLMs) and text-to-image (T2I) models to provide the capability of multi-turn image generation. Within this framework, LLMs, acting as a "Screenwriter", engage in multi-turn interaction, generating and managing a standardized prompt book that encompasses prompts and layout designs for each character in the target image. Based on these, Theatergen generate a list of character images and extract guidance information, akin to the "Rehearsal". Subsequently, through incorporating the prompt book and guidance information into the reverse denoising process of T2I diffusion models, Theatergen generate the final image, as conducting the "Final Performance". With the effective management of prompt books and character images, TheaterGen significantly improves semantic and contextual consistency in synthesized images. Furthermore, we introduce a dedicated benchmark, CMIGBench (Consistent Multi-turn Image Generation Benchmark) with 8000 multi-turn instructions. Different from previous multi-turn benchmarks, CMIGBench does not define characters in advance. Both the tasks of story generation and multi-turn editing are included on CMIGBench for comprehensive evaluation. Extensive experimental results show that TheaterGen outperforms state-of-the-art methods significantly. It raises the performance bar of the cutting-edge Mini DALLE 3 model by 21% in average character-character similarity and 19% in average text-image similarity.
Authors: Feng Wang, Jiahao Wang, Sucheng Ren, Guoyizhe Wei, Jieru Mei, Wei Shao, Yuyin Zhou, Alan Yuille, Cihang Xie
Abstract: Similar to Vision Transformers, this paper identifies artifacts also present within the feature maps of Vision Mamba. These artifacts, corresponding to high-norm tokens emerging in low-information background areas of images, appear much more severe in Vision Mamba -- they exist prevalently even with the tiny-sized model and activate extensively across background regions. To mitigate this issue, we follow the prior solution of introducing register tokens into Vision Mamba. To better cope with Mamba blocks' uni-directional inference paradigm, two key modifications are introduced: 1) evenly inserting registers throughout the input token sequence, and 2) recycling registers for final decision predictions. We term this new architecture Mamba-R. Qualitative observations suggest, compared to vanilla Vision Mamba, Mamba-R's feature maps appear cleaner and more focused on semantically meaningful regions. Quantitatively, Mamba-R attains stronger performance and scales better. For example, on the ImageNet benchmark, our base-size Mamba-R attains 83.0% accuracy, significantly outperforming Vim-B's 81.8%; furthermore, we provide the first successful scaling to the large model size (i.e., with 341M parameters), attaining a competitive accuracy of 83.6% (84.5% if finetuned with 384x384 inputs). Additional validation on the downstream semantic segmentation task also supports Mamba-R's efficacy. Code is available at https://github.com/wangf3014/Mamba-Reg.
Authors: Chaoyou Fu, Yuhan Dai, Yongdong Luo, Lei Li, Shuhuai Ren, Renrui Zhang, Zihan Wang, Chenyu Zhou, Yunhang Shen, Mengdan Zhang, Peixian Chen, Yanwei Li, Shaohui Lin, Sirui Zhao, Ke Li, Tong Xu, Xiawu Zheng, Enhong Chen, Caifeng Shan, Ran He, Xing Sun
Abstract: In the quest for artificial general intelligence, Multi-modal Large Language Models (MLLMs) have emerged as a focal point in recent advancements. However, the predominant focus remains on developing their capabilities in static image understanding. The potential of MLLMs in processing sequential visual data is still insufficiently explored, highlighting the absence of a comprehensive, high-quality assessment of their performance. In this paper, we introduce Video-MME, the first-ever full-spectrum, Multi-Modal Evaluation benchmark of MLLMs in Video analysis. Our work distinguishes from existing benchmarks through four key features: 1) Diversity in video types, spanning 6 primary visual domains with 30 subfields to ensure broad scenario generalizability; 2) Duration in temporal dimension, encompassing both short-, medium-, and long-term videos, ranging from 11 seconds to 1 hour, for robust contextual dynamics; 3) Breadth in data modalities, integrating multi-modal inputs besides video frames, including subtitles and audios, to unveil the all-round capabilities of MLLMs; 4) Quality in annotations, utilizing rigorous manual labeling by expert annotators to facilitate precise and reliable model assessment. 900 videos with a total of 254 hours are manually selected and annotated by repeatedly viewing all the video content, resulting in 2,700 question-answer pairs. With Video-MME, we extensively evaluate various state-of-the-art MLLMs, including GPT-4 series and Gemini 1.5 Pro, as well as open-source image models like InternVL-Chat-V1.5 and video models like LLaVA-NeXT-Video. Our experiments reveal that Gemini 1.5 Pro is the best-performing commercial model, significantly outperforming the open-source models. Our dataset along with these findings underscores the need for further improvements in handling longer sequences and multi-modal data. Project Page: https://video-mme.github.io
Authors: Junhao Cheng, Xi Lu, Hanhui Li, Khun Loun Zai, Baiqiao Yin, Yuhao Cheng, Yiqiang Yan, Xiaodan Liang
Abstract: As cutting-edge Text-to-Image (T2I) generation models already excel at producing remarkable single images, an even more challenging task, i.e., multi-turn interactive image generation begins to attract the attention of related research communities. This task requires models to interact with users over multiple turns to generate a coherent sequence of images. However, since users may switch subjects frequently, current efforts struggle to maintain subject consistency while generating diverse images. To address this issue, we introduce a training-free multi-agent framework called AutoStudio. AutoStudio employs three agents based on large language models (LLMs) to handle interactions, along with a stable diffusion (SD) based agent for generating high-quality images. Specifically, AutoStudio consists of (i) a subject manager to interpret interaction dialogues and manage the context of each subject, (ii) a layout generator to generate fine-grained bounding boxes to control subject locations, (iii) a supervisor to provide suggestions for layout refinements, and (iv) a drawer to complete image generation. Furthermore, we introduce a Parallel-UNet to replace the original UNet in the drawer, which employs two parallel cross-attention modules for exploiting subject-aware features. We also introduce a subject-initialized generation method to better preserve small subjects. Our AutoStudio hereby can generate a sequence of multi-subject images interactively and consistently. Extensive experiments on the public CMIGBench benchmark and human evaluations show that AutoStudio maintains multi-subject consistency across multiple turns well, and it also raises the state-of-the-art performance by 13.65% in average Frechet Inception Distance and 2.83% in average character-character similarity.
Authors: Samar Khanna, Medhanie Irgau, David B. Lobell, Stefano Ermon
Abstract: Parameter-efficient fine-tuning (PEFT) techniques such as low-rank adaptation (LoRA) can effectively adapt large pre-trained foundation models to downstream tasks using only a small fraction (0.1%-10%) of the original trainable weights. An under-explored question of PEFT is in extending the pre-training phase without supervised labels; that is, can we adapt a pre-trained foundation model to a new domain via efficient self-supervised pre-training on this domain? In this work, we introduce ExPLoRA, a highly effective technique to improve transfer learning of pre-trained vision transformers (ViTs) under domain shifts. Initializing a ViT with pre-trained weights on large, natural-image datasets such as from DinoV2 or MAE, ExPLoRA continues the unsupervised pre-training objective on a new domain, unfreezing 1-2 pre-trained ViT blocks and tuning all other layers with LoRA. We then fine-tune the resulting model only with LoRA on this new domain for supervised learning. Our experiments demonstrate state-of-the-art results on satellite imagery, even outperforming fully pre-training and fine-tuning ViTs. Using the DinoV2 training objective, we demonstrate up to 8% improvement in linear probing top-1 accuracy on downstream tasks while using <10% of the number of parameters that are used in prior fully-tuned state-of-the art approaches. Our ablation studies confirm the efficacy of our approach over other baselines such as PEFT. Code is available on the project website: https://samar-khanna.github.io/ExPLoRA/
Authors: Chaoyou Fu, Haojia Lin, Zuwei Long, Yunhang Shen, Yuhang Dai, Meng Zhao, Yi-Fan Zhang, Shaoqi Dong, Yangze Li, Xiong Wang, Haoyu Cao, Di Yin, Long Ma, Xiawu Zheng, Rongrong Ji, Yunsheng Wu, Ran He, Caifeng Shan, Xing Sun
Abstract: The remarkable multimodal capabilities and interactive experience of GPT-4o underscore their necessity in practical applications, yet open-source models rarely excel in both areas. In this paper, we introduce VITA, the first-ever open-source Multimodal Large Language Model (MLLM) adept at simultaneous processing and analysis of Video, Image, Text, and Audio modalities, and meanwhile has an advanced multimodal interactive experience. Starting from Mixtral 8x7B as a language foundation, we expand its Chinese vocabulary followed by bilingual instruction tuning. We further endow the language model with visual and audio capabilities through two-stage multi-task learning of multimodal alignment and instruction tuning. VITA demonstrates robust foundational capabilities of multilingual, vision, and audio understanding, as evidenced by its strong performance across a range of both unimodal and multimodal benchmarks. Beyond foundational capabilities, we have made considerable progress in enhancing the natural multimodal human-computer interaction experience. VITA is the first step for the open-source community to explore the seamless integration of multimodal understanding and interaction. While there is still lots of work to be done on VITA to get close to close-source counterparts, we hope that its role as a pioneer can serve as a cornerstone for subsequent research. Project Page: https://vita-home.github.io.
Authors: Yuxia Geng, Runkai Zhu, Jiaoyan Chen, Jintai Chen, Xiang Chen, Zhuo Chen, Shuofei Qiao, Yuxiang Wang, Xiaoliang Xu, Sheng-Jun Huang
Abstract: Disentanglement of visual features of primitives (i.e., attributes and objects) has shown exceptional results in Compositional Zero-shot Learning (CZSL). However, due to the feature divergence of an attribute (resp. object) when combined with different objects (resp. attributes), it is challenging to learn disentangled primitive features that are general across different compositions. To this end, we propose the solution of cross-composition feature disentanglement, which takes multiple primitive-sharing compositions as inputs and constrains the disentangled primitive features to be general across these compositions. More specifically, we leverage a compositional graph to define the overall primitive-sharing relationships between compositions, and build a task-specific architecture upon the recently successful large pre-trained vision-language model (VLM) CLIP, with dual cross-composition disentangling adapters (called L-Adapter and V-Adapter) inserted into CLIP's frozen text and image encoders, respectively. Evaluation on three popular CZSL benchmarks shows that our proposed solution significantly improves the performance of CZSL, and its components have been verified by solid ablation studies. Our code and data are available at:https://github.com/zhurunkai/DCDA.
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: Giorgos Savathrakis, Antonis Argyros
Abstract: Transformers demonstrate competitive performance in terms of precision on the problem of vision-based object detection. However, they require considerable computational resources due to the quadratic size of the attention weights. In this work, we propose to cluster the transformer input on the basis of its entropy, due to its similarity between same object pixels. This is expected to reduce GPU usage during training, while maintaining reasonable accuracy. This idea is realized with an implemented module that is called ENtropy-based Attention Clustering for detection Transformers (ENACT), which serves as a plug-in to any multi-head self-attention based transformer network. Experiments on the COCO object detection dataset and three detection transformers demonstrate that the requirements on memory are reduced, while the detection accuracy is degraded only slightly. The code of the ENACT module is available at https://github.com/GSavathrakis/ENACT.
Authors: Taekyung Lee, Jaemoo Choi, Jaewoong Choi, Myungjoo Kang
Abstract: Unpaired point cloud completion is crucial for real-world applications, where ground-truth data for complete point clouds are often unavailable. By learning a completion map from unpaired incomplete and complete point cloud data, this task avoids the reliance on paired datasets. In this paper, we propose the \textit{Unbalanced Optimal Transport Map for Unpaired Point Cloud Completion (\textbf{UOT-UPC})} model, which formulates the unpaired completion task as the (Unbalanced) Optimal Transport (OT) problem. Our method employs a Neural OT model learning the UOT map using neural networks. Our model is the first attempt to leverage UOT for unpaired point cloud completion, achieving competitive or superior performance on both single-category and multi-category benchmarks. In particular, our approach is especially robust under the class imbalance problem, which is frequently encountered in real-world unpaired point cloud completion scenarios. The code is available at https://github.com/LEETK99/UOT-UPC.
Authors: Ye Wang, Sipeng Zheng, Bin Cao, Qianshan Wei, Weishuai Zeng, Qin Jin, Zongqing Lu
Abstract: Inspired by the recent success of LLMs, the field of human motion understanding has increasingly shifted toward developing large motion models. Despite some progress, current efforts remain far from achieving truly generalist models, primarily due to the lack of massive high-quality data. To address this gap, we present MotionLib, the first million-level dataset for motion generation, which is at least 15$\times$ larger than existing counterparts and enriched with hierarchical text descriptions. Using MotionLib, we train a large motion model named \projname, demonstrating robust performance across a wide range of human activities, including unseen ones. Through systematic investigation, for the first time, we highlight the importance of scaling both data and model size for advancing motion generation, along with key insights to achieve this goal. To better integrate the motion modality, we propose Motionbook, an innovative motion encoding approach including (1) a compact yet lossless feature to represent motions; (2) a novel 2D lookup-free motion tokenizer that preserves fine-grained motion details while expanding codebook capacity, significantly enhancing the representational power of motion tokens. We believe this work lays the groundwork for developing more versatile and powerful motion generation models in the future. For further details, visit https://beingbeyond.github.io/Being-M0/.
Authors: Feng Wang, Timing Yang, Yaodong Yu, Sucheng Ren, Guoyizhe Wei, Angtian Wang, Wei Shao, Yuyin Zhou, Alan Yuille, Cihang Xie
Abstract: In this work, we introduce the Adventurer series models where we treat images as sequences of patch tokens and employ uni-directional language models to learn visual representations. This modeling paradigm allows us to process images in a recurrent formulation with linear complexity relative to the sequence length, which can effectively address the memory and computation explosion issues posed by high-resolution and fine-grained images. In detail, we introduce two simple designs that seamlessly integrate image inputs into the causal inference framework: a global pooling token placed at the beginning of the sequence and a flipping operation between every two layers. Extensive empirical studies highlight that compared with the existing plain architectures such as DeiT and Vim, Adventurer offers an optimal efficiency-accuracy trade-off. For example, our Adventurer-Base attains a competitive test accuracy of 84.3% on the standard ImageNet-1k benchmark with 216 images/s training throughput, which is 3.8 and 6.2 times faster than Vim and DeiT to achieve the same result. As Adventurer offers great computation and memory efficiency and allows scaling with linear complexity, we hope this architecture can benefit future explorations in modeling long sequences for high-resolution or fine-grained images. Code is available at https://github.com/wangf3014/Adventurer.
Authors: Jingqi Zhou, Sheng Wang, Jingwei Dong, Kai Liu, Lei Li, Jiahui Gao, Jiyue Jiang, Lingpeng Kong, Chuan Wu
Abstract: Large vision-language models (LVLMs) have witnessed significant progress on visual understanding tasks. However, they often prioritize language knowledge over image information on visual reasoning tasks, incurring performance degradation. To tackle this issue, we first identify the drawbacks of existing solutions (i.e., limited multi-modal reasoning capacities, and insufficient and irrelevant visual descriptions). We then decompose visual reasoning process into two stages: proactive visual perception (i.e., eyesight) and textual reasoning (i.e., wisdom), and introduce a novel visual reasoning framework named ProReason. This framework features decoupled vision-reasoning capabilities and multi-run proactive perception. Briefly, given a multi-modal question, ProReason iterates proactive information collection and reasoning until the answer can be concluded with necessary and sufficient visual descriptions. Notably, the disassociation of capabilities allows seamless integration of existing large language models (LLMs) to compensate for the reasoning deficits of LVLMs. Our extensive experiments demonstrate that ProReason outperforms existing multi-step reasoning frameworks on various benchmarks for both open-source and closed-source models, with the average performance gain reaching 13.2%. Besides, the integration of LLMs allows ProReason to produce high-quality visual reasoning data, which empowers ProReason-distilled models (i.e., ProReason-VL and ProReason-Q3) to achieve superior performance in downstream tasks. Our insights into existing solutions and the decoupled perspective for feasible integration of LLMs illuminate future research on visual reasoning techniques, especially LLM-assisted ones.
Authors: Yongrui Yu, Yannian Gu, Shaoting Zhang, Xiaofan Zhang
Abstract: Diffusion models have achieved significant success in both natural image and medical image domains, encompassing a wide range of applications. Previous investigations in medical images have often been constrained to specific anatomical regions, particular applications, and limited datasets, resulting in isolated diffusion models. This paper introduces a diffusion-based foundation model to address a diverse range of medical image tasks, namely MedDiff-FM. MedDiff-FM leverages 3D CT images from multiple publicly available datasets, covering anatomical regions from head to abdomen, to pre-train a diffusion foundation model, and explores the capabilities of the diffusion foundation model across a variety of application scenarios. The diffusion foundation model handles multi-level integrated image processing both at the image-level and patch-level, utilizes position embedding to establish multi-level spatial relationships, and leverages region classes and anatomical structures to capture certain anatomical regions. MedDiff-FM manages several downstream tasks seamlessly, including image denoising, anomaly detection, and image synthesis. MedDiff-FM is also capable of performing super-resolution, lesion generation, and lesion inpainting by rapidly fine-tuning the diffusion foundation model using ControlNet with task-specific conditions. The experimental results demonstrate the effectiveness of MedDiff-FM in addressing diverse downstream medical image tasks.
Authors: Nitish Shukla, Arun Ross
Abstract: A facial morph is an image strategically created by combining two face images pertaining to two distinct identities. The goal is to create a face image that can be matched to two different identities by a face matcher. Face demorphing inverts this process and attempts to recover the original images constituting a facial morph. Existing demorphing techniques have two major limitations: (a) they assume that some identities are common in the train and test sets; and (b) they are prone to the morph replication problem, where the outputs are merely replicates of the input morph. In this paper, we overcome these issues by proposing dc-GAN (dual-conditioned GAN), a novel demorphing method conditioned on the morph image as well as the embedding extracted from the image. Our method overcomes the morph replication problem and produces high-fidelity reconstructions of the constituent images. Moreover, the proposed method is highly generalizable and applicable to both reference-based and reference-free demorphing methods. Experiments were conducted using the AMSL, FRLL-Morphs, and MorDiff datasets to demonstrate the efficacy of the method.
Authors: Dahye Kim, Xavier Thomas, Deepti Ghadiyaram
Abstract: We study $\textit{how}$ rich visual semantic information is represented within various layers and denoising timesteps of different diffusion architectures. We uncover monosemantic interpretable features by leveraging k-sparse autoencoders (k-SAE). We substantiate our mechanistic interpretations via transfer learning using light-weight classifiers on off-the-shelf diffusion models' features. On $4$ datasets, we demonstrate the effectiveness of diffusion features for representation learning. We provide an in-depth analysis of how different diffusion architectures, pre-training datasets, and language model conditioning impacts visual representation granularity, inductive biases, and transfer learning capabilities. Our work is a critical step towards deepening interpretability of black-box diffusion models. Code and visualizations available at: https://github.com/revelio-diffusion/revelio
Authors: Fan Yang, Ru Zhen, Jianing Wang, Yanhao Zhang, Haoxiang Chen, Haonan Lu, Sicheng Zhao, Guiguang Ding
Abstract: AIGC images are prevalent across various fields, yet they frequently suffer from quality issues like artifacts and unnatural textures. Specialized models aim to predict defect region heatmaps but face two primary challenges: (1) lack of explainability, failing to provide reasons and analyses for subtle defects, and (2) inability to leverage common sense and logical reasoning, leading to poor generalization. Multimodal large language models (MLLMs) promise better comprehension and reasoning but face their own challenges: (1) difficulty in fine-grained defect localization due to the limitations in capturing tiny details, and (2) constraints in providing pixel-wise outputs necessary for precise heatmap generation. To address these challenges, we propose HEIE: a novel MLLM-Based Hierarchical Explainable Image Implausibility Evaluator. We introduce the CoT-Driven Explainable Trinity Evaluator, which integrates heatmaps, scores, and explanation outputs, using CoT to decompose complex tasks into subtasks of increasing difficulty and enhance interpretability. Our Adaptive Hierarchical Implausibility Mapper synergizes low-level image features with high-level mapper tokens from LLMs, enabling precise local-to-global hierarchical heatmap predictions through an uncertainty-based adaptive token approach. Moreover, we propose a new dataset: Expl-AIGI-Eval, designed to facilitate interpretable implausibility evaluation of AIGC images. Our method demonstrates state-of-the-art performance through extensive experiments. Our project is at https://yfthu.github.io/HEIE/.
Authors: Xinyu Zhang, Lingling Zhang, Yanrui Wu, Muye Huang, Wenjun Wu, Bo Li, Shaowei Wang, Basura Fernando, Jun Liu
Abstract: Visual Question Generation (VQG) research focuses predominantly on natural images while neglecting the diagram, which is a critical component in educational materials. To meet the needs of pedagogical assessment, we propose the Diagram-Driven Course Questions Generation (DDCQG) task and construct DiagramQG, a comprehensive dataset with 15,720 diagrams and 25,798 questions across 37 subjects and 371 courses. Our approach employs course and input text constraints to generate course-relevant questions about specific diagram elements. We reveal three challenges of DDCQG: domain-specific knowledge requirements across courses, long-tail distribution in course coverage, and high information density in diagrams. To address these, we propose the Hierarchical Knowledge Integration framework (HKI-DDCQG), which utilizes trainable CLIP for identifying relevant diagram patches, leverages frozen vision-language models for knowledge extraction, and generates questions with trainable T5. Experiments demonstrate that HKI-DDCQG outperforms existing models on DiagramQG while maintaining strong generalizability across natural image datasets, establishing a strong baseline for DDCQG.
Authors: Alice Heiman, Xiaoman Zhang, Emma Chen, Sung Eun Kim, Pranav Rajpurkar
Abstract: Medical vision-language model models often struggle with generating accurate quantitative measurements in radiology reports, leading to hallucinations that undermine clinical reliability. We introduce FactCheXcker, a modular framework that de-hallucinates radiology report measurements by leveraging an improved query-code-update paradigm. Specifically, FactCheXcker employs specialized modules and the code generation capabilities of large language models to solve measurement queries generated based on the original report. After extracting measurable findings, the results are incorporated into an updated report. We evaluate FactCheXcker on endotracheal tube placement, which accounts for an average of 78% of report measurements, using the MIMIC-CXR dataset and 11 medical report-generation models. Our results show that FactCheXcker significantly reduces hallucinations, improves measurement precision, and maintains the quality of the original reports. Specifically, FactCheXcker improves the performance of all 11 models and achieves an average improvement of 94.0% in reducing measurement hallucinations measured by mean absolute error.
Authors: Jeongsoo Choi, Ji-Hoon Kim, Jinyu Li, Joon Son Chung, Shujie Liu
Abstract: In this paper, we introduce V2SFlow, a novel Video-to-Speech (V2S) framework designed to generate natural and intelligible speech directly from silent talking face videos. While recent V2S systems have shown promising results on constrained datasets with limited speakers and vocabularies, their performance often degrades on real-world, unconstrained datasets due to the inherent variability and complexity of speech signals. To address these challenges, we decompose the speech signal into manageable subspaces (content, pitch, and speaker information), each representing distinct speech attributes, and predict them directly from the visual input. To generate coherent and realistic speech from these predicted attributes, we employ a rectified flow matching decoder built on a Transformer architecture, which models efficient probabilistic pathways from random noise to the target speech distribution. Extensive experiments demonstrate that V2SFlow significantly outperforms state-of-the-art methods, even surpassing the naturalness of ground truth utterances. Code and models are available at: https://github.com/kaistmm/V2SFlow
Authors: Muhammad Fetrat Qharabagh, Mohammadreza Ghofrani, Kimon Fountoulakis
Abstract: Counting is a fundamental operation for various real-world visual tasks, requiring both object recognition and robust counting capabilities. Despite their advanced visual perception, large vision-language models (LVLMs) are known to struggle with counting tasks. In this work, we evaluate the performance of several recent LVLMs on visual counting tasks across multiple counting and vision datasets. We observe that while their performance may be less prone to error for small numbers of objects, they exhibit significant weaknesses as the number of objects increases. To alleviate this issue, we propose a simple yet effective baseline method that enhances LVLMs' counting ability for large numbers of objects using a divide-and-conquer approach. Our method decomposes counting problems into sub-tasks. Moreover, it incorporates a mechanism to prevent objects from being split during division, which could otherwise lead to repetitive counting -- a common issue in a naive divide-and-conquer implementation. We demonstrate the effectiveness of this approach across various datasets and benchmarks, establishing it as a valuable reference for evaluating future solutions.
Authors: Jianfeng Xiang, Zelong Lv, Sicheng Xu, Yu Deng, Ruicheng Wang, Bowen Zhang, Dong Chen, Xin Tong, Jiaolong Yang
Abstract: We introduce a novel 3D generation method for versatile and high-quality 3D asset creation. The cornerstone is a unified Structured LATent (SLAT) representation which allows decoding to different output formats, such as Radiance Fields, 3D Gaussians, and meshes. This is achieved by integrating a sparsely-populated 3D grid with dense multiview visual features extracted from a powerful vision foundation model, comprehensively capturing both structural (geometry) and textural (appearance) information while maintaining flexibility during decoding. We employ rectified flow transformers tailored for SLAT as our 3D generation models and train models with up to 2 billion parameters on a large 3D asset dataset of 500K diverse objects. Our model generates high-quality results with text or image conditions, significantly surpassing existing methods, including recent ones at similar scales. We showcase flexible output format selection and local 3D editing capabilities which were not offered by previous models. Code, model, and data will be released.
Authors: Anton Baumann, Rui Li, Marcus Klasson, Santeri Mentu, Shyamgopal Karthik, Zeynep Akata, Arno Solin, Martin Trapp
Abstract: Vision-language models (VLMs), such as CLIP and SigLIP, have found remarkable success in classification, retrieval, and generative tasks. For this, VLMs deterministically map images and text descriptions to a joint latent space in which their similarity is assessed using the cosine similarity. However, a deterministic mapping of inputs fails to capture uncertainties over concepts arising from domain shifts when used in downstream tasks. In this work, we propose post-hoc uncertainty estimation in VLMs that does not require additional training. Our method leverages a Bayesian posterior approximation over the last layers in VLMs and analytically quantifies uncertainties over cosine similarities. We demonstrate its effectiveness for uncertainty quantification and support set selection in active learning. Compared to baselines, we obtain improved and well-calibrated predictive uncertainties, interpretable uncertainty estimates, and sample-efficient active learning. Our results show promise for safety-critical applications of large-scale models.
Authors: Liyang He, Yuren Zhang, Rui Li, Zhenya Huang, Runze Wu, Enhong Chen
Abstract: Deep supervised hashing is essential for efficient storage and search in large-scale image retrieval. Traditional deep supervised hashing models generate single-length hash codes, but this creates a trade-off between efficiency and effectiveness for different code lengths. To find the optimal length for a task, multiple models must be trained, increasing time and computation. Furthermore, relationships between hash codes of different lengths are often ignored. To address these issues, we propose the Nested Hash Layer (NHL), a plug-and-play module for deep supervised hashing models. NHL generates hash codes of multiple lengths simultaneously in a nested structure. To resolve optimization conflicts from multiple learning objectives, we introduce a dominance-aware dynamic weighting strategy to adjust gradients. Additionally, we propose a long-short cascade self-distillation method, where long hash codes guide the learning of shorter ones, improving overall code quality. Experiments indicate that the NHL achieves an overall training speed improvement of approximately 5 to 8 times across various deep supervised hashing models and enhances the average performance of these models by about 3.4%.
Authors: Wengyi Zhan, Mingbao Lin, Shuicheng Yan, Rongrong Ji
Abstract: We introduce DiffusionTrend for virtual fashion try-on, which forgoes the need for retraining diffusion models. Using advanced diffusion models, DiffusionTrend harnesses latent information rich in prior information to capture the nuances of garment details. Throughout the diffusion denoising process, these details are seamlessly integrated into the model image generation, expertly directed by a precise garment mask crafted by a lightweight and compact CNN. Although our DiffusionTrend model initially demonstrates suboptimal metric performance, our exploratory approach offers some important advantages: (1) It circumvents resource-intensive retraining of diffusion models on large datasets. (2) It eliminates the necessity for various complex and user-unfriendly model inputs. (3) It delivers a visually compelling try-on experience, underscoring the potential of training-free diffusion model. This initial foray into the application of untrained diffusion models in virtual try-on technology potentially paves the way for further exploration and refinement in this industrially and academically valuable field.
Authors: Mang Ning, Mingxiao Li, Jianlin Su, Haozhe Jia, Lanmiao Liu, Martin Bene\v{s}, Wenshuo Chen, Albert Ali Salah, Itir Onal Ertugrul
Abstract: This paper explores image modeling from the frequency space and introduces DCTdiff, an end-to-end diffusion generative paradigm that efficiently models images in the discrete cosine transform (DCT) space. We investigate the design space of DCTdiff and reveal the key design factors. Experiments on different frameworks (UViT, DiT), generation tasks, and various diffusion samplers demonstrate that DCTdiff outperforms pixel-based diffusion models regarding generative quality and training efficiency. Remarkably, DCTdiff can seamlessly scale up to 512$\times$512 resolution without using the latent diffusion paradigm and beats latent diffusion (using SD-VAE) with only 1/4 training cost. Finally, we illustrate several intriguing properties of DCT image modeling. For example, we provide a theoretical proof of why 'image diffusion can be seen as spectral autoregression', bridging the gap between diffusion and autoregressive models. The effectiveness of DCTdiff and the introduced properties suggest a promising direction for image modeling in the frequency space. The code is https://github.com/forever208/DCTdiff.
Authors: Yitong Zhou, Mingyue Cheng, Qingyang Mao, Feiyang Xu, Xin Li
Abstract: Pre-trained foundation models have recently made significant progress in table-related tasks such as table understanding and reasoning. However, recognizing the structure and content of unstructured tables using Vision Large Language Models (VLLMs) remains under-explored. To bridge this gap, we propose a benchmark based on a hierarchical design philosophy to evaluate the recognition capabilities of VLLMs in training-free scenarios. Through in-depth evaluations, we find that low-quality image input is a significant bottleneck in the recognition process. Drawing inspiration from this, we propose the Neighbor-Guided Toolchain Reasoner (NGTR) framework, which is characterized by integrating diverse lightweight tools for visual operations aimed at mitigating issues with low-quality images. Specifically, we transfer a tool selection experience from a similar neighbor to the input and design a reflection module to supervise the tool invocation process. Extensive experiments on public datasets demonstrate that our approach significantly enhances the recognition capabilities of the vanilla VLLMs. We believe that the benchmark and framework could provide an alternative solution to table recognition.
Authors: Hongliang Li, Jiaxin Zhang, Wenhui Liao, Dezhi Peng, Kai Ding, Lianwen Jin
Abstract: Current Multimodal Large Language Model (MLLM) architectures face a critical tradeoff between performance and efficiency: decoder-only architectures achieve higher performance but lower efficiency, while cross-attention-based architectures offer greater efficiency but lower performance. The key distinction lies in how visual tokens are processed. Decoder-only architectures apply self-attention and FFN operations on visual tokens, while cross-attention architectures skip these computations. To investigate whether redundancy exists in this computationally expensive process, we propose a training-free framework for analyzing trained MLLMs. It consists of Probe-Activated Dynamic FFN and Hollow Attention, which enable adjustable reductions in computations for visual tokens, as well as a Layer Ranking Algorithm that prioritizes layers for these reductions. Extensive experiments demonstrate substantial, structured, and clustered redundancy unique to decoder-only MLLMs, offering valuable insights for future MLLM architecture design. Furthermore, by leveraging our reduction framework as a training-free inference acceleration approach, we achieve performance comparable to or better than state-of-the-art methods while remaining compatible with them. Code will be publicly available at https://github.com/L-Hugh/RedundancyLens.
Authors: Junhao Cheng, Wei-Ting Chen, Xi Lu, Ming-Hsuan Yang
Abstract: Generative diffusion models trained on large-scale datasets have achieved remarkable progress in image synthesis. In favor of their ability to supplement missing details and generate aesthetically pleasing contents, recent works have applied them to image deblurring via training an adapter on blurry-sharp image pairs to provide structural conditions for restoration. However, acquiring substantial amounts of realistic paired data is challenging and costly in real-world scenarios. On the other hand, relying solely on synthetic data often results in overfitting, leading to unsatisfactory performance when confronted with unseen blur patterns. To tackle this issue, we propose UID-Diff, a generative-diffusion-based model designed to enhance deblurring performance on unknown domains by decoupling structural features and blur patterns through joint training on three specially designed tasks. We employ two Q-Formers as structural features and blur patterns extractors separately. The features extracted by them will be used for the supervised deblurring task on synthetic data and the unsupervised blur-transfer task by leveraging unpaired blurred images from the target domain simultaneously. We further introduce a reconstruction task to make the structural features and blur patterns complementary. This blur-decoupled learning process enhances the generalization capabilities of UID-Diff when encountering unknown blur patterns. Experiments on real-world datasets demonstrate that UID-Diff outperforms existing state-of-the-art methods in blur removal and structural preservation in various challenging scenarios.
Authors: Hao Chen, Yujin Han, Fangyi Chen, Xiang Li, Yidong Wang, Jindong Wang, Ze Wang, Zicheng Liu, Difan Zou, Bhiksha Raj
Abstract: Recent advances in latent diffusion models have demonstrated their effectiveness for high-resolution image synthesis. However, the properties of the latent space from tokenizer for better learning and generation of diffusion models remain under-explored. Theoretically and empirically, we find that improved generation quality is closely tied to the latent distributions with better structure, such as the ones with fewer Gaussian Mixture modes and more discriminative features. Motivated by these insights, we propose MAETok, an autoencoder (AE) leveraging mask modeling to learn semantically rich latent space while maintaining reconstruction fidelity. Extensive experiments validate our analysis, demonstrating that the variational form of autoencoders is not necessary, and a discriminative latent space from AE alone enables state-of-the-art performance on ImageNet generation using only 128 tokens. MAETok achieves significant practical improvements, enabling a gFID of 1.69 with 76x faster training and 31x higher inference throughput for 512x512 generation. Our findings show that the structure of the latent space, rather than variational constraints, is crucial for effective diffusion models. Code and trained models are released.
Authors: Yibo Xu, Dawei Zhou, Decheng Liu, Nannan Wang
Abstract: Deep neural networks are found to be vulnerable to adversarial perturbations. The prompt-based defense has been increasingly studied due to its high efficiency. However, existing prompt-based defenses mainly exploited mixed prompt patterns, where critical patterns closely related to object semantics lack sufficient focus. The phase and amplitude spectra have been proven to be highly related to specific semantic patterns and crucial for robustness. To this end, in this paper, we propose a Phase and Amplitude-aware Prompting (PAP) defense. Specifically, we construct phase-level and amplitude-level prompts for each class, and adjust weights for prompting according to the model's robust performance under these prompts during training. During testing, we select prompts for each image using its predicted label to obtain the prompted image, which is inputted to the model to get the final prediction. Experimental results demonstrate the effectiveness of our method.
Authors: Xilin Wei, Xiaoran Liu, Yuhang Zang, Xiaoyi Dong, Pan Zhang, Yuhang Cao, Jian Tong, Haodong Duan, Qipeng Guo, Jiaqi Wang, Xipeng Qiu, Dahua Lin
Abstract: While Rotary Position Embedding (RoPE) and its variants are widely adopted for their long-context capabilities, the extension of the 1D RoPE to video, with its complex spatio-temporal structure, remains an open challenge. This work first introduces a comprehensive analysis that identifies four key characteristics essential for the effective adaptation of RoPE to video, which have not been fully considered in prior work. As part of our analysis, we introduce a challenging V-NIAH-D (Visual Needle-In-A-Haystack with Distractors) task, which adds periodic distractors into V-NIAH. The V-NIAH-D task demonstrates that previous RoPE variants, lacking appropriate temporal dimension allocation, are easily misled by distractors. Based on our analysis, we introduce \textbf{VideoRoPE}, with a \textit{3D structure} designed to preserve spatio-temporal relationships. VideoRoPE features \textit{low-frequency temporal allocation} to mitigate periodic oscillations, a \textit{diagonal layout} to maintain spatial symmetry, and \textit{adjustable temporal spacing} to decouple temporal and spatial indexing. VideoRoPE consistently surpasses previous RoPE variants, across diverse downstream tasks such as long video retrieval, video understanding, and video hallucination. Our code will be available at \href{https://github.com/Wiselnn570/VideoRoPE}{https://github.com/Wiselnn570/VideoRoPE}.
URLs: https://github.com/Wiselnn570/VideoRoPE, https://github.com/Wiselnn570/VideoRoPE
Authors: Kafi Anan, Anindya Bhattacharjee, Ashir Intesher, Kaidul Islam, Abrar Assaeem Fuad, Utsab Saha, Hafiz Imtiaz
Abstract: Effective deepfake detection tools are becoming increasingly essential to the growing usage of deepfakes in unethical practices. There exists a wide range of deepfake generation techniques, which makes it challenging to develop an accurate universal detection mechanism. The 2025 IEEE Signal Processing Cup (\textit{DFWild-Cup} competition) provided a diverse dataset of deepfake images containing significant class imbalance. The images in the dataset are generated from multiple deepfake image generators, for training machine learning model(s) to emphasize the generalization of deepfake detection. To this end, we proposed a disjoint set-based multistage training method to address the class imbalance and devised an ensemble-based architecture \emph{CAE-Net}. Our architecture consists of a convolution- and attention-based ensemble network, and employs three different neural network architectures: EfficientNet, Data-Efficient Image Transformer (DeiT), and ConvNeXt with wavelet transform to capture both local and global features of deepfakes. We visualize the specific regions that these models focus on for classification using Grad-CAM, and empirically demonstrate the effectiveness of these models in grouping real and fake images into cohesive clusters using t-SNE plots. Individually, the EfficientNet B0 architecture has achieved 90.79\% accuracy, whereas the ConvNeXt and the DeiT architecture have achieved 89.49\% and 89.32\% accuracy, respectively. With these networks, our weighted ensemble model achieves an excellent accuracy of 94.63\% on the validation dataset of the SP Cup 2025 competition. The equal error rate of 4.72\% and the Area Under the ROC curve of 97.37\% further confirm the stability of our proposed method. Finally, the robustness of our proposed model against adversarial perturbation attacks is tested as well, showing the inherent defensive properties of the ensemble approach.
Authors: Shuyong Gao, Yu'ang Feng, Qishan Wang, Lingyi Hong, Xinyu Zhou, Liu Fei, Yan Wang, Wenqiang Zhang
Abstract: Video Camouflaged Object Detection (VCOD) is a challenging task which aims to identify objects that seamlessly concealed within the background in videos. The dynamic properties of video enable detection of camouflaged objects through motion cues or varied perspectives. Previous VCOD datasets primarily contain animal objects, limiting the scope of research to wildlife scenarios. However, the applications of VCOD extend beyond wildlife and have significant implications in security, art, and medical fields. Addressing this problem, we construct a new large-scale multi-domain VCOD dataset MSVCOD. To achieve high-quality annotations, we design a semi-automatic iterative annotation pipeline that reduces costs while maintaining annotation accuracy. Our MSVCOD is the largest VCOD dataset to date, introducing multiple object categories including human, animal, medical, and vehicle objects for the first time, while also expanding background diversity across various environments. This expanded scope increases the practical applicability of the VCOD task in camouflaged object detection. Alongside this dataset, we introduce a one-steam video camouflage object detection model that performs both feature extraction and information fusion without additional motion feature fusion modules. Our framework achieves state-of-the-art results on the existing VCOD animal dataset and the proposed MSVCOD. The dataset and code will be made publicly available.
Authors: Hai Wang, Xiaoyu Xiang, Weihao Xia, Jing-Hao Xue
Abstract: The advent of text-driven 360-degree panorama generation, enabling the synthesis of 360-degree panoramic images directly from textual descriptions, marks a transformative advancement in immersive visual content creation. This innovation significantly simplifies the traditionally complex process of producing such content. Recent progress in text-to-image diffusion models has accelerated the rapid development in this emerging field. This survey presents a comprehensive review of text-driven 360-degree panorama generation, offering an in-depth analysis of state-of-the-art algorithms and their expanding applications in 360-degree 3D scene generation. Furthermore, we critically examine current limitations and propose promising directions for future research. A curated project page with relevant resources and research papers is available at https://littlewhitesea.github.io/Text-Driven-Pano-Gen/.
URLs: https://littlewhitesea.github.io/Text-Driven-Pano-Gen/.
Authors: Asifullah Khan, Laiba Asmatullah, Anza Malik, Shahzaib Khan, Hamna Asif
Abstract: Self-supervised learning is a machine learning approach that generates implicit labels by learning underlined patterns and extracting discriminative features from unlabeled data without manual labelling. Contrastive learning introduces the concept of "positive" and "negative" samples, where positive pairs (e.g., variation of the same image/object) are brought together in the embedding space, and negative pairs (e.g., views from different images/objects) are pushed farther away. This methodology has shown significant improvements in image understanding and image text analysis without much reliance on labeled data. In this paper, we comprehensively discuss the terminologies, recent developments and applications of contrastive learning with respect to text-image models. Specifically, we provide an overview of the approaches of contrastive learning in text-image models in recent years. Secondly, we categorize the approaches based on different model structures. Thirdly, we further introduce and discuss the latest advances of the techniques used in the process such as pretext tasks for both images and text, architectural structures, and key trends. Lastly, we discuss the recent state-of-art applications of self-supervised contrastive learning Text-Image based models.
Authors: Sara Sarto, Marcella Cornia, Rita Cucchiara
Abstract: The evaluation of machine-generated image captions is a complex and evolving challenge. With the advent of Multimodal Large Language Models (MLLMs), image captioning has become a core task, increasing the need for robust and reliable evaluation metrics. This survey provides a comprehensive overview of advancements in image captioning evaluation, analyzing the evolution, strengths, and limitations of existing metrics. We assess these metrics across multiple dimensions, including correlation with human judgment, ranking accuracy, and sensitivity to hallucinations. Additionally, we explore the challenges posed by the longer and more detailed captions generated by MLLMs and examine the adaptability of current metrics to these stylistic variations. Our analysis highlights some limitations of standard evaluation approaches and suggests promising directions for future research in image captioning assessment.
Authors: Ketan Suhaas Saichandran, Xavier Thomas, Prakhar Kaushik, Deepti Ghadiyaram
Abstract: Text-to-image generative models often struggle with long prompts detailing complex scenes, diverse objects with distinct visual characteristics and spatial relationships. In this work, we propose SCoPE (Scheduled interpolation of Coarse-to-fine Prompt Embeddings), a training-free method to improve text-to-image alignment by progressively refining the input prompt in a coarse-to-fine-grained manner. Given a detailed input prompt, we first decompose it into multiple sub-prompts which evolve from describing broad scene layout to highly intricate details. During inference, we interpolate between these sub-prompts and thus progressively introduce finer-grained details into the generated image. Our training-free plug-and-play approach significantly enhances prompt alignment, achieves an average improvement of more than +8 in Visual Question Answering (VQA) scores over the Stable Diffusion baselines on 83% of the prompts from the GenAI-Bench dataset.
Authors: Qiao Liang, Yanjiang Liu, Weixiang Zhou, Ben He, Yaojie Lu, Hongyu Lin, Jia Zheng, Xianpei Han, Le Sun, Yingfei Sun
Abstract: Does the prior knowledge of the vision encoder constrain the capability boundary of Multi-modal Large Language Models (MLLMs)? While most existing research treats MLLMs as unified systems optimized through end-to-end training, the impact of vision encoder's prior knowledge is seldom investigated. In this work, we introduce a novel metric, $Rank_e$, to quantify the effect of prior knowledge of the vision encoder on MLLM performance. Our analysis reveals a positive correlation between prior knowledge and MLLM performance. Moreover, we find that domain-specific fine-tuning using solely end-to-end visual question answering (VQA) data is insufficient, particularly for entities with low inherent visual prior knowledge. To address this issue, we propose VisPRE (Vision Prior Remediation), a two-stage training framework that explicitly incorporates prior knowledge at the vision encoder level. Experimental results demonstrate that augmenting vision encoder's prior knowledge substantially boosts the visual understanding capabilities of MLLMs, offering a novel and effective strategy for improving performance, especially in scenarios involving uncommon visual entities.
Authors: Konstantin Pakulev, Alexander Vakhitov, Gonzalo Ferrer
Abstract: Local features are essential to many modern downstream applications. Therefore, it is of interest to determine the properties of local features that contribute to the downstream performance for a better design of feature detectors and descriptors. In our work, we propose a new theoretical model for scoring feature points (keypoints) in the context of the two-view geometry estimation problem. The model determines two properties that a good keypoint for solving the homography estimation problem should have: be repeatable and have a small expected measurement error. This result provides key insights into why maximizing the number of correspondences doesn't always lead to better homography estimation accuracy. We use the developed model to design a method that detects keypoints that benefit the homography estimation and introduce the Bounded NeSS-ST (BoNeSS-ST) keypoint detector. The novelty of BoNeSS-ST comes from strong theoretical foundations, a more accurate keypoint scoring due to subpixel refinement and a cost designed for superior robustness to low saliency keypoints. As a result, BoNeSS-ST outperforms prior self-supervised local feature detectors on the planar homography estimation task and is on par with them on the epipolar geometry estimation task.
Authors: Masoumeh Sharafi, Emma Ollivier, Muhammad Osama Zeeshan, Soufiane Belharbi, Marco Pedersoli, Alessandro Lameiras Koerich, Simon Bacon, Eric Granger
Abstract: Facial Expression Recognition (FER) from videos is a crucial task in various application areas, such as human-computer interaction and health diagnosis and monitoring (e.g., assessing pain and depression). Beyond the challenges of recognizing subtle emotional or health states, the effectiveness of deep FER models is often hindered by the considerable inter-subject variability in expressions. Source-free (unsupervised) domain adaptation (SFDA) methods may be employed to adapt a pre-trained source model using only unlabeled target domain data, thereby avoiding data privacy, storage, and transmission issues. Typically, SFDA methods adapt to a target domain dataset corresponding to an entire population and assume it includes data from all recognition classes. However, collecting such comprehensive target data can be difficult or even impossible for FER in healthcare applications. In many real-world scenarios, it may be feasible to collect a short neutral control video (which displays only neutral expressions) from target subjects before deployment. These videos can be used to adapt a model to better handle the variability of expressions among subjects. This paper introduces the Disentangled SFDA (DSFDA) method to address the challenge posed by adapting models with missing target expression data. DSFDA leverages data from a neutral target control video for end-to-end generation and adaptation of target data with missing non-neutral data. Our method learns to disentangle features related to expressions and identity while generating the missing non-neutral expression data for the target subject, thereby enhancing model accuracy. Additionally, our self-supervision strategy improves model adaptation by reconstructing target images that maintain the same identity and source expression.
Authors: Hasan Moughnieh, Mohamad Chalhoub, Hasan Nasrallah, Cristiano Nattero, Paolo Campanella, Giovanni Nico, Ali J. Ghandour
Abstract: Adapting pre-trained models has become an effective strategy in artificial intelligence, offering a scalable and efficient alternative to training models from scratch. In the context of remote sensing (RS), where visual grounding(VG) remains underexplored, this approach enables the deployment of powerful vision-language models to achieve robust cross-modal understanding while significantly reducing computational overhead. To address this, we applied Parameter Efficient Fine Tuning (PEFT) techniques to adapt these models for RS-specific VG tasks. Specifically, we evaluated LoRA placement across different modules in Grounding DINO and used BitFit and adapters to fine-tune the OFA foundation model pre-trained on general-purpose VG datasets. This approach achieved performance comparable to or surpassing current State Of The Art (SOTA) models while significantly reducing computational costs. This study highlights the potential of PEFT techniques to advance efficient and precise multi-modal analysis in RS, offering a practical and cost-effective alternative to full model training.
Authors: Junhao Cheng, Yuying Ge, Yixiao Ge, Jing Liao, Ying Shan
Abstract: Recent advancements in image and video synthesis have opened up new promise in generative games. One particularly intriguing application is transforming characters from anime films into interactive, playable entities. This allows players to immerse themselves in the dynamic anime world as their favorite characters for life simulation through language instructions. Such games are defined as infinite game since they eliminate predetermined boundaries and fixed gameplay rules, where players can interact with the game world through open-ended language and experience ever-evolving storylines and environments. Recently, a pioneering approach for infinite anime life simulation employs large language models (LLMs) to translate multi-turn text dialogues into language instructions for image generation. However, it neglects historical visual context, leading to inconsistent gameplay. Furthermore, it only generates static images, failing to incorporate the dynamics necessary for an engaging gaming experience. In this work, we propose AnimeGamer, which is built upon Multimodal Large Language Models (MLLMs) to generate each game state, including dynamic animation shots that depict character movements and updates to character states, as illustrated in Figure 1. We introduce novel action-aware multimodal representations to represent animation shots, which can be decoded into high-quality video clips using a video diffusion model. By taking historical animation shot representations as context and predicting subsequent representations, AnimeGamer can generate games with contextual consistency and satisfactory dynamics. Extensive evaluations using both automated metrics and human evaluations demonstrate that AnimeGamer outperforms existing methods in various aspects of the gaming experience. Codes and checkpoints are available at https://github.com/TencentARC/AnimeGamer.
Authors: Shabnam Choudhury, Yash Salunkhe, Sarthak Mehrotra, Biplab Banerjee
Abstract: The rapid expansion of remote sensing image archives demands the development of strong and efficient techniques for content-based image retrieval (RS-CBIR). This paper presents REJEPA (Retrieval with Joint-Embedding Predictive Architecture), an innovative self-supervised framework designed for unimodal RS-CBIR. REJEPA utilises spatially distributed context token encoding to forecast abstract representations of target tokens, effectively capturing high-level semantic features and eliminating unnecessary pixel-level details. In contrast to generative methods that focus on pixel reconstruction or contrastive techniques that depend on negative pairs, REJEPA functions within feature space, achieving a reduction in computational complexity of 40-60% when compared to pixel-reconstruction baselines like Masked Autoencoders (MAE). To guarantee strong and varied representations, REJEPA incorporates Variance-Invariance-Covariance Regularisation (VICReg), which prevents encoder collapse by promoting feature diversity and reducing redundancy. The method demonstrates an estimated enhancement in retrieval accuracy of 5.1% on BEN-14K (S1), 7.4% on BEN-14K (S2), 6.0% on FMoW-RGB, and 10.1% on FMoW-Sentinel compared to prominent SSL techniques, including CSMAE-SESD, Mask-VLM, SatMAE, ScaleMAE, and SatMAE++, on extensive RS benchmarks BEN-14K (multispectral and SAR data), FMoW-RGB and FMoW-Sentinel. Through effective generalisation across sensor modalities, REJEPA establishes itself as a sensor-agnostic benchmark for efficient, scalable, and precise RS-CBIR, addressing challenges like varying resolutions, high object density, and complex backgrounds with computational efficiency.
Authors: Xiyao Wang, Zhengyuan Yang, Chao Feng, Hongjin Lu, Linjie Li, Chung-Ching Lin, Kevin Lin, Furong Huang, Lijuan Wang
Abstract: We introduce ThinkLite-VL, a family of visual reasoning models that achieve state-of-the-art (SoTA) performance using an order of magnitude fewer training samples, relying purely on reinforcement fine-tuning (RFT) self-improvement without any knowledge distillation. Our central insight is that sample difficulty critically influences RFT effectiveness: appropriately challenging examples can drive substantial reasoning improvements, even in low-data regimes. However, quantifying sample difficulty in a reliable and scalable manner remains non-trivial. To address this, we repurpose Monte Carlo Tree Search (MCTS) to measure sample difficulty via the number of reasoning iterations a vision-language model (VLM) requires to solve each instance. This MCTS-based selection procedure identifies samples that induce deeper reasoning while remaining solvable, allowing us to filter a high-quality subset from 70k open-source examples spanning math, natural image understanding, and chart comprehension. Using this approach, we select just 11k challenging samples for RFT on Qwen2.5-VL-7B-Instruct and 7.5k samples for Qwen2.5-VL-72B-Instruct. The resulting models, ThinkLite-VL-7B and ThinkLite-VL-72B, significantly outperform their respective base models across eight visual reasoning benchmarks. In particular, ThinkLite-VL-7B improves the average performance of Qwen2.5-VL-7B-Instruct by 7\% and surpasses all existing 7B-level models, as well as much larger models such as GPT-4o, O1 and Qwen2.5-VL-72B, achieving a new SoTA score of 75.1 on MathVista. ThinkLite-VL-72B further advances the SoTA frontier, achieving an accuracy of 79.7 on MathVista and an average benchmark improvement of 4.42 over the open-source SOTA. These results demonstrate that MCTS-guided difficulty filtering provides a scalable and effective path toward data-efficient self-improvement in multimodal reasoning.
Authors: Hasan Wehbi, Hasan Nasrallah, Mohamad Hasan Zahweh, Zeinab Takach, Veera Ganesh Yalla, Ali J. Ghandour
Abstract: Wheat accounts for approximately 20% of the world's caloric intake, making it a vital component of global food security. Given this importance, mapping wheat fields plays a crucial role in enabling various stakeholders, including policy makers, researchers, and agricultural organizations, to make informed decisions regarding food security, supply chain management, and resource allocation. In this paper, we tackle the problem of accurately mapping wheat fields out of satellite images by introducing an improved pipeline for winter wheat segmentation, as well as presenting a case study on a decade-long analysis of wheat mapping in Lebanon. We integrate a Temporal Spatial Vision Transformer (TSViT) with Parameter-Efficient Fine Tuning (PEFT) and a novel post-processing pipeline based on the Fields of The World (FTW) framework. Our proposed pipeline addresses key challenges encountered in existing approaches, such as the clustering of small agricultural parcels in a single large field. By merging wheat segmentation with precise field boundary extraction, our method produces geometrically coherent and semantically rich maps that enable us to perform in-depth analysis such as tracking crop rotation pattern over years. Extensive evaluations demonstrate improved boundary delineation and field-level precision, establishing the potential of the proposed framework in operational agricultural monitoring and historical trend analysis. By allowing for accurate mapping of wheat fields, this work lays the foundation for a range of critical studies and future advances, including crop monitoring and yield estimation.
Authors: Wangyu Wu, Zhenhong Chen, Siqi Song, Xianglin Qiu, Xiaowei Huang, Fei Ma, Jimin Xiao
Abstract: Cross-Domain Sequential Recommendation (CDSR) predicts user behavior by leveraging historical interactions across multiple domains, focusing on modeling cross-domain preferences through intra- and inter-sequence item relationships. Inspired by human cognitive processes, we propose Hierarchical Attention Fusion of Visual and Textual Representations (HAF-VT), a novel approach integrating visual and textual data to enhance cognitive modeling. Using the frozen CLIP model, we generate image and text embeddings, enriching item representations with multimodal data. A hierarchical attention mechanism jointly learns single-domain and cross-domain preferences, mimicking human information integration. Evaluated on four e-commerce datasets, HAF-VT outperforms existing methods in capturing cross-domain user interests, bridging cognitive principles with computational models and highlighting the role of multimodal data in sequential decision-making.
Authors: Banafsheh Karimian, Giulia Avanzato, Soufian Belharbi, Luke McCaffrey, Mohammadhadi Shateri, Eric Granger
Abstract: Multimodal learning has shown significant promise for improving medical image analysis by integrating information from complementary data sources. This is widely employed for training vision-language models (VLMs) for cancer detection based on histology images and text reports. However, one of the main limitations in training these VLMs is the requirement for large paired datasets, raising concerns over privacy, and data collection, annotation, and maintenance costs. To address this challenge, we introduce CLIP-IT method to train a vision backbone model to classify histology images by pairing them with privileged textual information from an external source. At first, the modality pairing step relies on a CLIP-based model to match histology images with semantically relevant textual report data from external sources, creating an augmented multimodal dataset without the need for manually paired samples. Then, we propose a multimodal training procedure that distills the knowledge from the paired text modality to the unimodal image classifier for enhanced performance without the need for the textual data during inference. A parameter-efficient fine-tuning method is used to efficiently address the misalignment between the main (image) and paired (text) modalities. During inference, the improved unimodal histology classifier is used, with only minimal additional computational complexity. Our experiments on challenging PCAM, CRC, and BACH histology image datasets show that CLIP-IT can provide a cost-effective approach to leverage privileged textual information and outperform unimodal classifiers for histology.
Authors: Chengzhi Wu, Yuxin Wan, Hao Fu, Julius Pfrommer, Zeyun Zhong, Junwei Zheng, Jiaming Zhang, J\"urgen Beyerer
Abstract: Driven by the increasing demand for accurate and efficient representation of 3D data in various domains, point cloud sampling has emerged as a pivotal research topic in 3D computer vision. Recently, learning-to-sample methods have garnered growing interest from the community, particularly for their ability to be jointly trained with downstream tasks. However, previous learning-based sampling methods either lead to unrecognizable sampling patterns by generating a new point cloud or biased sampled results by focusing excessively on sharp edge details. Moreover, they all overlook the natural variations in point distribution across different shapes, applying a similar sampling strategy to all point clouds. In this paper, we propose a Sparse Attention Map and Bin-based Learning method (termed SAMBLE) to learn shape-specific sampling strategies for point cloud shapes. SAMBLE effectively achieves an improved balance between sampling edge points for local details and preserving uniformity in the global shape, resulting in superior performance across multiple common point cloud downstream tasks, even in scenarios with few-point sampling.
Authors: Simon Ging, Sebastian Walter, Jelena Bratuli\'c, Johannes Dienert, Hannah Bast, Thomas Brox
Abstract: Training high-quality CLIP models typically requires enormous datasets, which limits the development of domain-specific models -- especially in areas that even the largest CLIP models do not cover well -- and drives up training costs. This poses challenges for scientific research that needs fine-grained control over the training procedure of CLIP models. In this work, we show that by employing smart web search strategies enhanced with knowledge graphs, a robust CLIP model can be trained from scratch with considerably less data. Specifically, we demonstrate that an expert foundation model for living organisms can be built using just 10M images. Moreover, we introduce EntityNet, a dataset comprising 33M images paired with 46M text descriptions, which enables the training of a generic CLIP model in significantly reduced time.
Authors: Zhihao Zhang, Abhinav Kumar, Girish Chandar Ganesan, Xiaoming Liu
Abstract: Accurately predicting 3D attributes is crucial for monocular 3D object detection (Mono3D), with depth estimation posing the greatest challenge due to the inherent ambiguity in mapping 2D images to 3D space. While existing methods leverage multiple depth cues (e.g., estimating depth uncertainty, modeling depth error) to improve depth accuracy, they overlook that accurate depth prediction requires conditioning on other 3D attributes, as these attributes are intrinsically inter-correlated through the 3D to 2D projection, which ultimately limits overall accuracy and stability. Inspired by Chain-of-Thought (CoT) in large language models (LLMs), this paper proposes MonoCoP, which leverages a Chain-of-Prediction (CoP) to predict attributes sequentially and conditionally via three key designs. First, it employs a lightweight AttributeNet (AN) for each 3D attribute to learn attribute-specific features. Next, MonoCoP constructs an explicit chain to propagate these learned features from one attribute to the next. Finally, MonoCoP uses a residual connection to aggregate features for each attribute along the chain, ensuring that later attribute predictions are conditioned on all previously processed attributes without forgetting the features of earlier ones. Experimental results show that our MonoCoP achieves state-of-the-art (SoTA) performance on the KITTI leaderboard without requiring additional data and further surpasses existing methods on the Waymo and nuScenes frontal datasets.
Authors: Hanxun Huang, Sarah Erfani, Yige Li, Xingjun Ma, James Bailey
Abstract: As Contrastive Language-Image Pre-training (CLIP) models are increasingly adopted for diverse downstream tasks and integrated into large vision-language models (VLMs), their susceptibility to adversarial perturbations has emerged as a critical concern. In this work, we introduce \textbf{X-Transfer}, a novel attack method that exposes a universal adversarial vulnerability in CLIP. X-Transfer generates a Universal Adversarial Perturbation (UAP) capable of deceiving various CLIP encoders and downstream VLMs across different samples, tasks, and domains. We refer to this property as \textbf{super transferability}--a single perturbation achieving cross-data, cross-domain, cross-model, and cross-task adversarial transferability simultaneously. This is achieved through \textbf{surrogate scaling}, a key innovation of our approach. Unlike existing methods that rely on fixed surrogate models, which are computationally intensive to scale, X-Transfer employs an efficient surrogate scaling strategy that dynamically selects a small subset of suitable surrogates from a large search space. Extensive evaluations demonstrate that X-Transfer significantly outperforms previous state-of-the-art UAP methods, establishing a new benchmark for adversarial transferability across CLIP models. The code is publicly available in our \href{https://github.com/HanxunH/XTransferBench}{GitHub repository}.
Authors: Yanbo Ding, Xirui Hu, Zhizhi Guo, Chi Zhang, Yali Wang
Abstract: Human image animation has gained increasing attention and developed rapidly due to its broad applications in digital humans. However, existing methods rely largely on 2D-rendered pose images for motion guidance, which limits generalization and discards essential 3D information for open-world animation. To tackle this problem, we propose MTVCrafter (Motion Tokenization Video Crafter), the first framework that directly models raw 3D motion sequences (i.e., 4D motion) for human image animation. Specifically, we introduce 4DMoT (4D motion tokenizer) to quantize 3D motion sequences into 4D motion tokens. Compared to 2D-rendered pose images, 4D motion tokens offer more robust spatio-temporal cues and avoid strict pixel-level alignment between pose image and character, enabling more flexible and disentangled control. Then, we introduce MV-DiT (Motion-aware Video DiT). By designing unique motion attention with 4D positional encodings, MV-DiT can effectively leverage motion tokens as 4D compact yet expressive context for human image animation in the complex 3D world. Hence, it marks a significant step forward in this field and opens a new direction for pose-guided human video generation. Experiments show that our MTVCrafter achieves state-of-the-art results with an FID-VID of 6.98, surpassing the second-best by 65%. Powered by robust motion tokens, MTVCrafter also generalizes well to diverse open-world characters (single/multiple, full/half-body) across various styles and scenarios. Our video demos and code are on: https://github.com/DINGYANB/MTVCrafter.
Authors: Wenchuan Zhang, Penghao Zhang, Jingru Guo, Tao Cheng, Jie Chen, Shuwan Zhang, Zhang Zhang, Yuhao Yi, Hong Bu
Abstract: Recent advances in vision language models (VLMs) have enabled broad progress in the general medical field. However, pathology still remains a more challenging subdomain, with current pathology specific VLMs exhibiting limitations in both diagnostic accuracy and reasoning plausibility. Such shortcomings are largely attributable to the nature of current pathology datasets, which are primarily composed of image description pairs that lack the depth and structured diagnostic paradigms employed by real world pathologists. In this study, we leverage pathology textbooks and real world pathology experts to construct high-quality, reasoning-oriented datasets. Building on this, we introduce Patho-R1, a multimodal RL-based pathology Reasoner, trained through a three-stage pipeline: (1) continued pretraining on 3.5 million image-text pairs for knowledge infusion; (2) supervised fine-tuning on 500k high-quality Chain-of-Thought samples for reasoning incentivizing; (3) reinforcement learning using Group Relative Policy Optimization and Decoupled Clip and Dynamic sAmpling Policy Optimization strategies for multimodal reasoning quality refinement. To further assess the alignment quality of our dataset, we propose Patho-CLIP, trained on the same figure-caption corpus used for continued pretraining. Comprehensive experimental results demonstrate that both Patho-CLIP and Patho-R1 achieve robust performance across a wide range of pathology-related tasks, including zero-shot classification, cross-modal retrieval, Visual Question Answering, and Multiple Choice Question. Our project is available at the Patho-R1 repository: https://github.com/Wenchuan-Zhang/Patho-R1.
Authors: Chaoya Jiang, Yongrui Heng, Wei Ye, Han Yang, Haiyang Xu, Ming Yan, Ji Zhang, Fei Huang, Shikun Zhang
Abstract: Recently, reasoning-based MLLMs have achieved a degree of success in generating long-form textual reasoning chains. However, they still struggle with complex tasks that necessitate dynamic and iterative focusing on and revisiting of visual regions to achieve precise grounding of textual reasoning in visual evidence. We introduce \textbf{VLM-R$^3$} (\textbf{V}isual \textbf{L}anguage \textbf{M}odel with \textbf{R}egion \textbf{R}ecognition and \textbf{R}easoning), a framework that equips an MLLM with the ability to (i) decide \emph{when} additional visual evidence is needed, (ii) determine \emph{where} to ground within the image, and (iii) seamlessly weave the relevant sub-image content back into an interleaved chain-of-thought. The core of our method is \textbf{Region-Conditioned Reinforcement Policy Optimization (R-GRPO)}, a training paradigm that rewards the model for selecting informative regions, formulating appropriate transformations (e.g.\ crop, zoom), and integrating the resulting visual context into subsequent reasoning steps. To bootstrap this policy, we compile a modest but carefully curated Visuo-Lingual Interleaved Rationale (VLIR) corpus that provides step-level supervision on region selection and textual justification. Extensive experiments on MathVista, ScienceQA, and other benchmarks show that VLM-R$^3$ sets a new state of the art in zero-shot and few-shot settings, with the largest gains appearing on questions demanding subtle spatial reasoning or fine-grained visual cue extraction.
Authors: Xiaoyu Ye, Songjie Cheng, Yongtao Wang, Yajiao Xiong, Yishen Li
Abstract: Recent advances in text-to-video (T2V) diffusion models have significantly enhanced the quality of generated videos. However, their ability to produce explicit or harmful content raises concerns about misuse and potential rights violations. Inspired by the success of unlearning techniques in erasing undesirable concepts from text-to-image (T2I) models, we extend unlearning to T2V models and propose a robust and precise unlearning method. Specifically, we adopt negatively-guided velocity prediction fine-tuning and enhance it with prompt augmentation to ensure robustness against LLM-refined prompts. To achieve precise unlearning, we incorporate a localization and a preservation regularization to preserve the model's ability to generate non-target concepts. Extensive experiments demonstrate that our method effectively erases a specific concept while preserving the model's generation capability for all other concepts, outperforming existing methods. We provide the unlearned models in \href{https://github.com/VDIGPKU/T2VUnlearning.git}{https://github.com/VDIGPKU/T2VUnlearning.git}.
URLs: https://github.com/VDIGPKU/T2VUnlearning.git, https://github.com/VDIGPKU/T2VUnlearning.git
Authors: Ming Hu, Zhengdi Yu, Feilong Tang, Kaiwen Chen, Yulong Li, Imran Razzak, Junjun He, Tolga Birdal, Kaijing Zhou, Zongyuan Ge
Abstract: Accurate 3D reconstruction of hands and instruments is critical for vision-based analysis of ophthalmic microsurgery, yet progress has been hampered by the lack of realistic, large-scale datasets and reliable annotation tools. In this work, we introduce OphNet-3D, the first extensive RGB-D dynamic 3D reconstruction dataset for ophthalmic surgery, comprising 41 sequences from 40 surgeons and totaling 7.1 million frames, with fine-grained annotations of 12 surgical phases, 10 instrument categories, dense MANO hand meshes, and full 6-DoF instrument poses. To scalably produce high-fidelity labels, we design a multi-stage automatic annotation pipeline that integrates multi-view data observation, data-driven motion prior with cross-view geometric consistency and biomechanical constraints, along with a combination of collision-aware interaction constraints for instrument interactions. Building upon OphNet-3D, we establish two challenging benchmarks-bimanual hand pose estimation and hand-instrument interaction reconstruction-and propose two dedicated architectures: H-Net for dual-hand mesh recovery and OH-Net for joint reconstruction of two-hand-two-instrument interactions. These models leverage a novel spatial reasoning module with weak-perspective camera modeling and collision-aware center-based representation. Both architectures outperform existing methods by substantial margins, achieving improvements of over 2mm in Mean Per Joint Position Error (MPJPE) and up to 23% in ADD-S metrics for hand and instrument reconstruction, respectively.
Authors: Anjie Le, Henan Liu, Yue Wang, Zhenyu Liu, Rongkun Zhu, Taohan Weng, Jinze Yu, Boyang Wang, Yalun Wu, Kaiwen Yan, Quanlin Sun, Meirui Jiang, Jialun Pei, Siya Liu, Haoyun Zheng, Zhoujun Li, Alison Noble, Jacques Souquet, Xiaoqing Guo, Manxi Lin, Hongcheng Guo
Abstract: Ultrasound is a widely-used imaging modality critical to global healthcare, yet its interpretation remains challenging due to its varying image quality on operators, noises, and anatomical structures. Although large vision-language models (LVLMs) have demonstrated impressive multimodal capabilities across natural and medical domains, their performance on ultrasound remains largely unexplored. We introduce U2-BENCH, the first comprehensive benchmark to evaluate LVLMs on ultrasound understanding across classification, detection, regression, and text generation tasks. U2-BENCH aggregates 7,241 cases spanning 15 anatomical regions and defines 8 clinically inspired tasks, such as diagnosis, view recognition, lesion localization, clinical value estimation, and report generation, across 50 ultrasound application scenarios. We evaluate 20 state-of-the-art LVLMs, both open- and closed-source, general-purpose and medical-specific. Our results reveal strong performance on image-level classification, but persistent challenges in spatial reasoning and clinical language generation. U2-BENCH establishes a rigorous and unified testbed to assess and accelerate LVLM research in the uniquely multimodal domain of medical ultrasound imaging.
Authors: Jiaxin Song, Yixu Wang, Jie Li, Rui Yu, Yan Teng, Xingjun Ma, Yingchun Wang
Abstract: Vision-Language Models (VLMs) exhibit impressive performance, yet the integration of powerful vision encoders has significantly broadened their attack surface, rendering them increasingly susceptible to jailbreak attacks. However, lacking well-defined attack objectives, existing jailbreak methods often struggle with gradient-based strategies prone to local optima and lacking precise directional guidance, and typically decouple visual and textual modalities, thereby limiting their effectiveness by neglecting crucial cross-modal interactions. Inspired by the Eliciting Latent Knowledge (ELK) framework, we posit that VLMs encode safety-relevant information within their internal fusion-layer representations, revealing an implicit safety decision boundary in the latent space. This motivates exploiting boundary to steer model behavior. Accordingly, we propose JailBound, a novel latent space jailbreak framework comprising two stages: (1) Safety Boundary Probing, which addresses the guidance issue by approximating decision boundary within fusion layer's latent space, thereby identifying optimal perturbation directions towards the target region; and (2) Safety Boundary Crossing, which overcomes the limitations of decoupled approaches by jointly optimizing adversarial perturbations across both image and text inputs. This latter stage employs an innovative mechanism to steer the model's internal state towards policy-violating outputs while maintaining cross-modal semantic consistency. Extensive experiments on six diverse VLMs demonstrate JailBound's efficacy, achieves 94.32% white-box and 67.28% black-box attack success averagely, which are 6.17% and 21.13% higher than SOTA methods, respectively. Our findings expose a overlooked safety risk in VLMs and highlight the urgent need for more robust defenses. Warning: This paper contains potentially sensitive, harmful and offensive content.
Authors: Shintaro Ito, Natsuki Takama, Koichi Ito, Hwann-Tzong Chen, Takafumi Aoki
Abstract: The use of multi-view images acquired by a 360-degree camera can reconstruct a 3D space with a wide area. There are 3D reconstruction methods from equirectangular images based on NeRF and 3DGS, as well as Novel View Synthesis (NVS) methods. On the other hand, it is necessary to overcome the large distortion caused by the projection model of a 360-degree camera when equirectangular images are used. In 3DGS-based methods, the large distortion of the 360-degree camera model generates extremely large 3D Gaussians, resulting in poor rendering accuracy. We propose ErpGS, which is Omnidirectional GS based on 3DGS to realize NVS addressing the problems. ErpGS introduce some rendering accuracy improvement techniques: geometric regularization, scale regularization, and distortion-aware weights and a mask to suppress the effects of obstacles in equirectangular images. Through experiments on public datasets, we demonstrate that ErpGS can render novel view images more accurately than conventional methods.
Authors: Zitong Wang, Hang Zhao, Qianyu Zhou, Xuequan Lu, Xiangtai Li, Yiren Song
Abstract: Diffusion models have recently motivated great success in many generation tasks like object removal. Nevertheless, existing image decomposition methods struggle to disentangle semi-transparent or transparent layer occlusions due to mask prior dependencies, static object assumptions, and the lack of datasets. In this paper, we delve into a novel task: Layer-Wise Decomposition of Alpha-Composited Images, aiming to recover constituent layers from single overlapped images under the condition of semi-transparent/transparent alpha layer non-linear occlusion. To address challenges in layer ambiguity, generalization, and data scarcity, we first introduce AlphaBlend, the first large-scale and high-quality dataset for transparent and semi-transparent layer decomposition, supporting six real-world subtasks (e.g., translucent flare removal, semi-transparent cell decomposition, glassware decomposition). Building on this dataset, we present DiffDecompose, a diffusion Transformer-based framework that learns the posterior over possible layer decompositions conditioned on the input image, semantic prompts, and blending type. Rather than regressing alpha mattes directly, DiffDecompose performs In-Context Decomposition, enabling the model to predict one or multiple layers without per-layer supervision, and introduces Layer Position Encoding Cloning to maintain pixel-level correspondence across layers. Extensive experiments on the proposed AlphaBlend dataset and public LOGO dataset verify the effectiveness of DiffDecompose. The code and dataset will be available upon paper acceptance. Our code will be available at: https://github.com/Wangzt1121/DiffDecompose.
Authors: Keanu Nichols, Nazia Tasnim, Yuting Yan, Nicholas Ikechukwu, Elva Zou, Deepti Ghadiyaram, Bryan A. Plummer
Abstract: Object orientation understanding represents a fundamental challenge in visual perception critical for applications like robotic manipulation and augmented reality. Current vision-language benchmarks fail to isolate this capability, often conflating it with positional relationships and general scene understanding. We introduce DORI (Discriminative Orientation Reasoning Intelligence), a comprehensive benchmark establishing object orientation perception as a primary evaluation target. DORI assesses four dimensions of orientation comprehension: frontal alignment, rotational transformations, relative directional relationships, and canonical orientation understanding. Through carefully curated tasks from 11 datasets spanning 67 object categories across synthetic and real-world scenarios, DORI provides insights on how multi-modal systems understand object orientations. Our evaluation of 15 state-of-the-art vision-language models reveals critical limitations: even the best models achieve only 54.2% accuracy on coarse tasks and 33.0% on granular orientation judgments, with performance deteriorating for tasks requiring reference frame shifts or compound rotations. These findings demonstrate the need for dedicated orientation representation mechanisms, as models show systematic inability to perform precise angular estimations, track orientation changes across viewpoints, and understand compound rotations - suggesting limitations in their internal 3D spatial representations. As the first diagnostic framework specifically designed for orientation awareness in multimodal systems, DORI offers implications for improving robotic control, 3D scene reconstruction, and human-AI interaction in physical environments. DORI data: https://huggingface.co/datasets/appledora/DORI-Benchmark
URLs: https://huggingface.co/datasets/appledora/DORI-Benchmark
Authors: Yanbei Jiang, Yihao Ding, Chao Lei, Jiayang Ao, Jey Han Lau, Krista A. Ehinger
Abstract: Current Multimodal Large Language Models (MLLMs) excel in general visual reasoning but remain underexplored in Abstract Visual Reasoning (AVR), which demands higher-order reasoning to identify abstract rules beyond simple perception. Existing AVR benchmarks focus on single-step reasoning, emphasizing the end result but neglecting the multi-stage nature of reasoning process. Past studies found MLLMs struggle with these benchmarks, but it doesn't explain how they fail. To address this gap, we introduce MultiStAR, a Multi-Stage AVR benchmark, based on RAVEN, designed to assess reasoning across varying levels of complexity. Additionally, existing metrics like accuracy only focus on the final outcomes while do not account for the correctness of intermediate steps. Therefore, we propose a novel metric, MSEval, which considers the correctness of intermediate steps in addition to the final outcomes. We conduct comprehensive experiments on MultiStAR using 17 representative close-source and open-source MLLMs. The results reveal that while existing MLLMs perform adequately on basic perception tasks, they continue to face challenges in more complex rule detection stages.
Authors: Woonho Ko, Jin Bok Park, Il Yong Chun
Abstract: Existing long-term video prediction methods often rely on an autoregressive video prediction mechanism. However, this approach suffers from error propagation, particularly in distant future frames. To address this limitation, this paper proposes the first AutoRegression-Free (ARFree) video prediction framework using diffusion models. Different from an autoregressive video prediction mechanism, ARFree directly predicts any future frame tuples from the context frame tuple. The proposed ARFree consists of two key components: 1) a motion prediction module that predicts a future motion using motion feature extracted from the context frame tuple; 2) a training method that improves motion continuity and contextual consistency between adjacent future frame tuples. Our experiments with two benchmark datasets show that the proposed ARFree video prediction framework outperforms several state-of-the-art video prediction methods.
Authors: Zobia Batool, Huseyin Ozkan, Erchan Aptoula
Abstract: Although Alzheimer's disease detection via MRIs has advanced significantly thanks to contemporary deep learning models, challenges such as class imbalance, protocol variations, and limited dataset diversity often hinder their generalization capacity. To address this issue, this article focuses on the single domain generalization setting, where given the data of one domain, a model is designed and developed with maximal performance w.r.t. an unseen domain of distinct distribution. Since brain morphology is known to play a crucial role in Alzheimer's diagnosis, we propose the use of learnable pseudo-morphological modules aimed at producing shape-aware, anatomically meaningful class-specific augmentations in combination with a supervised contrastive learning module to extract robust class-specific representations. Experiments conducted across three datasets show improved performance and generalization capacity, especially under class imbalance and imaging protocol variations. The source code will be made available upon acceptance at https://github.com/zobia111/SDG-Alzheimer.
Authors: Ruixuan Zhang, He Wang, Zhengyu Zhao, Zhiqing Guo, Xun Yang, Yunfeng Diao, Meng Wang
Abstract: Rapid advances in Artificial Intelligence Generated Images (AIGI) have facilitated malicious use, such as forgery and misinformation. Therefore, numerous methods have been proposed to detect fake images. Although such detectors have been proven to be universally vulnerable to adversarial attacks, defenses in this field are scarce. In this paper, we first identify that adversarial training (AT), widely regarded as the most effective defense, suffers from performance collapse in AIGI detection. Through an information-theoretic lens, we further attribute the cause of collapse to feature entanglement, which disrupts the preservation of feature-label mutual information. Instead, standard detectors show clear feature separation. Motivated by this difference, we propose Training-free Robust Detection via Information-theoretic Measures (TRIM), the first training-free adversarial defense for AIGI detection. TRIM builds on standard detectors and quantifies feature shifts using prediction entropy and KL divergence. Extensive experiments across multiple datasets and attacks validate the superiority of our TRIM, e.g., outperforming the state-of-the-art defense by 33.88% (28.91%) on ProGAN (GenImage), while well maintaining original accuracy.
Authors: Zhangyi Hu, Jiemin Wu, Hua Xu, Mingqian Liao, Ninghui Feng, Bo Gao, Songning Lai, Yutao Yue
Abstract: Irregular Multivariate Time Series (IMTS) forecasting is challenging due to the unaligned nature of multi-channel signals and the prevalence of extensive missing data. Existing methods struggle to capture reliable temporal patterns from such data due to significant missing values. While pre-trained foundation models show potential for addressing these challenges, they are typically designed for Regularly Sampled Time Series (RTS). Motivated by the visual Mask AutoEncoder's (MAE) powerful capability for modeling sparse multi-channel information and its success in RTS forecasting, we propose VIMTS, a framework adapting Visual MAE for IMTS forecasting. To mitigate the effect of missing values, VIMTS first processes IMTS along the timeline into feature patches at equal intervals. These patches are then complemented using learned cross-channel dependencies. Then it leverages visual MAE's capability in handling sparse multichannel data for patch reconstruction, followed by a coarse-to-fine technique to generate precise predictions from focused contexts. In addition, we integrate self-supervised learning for improved IMTS modeling by adapting the visual MAE to IMTS data. Extensive experiments demonstrate VIMTS's superior performance and few-shot capability, advancing the application of visual foundation models in more general time series tasks. Our code is available at https://github.com/WHU-HZY/VIMTS.
Authors: Ruichen Chen, Keith G. Mills, Liyao Jiang, Chao Gao, Di Niu
Abstract: Diffusion Transformers (DiT) have become the de-facto model for generating high-quality visual content like videos and images. A huge bottleneck is the attention mechanism where complexity scales quadratically with resolution and video length. One logical way to lessen this burden is sparse attention, where only a subset of tokens or patches are included in the calculation. However, existing techniques fail to preserve visual quality at extremely high sparsity levels and might even incur non-negligible compute overheads. % To address this concern, we propose Re-ttention, which implements very high sparse attention for visual generation models by leveraging the temporal redundancy of Diffusion Models to overcome the probabilistic normalization shift within the attention mechanism. Specifically, Re-ttention reshapes attention scores based on the prior softmax distribution history in order to preserve the visual quality of the full quadratic attention at very high sparsity levels. % Experimental results on T2V/T2I models such as CogVideoX and the PixArt DiTs demonstrate that Re-ttention requires as few as 3.1\% of the tokens during inference, outperforming contemporary methods like FastDiTAttn, Sparse VideoGen and MInference. Further, we measure latency to show that our method can attain over 45\% end-to-end % and over 92\% self-attention latency reduction on an H100 GPU at negligible overhead cost. Code available online here: \href{https://github.com/cccrrrccc/Re-ttention}{https://github.com/cccrrrccc/Re-ttention}
URLs: https://github.com/cccrrrccc/Re-ttention, https://github.com/cccrrrccc/Re-ttention
Authors: Hao Wu, Junzhou Chen, Ronghui Zhang, Nengchao Lyu, Hongyu Hu, Yanyong Guo, Tony Z. Qiu
Abstract: Object detection is a cornerstone of environmental perception in advanced driver assistance systems(ADAS). However, most existing methods rely on RGB cameras, which suffer from significant performance degradation under low-light conditions due to poor image quality. To address this challenge, we proposes WTEFNet, a real-time object detection framework specifically designed for low-light scenarios, with strong adaptability to mainstream detectors. WTEFNet comprises three core modules: a Low-Light Enhancement (LLE) module, a Wavelet-based Feature Extraction (WFE) module, and an Adaptive Fusion Detection (AFFD) module. The LLE enhances dark regions while suppressing overexposed areas; the WFE applies multi-level discrete wavelet transforms to isolate high- and low-frequency components, enabling effective denoising and structural feature retention; the AFFD fuses semantic and illumination features for robust detection. To support training and evaluation, we introduce GSN, a manually annotated dataset covering both clear and rainy night-time scenes. Extensive experiments on BDD100K, SHIFT, nuScenes, and GSN demonstrate that WTEFNet achieves state-of-the-art accuracy under low-light conditions. Furthermore, deployment on a embedded platform (NVIDIA Jetson AGX Orin) confirms the framework's suitability for real-time ADAS applications.
Authors: Yazhou Zhang, Chunwang Zou, Qimeng Liu, Lu Rong, Ben Yao, Zheng Lian, Qiuchi Li, Peng Zhang, Jing Qin
Abstract: Can multi-modal large models (MLMs) that can ``see'' an image be said to ``understand'' it? Drawing inspiration from Searle's Chinese Room, we propose the \textbf{Visual Room} argument: a system may process and describe every detail of visual inputs by following algorithmic rules, without genuinely comprehending the underlying intention. This dilemma challenges the prevailing assumption that perceptual mastery implies genuine understanding. In implementation, we introduce a two-tier evaluation framework spanning perception and cognition. The perception component evaluates whether MLMs can accurately capture the surface-level details of visual contents, where the cognitive component examines their ability to infer sarcasm polarity. To support this framework, We further introduce a high-quality multi-modal sarcasm dataset comprising both 924 static images and 100 dynamic videos. All sarcasm labels are annotated by the original authors and verified by independent reviewers to ensure clarity and consistency. We evaluate eight state-of-the-art (SoTA) MLMs. Our results highlight three key findings: (1) MLMs demonstrate high accuracy in visual perception; (2) even with correct perception, MLMs exhibit an average error rate of ~17.1\% in sarcasm understanding, revealing a significant gap between seeing and understanding; (3) this gap stems from weaknesses in context integration, emotional reasoning, and pragmatic inference. This work provides empirical grounding for the proposed Visual Room argument and offers a new evaluation paradigm for MLMs.
Authors: Xu Chu, Xinrong Chen, Guanyu Wang, Zhijie Tan, Kui Huang, Wenyu Lv, Tong Mo, Weiping Li
Abstract: Inference time scaling drives extended reasoning to enhance the performance of Vision-Language Models (VLMs), thus forming powerful Vision-Language Reasoning Models (VLRMs). However, long reasoning dilutes visual tokens, causing visual information to receive less attention and may trigger hallucinations. Although introducing text-only reflection processes shows promise in language models, we demonstrate that it is insufficient to suppress hallucinations in VLMs. To address this issue, we introduce Qwen-LookAgain (Qwen-LA), a novel VLRM designed to mitigate hallucinations by incorporating a vision-text reflection process that guides the model to re-attention visual information during reasoning. We first propose a reinforcement learning method Balanced Reflective Policy Optimization (BRPO), which guides the model to decide when to generate vision-text reflection on its own and balance the number and length of reflections. Then, we formally prove that VLRMs lose attention to visual tokens as reasoning progresses, and demonstrate that supplementing visual information during reflection enhances visual attention. Therefore, during training and inference, Visual Token COPY and Visual Token ROUTE are introduced to force the model to re-attention visual information at the visual level, addressing the limitations of text-only reflection. Experiments on multiple visual QA datasets and hallucination metrics indicate that Qwen-LA achieves leading accuracy performance while reducing hallucinations. Our code is available at: https://github.com/Liar406/Look_Again
Authors: Size Wu, Zhonghua Wu, Zerui Gong, Qingyi Tao, Sheng Jin, Qinyue Li, Wei Li, Chen Change Loy
Abstract: In this report, we present OpenUni, a simple, lightweight, and fully open-source baseline for unifying multimodal understanding and generation. Inspired by prevailing practices in unified model learning, we adopt an efficient training strategy that minimizes the training complexity and overhead by bridging the off-the-shelf multimodal large language models (LLMs) and diffusion models through a set of learnable queries and a light-weight transformer-based connector. With a minimalist choice of architecture, we demonstrate that OpenUni can: 1) generate high-quality and instruction-aligned images, and 2) achieve exceptional performance on standard benchmarks such as GenEval, DPG- Bench, and WISE, with only 1.1B and 3.1B activated parameters. To support open research and community advancement, we release all model weights, training code, and our curated training datasets (including 23M image-text pairs) at https://github.com/wusize/OpenUni.
Authors: Weijie Wang, Donny Y. Chen, Zeyu Zhang, Duochao Shi, Akide Liu, Bohan Zhuang
Abstract: Feed-forward 3D Gaussian Splatting (3DGS) models have recently emerged as a promising solution for novel view synthesis, enabling one-pass inference without the need for per-scene 3DGS optimization. However, their scalability is fundamentally constrained by the limited capacity of their encoders, leading to degraded performance or excessive memory consumption as the number of input views increases. In this work, we analyze feed-forward 3DGS frameworks through the lens of the Information Bottleneck principle and introduce ZPressor, a lightweight architecture-agnostic module that enables efficient compression of multi-view inputs into a compact latent state $Z$ that retains essential scene information while discarding redundancy. Concretely, ZPressor enables existing feed-forward 3DGS models to scale to over 100 input views at 480P resolution on an 80GB GPU, by partitioning the views into anchor and support sets and using cross attention to compress the information from the support views into anchor views, forming the compressed latent state $Z$. We show that integrating ZPressor into several state-of-the-art feed-forward 3DGS models consistently improves performance under moderate input views and enhances robustness under dense view settings on two large-scale benchmarks DL3DV-10K and RealEstate10K. The video results, code and trained models are available on our project page: https://lhmd.top/zpressor.
Authors: Rickard Br\"uel-Gabrielsson, Tongzhou Wang, Manel Baradad, Justin Solomon
Abstract: Despite dropout's ubiquity in machine learning, its effectiveness as a form of data augmentation remains under-explored. We address two key questions: (i) When is dropout effective as an augmentation strategy? (ii) Is dropout uniquely effective under these conditions? To explore these questions, we propose Deep Augmentation, a network- and modality-agnostic method that applies dropout or PCA transformations to targeted layers in neural networks. Through extensive experiments on contrastive learning tasks in NLP, computer vision, and graph learning, we find that uniformly applying dropout across layers does not consistently improve performance. Instead, dropout proves most beneficial in deeper layers and can be matched by alternative augmentations (e.g., PCA). We also show that a stop-gradient operation is critical for ensuring dropout functions effectively as an augmentation, and that performance trends invert when moving from contrastive tasks to supervised tasks. Our analysis suggests that Deep Augmentation helps mitigate inter-layer co-adaptation -- a notable issue in self-supervised learning due to the absence of labeled data. Drawing on these insights, we outline a procedure for selecting the optimal augmentation layer and demonstrate that Deep Augmentation can outperform traditional input-level augmentations. This simple yet powerful approach can be seamlessly integrated into a wide range of architectures and modalities, yielding notable gains in both performance and generalization.
Authors: Ziyi Chang, George Alex Koulieris, Hyung Jin Chang, Hubert P. H. Shum
Abstract: Diffusion models are learning pattern-learning systems to model and sample from data distributions with three functional components namely the forward process, the reverse process, and the sampling process. The components of diffusion models have gained significant attention with many design factors being considered in common practice. Existing reviews have primarily focused on higher-level solutions, covering less on the design fundamentals of components. This study seeks to address this gap by providing a comprehensive and coherent review of seminal designable factors within each functional component of diffusion models. This provides a finer-grained perspective of diffusion models, benefiting future studies in the analysis of individual components, the design factors for different purposes, and the implementation of diffusion models.
Authors: Yihao Huang, Chong Wang, Xiaojun Jia, Qing Guo, Felix Juefei-Xu, Jian Zhang, Geguang Pu, Yang Liu
Abstract: Universal goal hijacking is a kind of prompt injection attack that forces LLMs to return a target malicious response for arbitrary normal user prompts. The previous methods achieve high attack performance while being too cumbersome and time-consuming. Also, they have concentrated solely on optimization algorithms, overlooking the crucial role of the prompt. To this end, we propose a method called POUGH that incorporates an efficient optimization algorithm and two semantics-guided prompt organization strategies. Specifically, our method starts with a sampling strategy to select representative prompts from a candidate pool, followed by a ranking strategy that prioritizes them. Given the sequentially ranked prompts, our method employs an iterative optimization algorithm to generate a fixed suffix that can concatenate to arbitrary user prompts for universal goal hijacking. Experiments conducted on four popular LLMs and ten types of target responses verified the effectiveness.
Authors: Weiyu Li, Jiarui Liu, Hongyu Yan, Rui Chen, Yixun Liang, Xuelin Chen, Ping Tan, Xiaoxiao Long
Abstract: We present a novel generative 3D modeling system, coined CraftsMan, which can generate high-fidelity 3D geometries with highly varied shapes, regular mesh topologies, and detailed surfaces, and, notably, allows for refining the geometry in an interactive manner. Despite the significant advancements in 3D generation, existing methods still struggle with lengthy optimization processes, irregular mesh topologies, noisy surfaces, and difficulties in accommodating user edits, consequently impeding their widespread adoption and implementation in 3D modeling software. Our work is inspired by the craftsman, who usually roughs out the holistic figure of the work first and elaborates the surface details subsequently. Specifically, we employ a 3D native diffusion model, which operates on latent space learned from latent set-based 3D representations, to generate coarse geometries with regular mesh topology in seconds. In particular, this process takes as input a text prompt or a reference image and leverages a powerful multi-view (MV) diffusion model to generate multiple views of the coarse geometry, which are fed into our MV-conditioned 3D diffusion model for generating the 3D geometry, significantly improving robustness and generalizability. Following that, a normal-based geometry refiner is used to significantly enhance the surface details. This refinement can be performed automatically, or interactively with user-supplied edits. Extensive experiments demonstrate that our method achieves high efficacy in producing superior-quality 3D assets compared to existing methods. HomePage: https://craftsman3d.github.io/, Code: https://github.com/wyysf-98/CraftsMan
URLs: https://craftsman3d.github.io/,, https://github.com/wyysf-98/CraftsMan
Authors: Wentong Chen, Junbo Cui, Jinyi Hu, Yujia Qin, Junjie Fang, Yue Zhao, Chongyi Wang, Jun Liu, Guirong Chen, Yupeng Huo, Yuan Yao, Yankai Lin, Zhiyuan Liu, Maosong Sun
Abstract: Utilizing Graphic User Interface (GUI) for human-computer interaction is essential for accessing a wide range of digital tools. Recent advancements in Vision Language Models (VLMs) highlight the compelling potential to develop versatile agents to help humans finish GUI navigation tasks. However, current VLMs are challenged in terms of fundamental abilities (OCR and grounding) and GUI knowledge (the functions and control methods of GUI elements), preventing them from becoming practical GUI agents. To solve these challenges, we contribute GUICourse, a suite of datasets to train visual-based GUI agents from general VLMs. First, we introduce the GUIEnv dataset to strengthen the OCR and grounding capabilities of VLMs. Then, we introduce the GUIAct and GUIChat datasets to enrich their knowledge of GUI components and interactions. Experiments demonstrate that our GUI agents have better performance on common GUI tasks than their baseline VLMs. Even the small-size GUI agent (with 3.1B parameters) can still work well on single-step and multi-step GUI tasks. Finally, we analyze the different varieties in the training stage of this agent by ablation study. Our source codes and datasets are released at https://github.com/yiye3/GUICourse.
Authors: ChaeHun Park, Yujin Baek, Jaeseok Kim, Yu-Jung Heo, Du-Seong Chang, Jaegul Choo
Abstract: To create culturally inclusive vision-language models (VLMs), developing a benchmark that tests their ability to address culturally relevant questions is essential. Existing approaches typically rely on human annotators, making the process labor-intensive and creating a cognitive burden in generating diverse questions. To address this, we propose a semi-automated framework for constructing cultural VLM benchmarks, specifically targeting multiple-choice QA. This framework combines human-VLM collaboration, where VLMs generate questions based on guidelines, a small set of annotated examples, and relevant knowledge, followed by a verification process by native speakers. We demonstrate the effectiveness of this framework through the creation of \texttt{K-Viscuit}, a dataset focused on Korean culture. Our experiments on this dataset reveal that open-source models lag behind proprietary ones in understanding Korean culture, highlighting key areas for improvement. We also present a series of further analyses, including human evaluation, augmenting VLMs with external knowledge, and the evaluation beyond multiple-choice QA. Our dataset is available at https://huggingface.co/datasets/ddehun/k-viscuit.
Authors: Kening Zheng, Junkai Chen, Yibo Yan, Xin Zou, Xuming Hu
Abstract: Hallucination issues continue to affect multimodal large language models (MLLMs), with existing research mainly addressing object-level or attribute-level hallucinations, neglecting the more complex relation hallucinations that require advanced reasoning. Current benchmarks for relation hallucinations lack detailed evaluation and effective mitigation, and their datasets often suffer from biases due to systematic annotation processes. To address these challenges, we introduce Reefknot, a comprehensive benchmark targeting relation hallucinations, comprising over 20,000 real-world samples. We provide a systematic definition of relation hallucinations, integrating perceptive and cognitive perspectives, and construct a relation-based corpus using the Visual Genome scene graph dataset. Our comparative evaluation reveals significant limitations in current MLLMs' ability to handle relation hallucinations. Additionally, we propose a novel confidence-based mitigation strategy, which reduces the hallucination rate by an average of 9.75% across three datasets, including Reefknot. Our work offers valuable insights for achieving trustworthy multimodal intelligence.
Authors: Giorgio Piras, Maura Pintor, Ambra Demontis, Battista Biggio, Giorgio Giacinto, Fabio Roli
Abstract: Recent work has proposed neural network pruning techniques to reduce the size of a network while preserving robustness against adversarial examples, i.e., well-crafted inputs inducing a misclassification. These methods, which we refer to as adversarial pruning methods, involve complex and articulated designs, making it difficult to analyze the differences and establish a fair and accurate comparison. In this work, we overcome these issues by surveying current adversarial pruning methods and proposing a novel taxonomy to categorize them based on two main dimensions: the pipeline, defining when to prune; and the specifics, defining how to prune. We then highlight the limitations of current empirical analyses and propose a novel, fair evaluation benchmark to address them. We finally conduct an empirical re-evaluation of current adversarial pruning methods and discuss the results, highlighting the shared traits of top-performing adversarial pruning methods, as well as common issues. We welcome contributions in our publicly-available benchmark at https://github.com/pralab/AdversarialPruningBenchmark
Authors: Chaohui Xu, Qi Cui, Jinxin Dong, Weiyang He, Chip-Hong Chang
Abstract: Illegitimate reproduction, distribution and derivation of Deep Neural Network (DNN) models can inflict economic loss, reputation damage and even privacy infringement. Passive DNN intellectual property (IP) protection methods such as watermarking and fingerprinting attempt to prove the ownership upon IP violation, but they are often too late to stop catastrophic damage of IP abuse and too feeble against strong adversaries. In this paper, we propose IDEA, an Inverse Domain Expert Adaptation based proactive DNN IP protection method featuring active authorization and source traceability. IDEA generalizes active authorization as an inverse problem of domain adaptation. The multi-adaptive optimization is solved by a mixture-of-experts model with one real and two fake experts. The real expert re-optimizes the source model to correctly classify test images with a unique model user key steganographically embedded. The fake experts are trained to output random prediction on test images without or with incorrect user key embedded by minimizing their mutual information (MI) with the real expert. The MoE model is knowledge distilled into a unified protected model to avoid leaking the expert model features by maximizing their MI with additional multi-layer attention and contrastive representation loss optimization. IDEA not only prevents unauthorized users without the valid key to access the functional model, but also enable the model owner to validate the deployed model and trace the source of IP infringement. We extensively evaluate IDEA on five datasets and four DNN models to demonstrate its effectiveness in authorization control, culprit tracing success rate, and robustness against various attacks.
Authors: Changdae Oh, Yixuan Li, Kyungwoo Song, Sangdoo Yun, Dongyoon Han
Abstract: Adapting a pre-trained foundation model on downstream tasks should ensure robustness against distribution shifts without the need to retrain the whole model. Although existing weight interpolation methods are simple yet effective, we argue that their static nature limits downstream performance while achieving efficiency. In this work, we propose DaWin, a training-free dynamic weight interpolation method that leverages the entropy of individual models over each unlabeled test sample to assess model expertise, and compute per-sample interpolation coefficients dynamically. Unlike previous works that typically rely on additional training to learn such coefficients, our approach requires no training. Then, we propose a mixture modeling approach that greatly reduces inference overhead raised by dynamic interpolation. We validate DaWin on the large-scale visual recognition benchmarks, spanning 14 tasks across robust fine-tuning -- ImageNet and derived five distribution shift benchmarks -- and multi-task learning with eight classification tasks. Results demonstrate that DaWin achieves significant performance gain in considered settings, with minimal computational overhead. We further discuss DaWin's analytic behavior to explain its empirical success.
Authors: Linger Deng, Linghao Zhu, Yuliang Liu, Yu Wang, Qunyi Xie, Jingjing Wu, Gang Zhang, Yingying Zhu, Xiang Bai
Abstract: Large Multimodal Models (LMMs) face limitations in geometric reasoning due to insufficient Chain of Thought (CoT) image-text training data. While existing approaches leverage template-based or LLM-assisted methods for geometric CoT data creation, they often face challenges in achieving both diversity and precision. To bridge this gap, we introduce a two-stage Theorem-Validated Reverse Chain-of-Thought Reasoning Synthesis (TR-CoT) framework. The first stage, TR-Engine, synthesizes theorem-grounded geometric diagrams with structured descriptions and properties. The second stage, TR-Reasoner, employs reverse reasoning to iteratively refine question-answer pairs by cross-validating geometric properties and description fragments. Our approach expands theorem-type coverage, corrects long-standing misunderstandings, and enhances geometric reasoning. Fine-grained CoT improves theorem understanding and increases logical consistency by 24.5%. Our best models surpass the baselines in MathVista and GeoQA by 10.1% and 4.7%, outperforming advanced closed-source models like GPT-4o.
Authors: Viacheslav Surkov, Chris Wendler, Antonio Mari, Mikhail Terekhov, Justin Deschenaux, Robert West, Caglar Gulcehre, David Bau
Abstract: For large language models (LLMs), sparse autoencoders (SAEs) have been shown to decompose intermediate representations that often are not interpretable directly into sparse sums of interpretable features, facilitating better control and subsequent analysis. However, similar analyses and approaches have been lacking for text-to-image models. We investigate the possibility of using SAEs to learn interpretable features for SDXL Turbo, a few-step text-to-image diffusion model. To this end, we train SAEs on the updates performed by transformer blocks within SDXL Turbo's denoising U-net in its 1-step setting. Interestingly, we find that they generalize to 4-step SDXL Turbo and even to the multi-step SDXL base model (i.e., a different model) without additional training. In addition, we show that their learned features are interpretable, causally influence the generation process, and reveal specialization among the blocks. We do so by creating RIEBench, a representation-based image editing benchmark, for editing images while they are generated by turning on and off individual SAE features. This allows us to track which transformer blocks' features are the most impactful depending on the edit category. Our work is the first investigation of SAEs for interpretability in text-to-image diffusion models and our results establish SAEs as a promising approach for understanding and manipulating the internal mechanisms of text-to-image models.
Authors: Jingyi Zhou, Senlin Luo, Haofan Chen
Abstract: Text emotion detection constitutes a crucial foundation for advancing artificial intelligence from basic comprehension to the exploration of emotional reasoning. Most existing emotion detection datasets rely on manual annotations, which are associated with high costs, substantial subjectivity, and severe label imbalances. This is particularly evident in the inadequate annotation of micro-emotions and the absence of emotional intensity representation, which fail to capture the rich emotions embedded in sentences and adversely affect the quality of downstream task completion. By proposing an all-labels and training-set label regression method, we map label values to energy intensity levels, thereby fully leveraging the learning capabilities of machine models and the interdependencies among labels to uncover multiple emotions within samples. This led to the establishment of the Emotion Quantization Network (EQN) framework for micro-emotion detection and annotation. Using five commonly employed sentiment datasets, we conducted comparative experiments with various models, validating the broad applicability of our framework within NLP machine learning models. Based on the EQN framework, emotion detection and annotation are conducted on the GoEmotions dataset. A comprehensive comparison with the results from Google literature demonstrates that the EQN framework possesses a high capability for automatic detection and annotation of micro-emotions. The EQN framework is the first to achieve automatic micro-emotion annotation with energy-level scores, providing strong support for further emotion detection analysis and the quantitative research of emotion computing.
Authors: Soobin Um, Beomsu Kim, Jong Chul Ye
Abstract: Minority samples are underrepresented instances located in low-density regions of a data manifold, and are valuable in many generative AI applications, such as data augmentation, creative content generation, etc. Unfortunately, existing diffusion-based minority generators often rely on computationally expensive guidance dedicated for minority generation. To address this, here we present a simple yet powerful guidance-free approach called Boost-and-Skip for generating minority samples using diffusion models. The key advantage of our framework requires only two minimal changes to standard generative processes: (i) variance-boosted initialization and (ii) timestep skipping. We highlight that these seemingly-trivial modifications are supported by solid theoretical and empirical evidence, thereby effectively promoting emergence of underrepresented minority features. Our comprehensive experiments demonstrate that Boost-and-Skip greatly enhances the capability of generating minority samples, even rivaling guidance-based state-of-the-art approaches while requiring significantly fewer computations. Code is available at https://github.com/soobin-um/BnS.
Authors: Jihao Gu, Yingyao Wang, Pi Bu, Chen Wang, Ziming Wang, Tengtao Song, Donglai Wei, Jiale Yuan, Yingxiu Zhao, Yancheng He, Shilong Li, Jiaheng Liu, Meng Cao, Jun Song, Yingshui Tan, Xiang Li, Wenbo Su, Zhicheng Zheng, Xiaoyong Zhu, Bo Zheng
Abstract: The evaluation of factual accuracy in large vision language models (LVLMs) has lagged behind their rapid development, making it challenging to fully reflect these models' knowledge capacity and reliability. In this paper, we introduce the first factuality-based visual question-answering benchmark in Chinese, named ChineseSimpleVQA, aimed at assessing the visual factuality of LVLMs across 8 major topics and 56 subtopics. The key features of this benchmark include a focus on the Chinese language, diverse knowledge types, a multi-hop question construction, high-quality data, static consistency, and easy-to-evaluate through short answers. Moreover, we contribute a rigorous data construction pipeline and decouple the visual factuality into two parts: seeing the world (i.e., object recognition) and discovering knowledge. This decoupling allows us to analyze the capability boundaries and execution mechanisms of LVLMs. Subsequently, we evaluate 34 advanced open-source and closed-source models, revealing critical performance gaps within this field. Our evaluation-friendly code and data have already been open-sourced.
Authors: Henrui Tian, Wenhui Lei, Linrui Dai, Hanyu Chen, Xiaofan Zhang
Abstract: Fully-supervised lesion recognition methods in medical imaging face challenges due to the reliance on large annotated datasets, which are expensive and difficult to collect. To address this, synthetic lesion generation has become a promising approach. However, existing models struggle with scalability, fine-grained control over lesion attributes, and the generation of complex structures. We propose LesionDiffusion, a text-controllable lesion synthesis framework for 3D CT imaging that generates both lesions and corresponding masks. By utilizing a structured lesion report template, our model provides greater control over lesion attributes and supports a wider variety of lesion types. We introduce a dataset of 1,505 annotated CT scans with paired lesion masks and structured reports, covering 14 lesion types across 8 organs. LesionDiffusion consists of two components: a lesion mask synthesis network (LMNet) and a lesion inpainting network (LINet), both guided by lesion attributes and image features. Extensive experiments demonstrate that LesionDiffusion significantly improves segmentation performance, with strong generalization to unseen lesion types and organs, outperforming current state-of-the-art models. Code is available at https://github.com/HengruiTianSJTU/LesionDiffusion.
Authors: Zixuan Wang, Chi-Keung Tang, Yu-Wing Tai
Abstract: Current audio generation conditioned by text or video focuses on aligning audio with text/video modalities. Despite excellent alignment results, these multimodal frameworks still cannot be directly applied to compelling movie storytelling involving multiple scenes, where "on-screen" sounds require temporally-aligned audio generation, while "off-screen" sounds contribute to appropriate environment sounds accompanied by background music when applicable. Inspired by professional movie production, this paper proposes a multi-agentic framework for audio generation supervised by an autonomous Sound Director agent, engaging multi-turn conversations with other agents for on-screen and off-screen sound generation through multimodal LLM. To address on-screen sound generation, after detecting any talking humans in videos, we capture semantically and temporally synchronized sound by training a prediction model that forecasts interpretable, time-varying audio control signals: loudness, pitch, and timbre, which are used by a Foley Artist agent to condition a cross-attention module in the sound generation. The Foley Artist works cooperatively with the Composer and Voice Actor agents, and together they autonomously generate off-screen sound to complement the overall production. Each agent takes on specific roles similar to those of a movie production team. To temporally ground audio language models, in ReelWave, text/video conditions are decomposed into atomic, specific sound generation instructions synchronized with visuals when applicable. Consequently, our framework can generate rich and relevant audio content conditioned on video clips extracted from movies.
Authors: Pascal Schlachter, Jonathan Fuss, Bin Yang
Abstract: A domain (distribution) shift between training and test data often hinders the real-world performance of deep neural networks, necessitating unsupervised domain adaptation (UDA) to bridge this gap. Online source-free UDA has emerged as a solution for practical scenarios where access to source data is restricted and target data is received as a continuous stream. However, the open-world nature of many real-world applications additionally introduces category shifts meaning that the source and target label spaces may differ. Online source-free universal domain adaptation (SF-UniDA) addresses this challenge. Existing methods mainly rely on self-training with pseudo-labels, yet the relationship between pseudo-labeling and adaptation outcomes has not been studied yet. To bridge this gap, we conduct a systematic analysis through controlled experiments with simulated pseudo-labeling, offering valuable insights into pseudo-labeling for online SF-UniDA. Our findings reveal a substantial gap between the current state-of-the-art and the upper bound of adaptation achieved with perfect pseudo-labeling. Moreover, we show that a contrastive loss enables effective adaptation even with moderate pseudo-label accuracy, while a cross-entropy (CE) loss, though less robust to pseudo-label errors, achieves superior results when pseudo-labeling approaches perfection. Lastly, our findings indicate that pseudo-label accuracy is in general more crucial than quantity, suggesting that prioritizing fewer but high-confidence pseudo-labels is beneficial. Overall, our study highlights the critical role of pseudo-labeling in (online) SF-UniDA and provides actionable insights to drive future advancements in the field. Our code is available at https://github.com/pascalschlachter/PLAnalysis.
Authors: Dawei Huang, Qing Li, Chuan Yan, Zebang Cheng, Yurong Huang, Xiang Li, Bin Li, Xiaohui Wang, Zheng Lian, Xiaojiang Peng
Abstract: Emotion understanding in videos aims to accurately recognize and interpret individuals' emotional states by integrating contextual, visual, textual, and auditory cues. While Large Multimodal Models (LMMs) have demonstrated significant progress in general vision-language (VL) tasks, their performance in emotion-specific scenarios remains limited. Moreover, fine-tuning LMMs on emotion-related tasks often leads to catastrophic forgetting, hindering their ability to generalize across diverse tasks. To address these challenges, we present Emotion-Qwen, a tailored multimodal framework designed to enhance both emotion understanding and general VL reasoning. Emotion-Qwen incorporates a sophisticated Hybrid Compressor based on the Mixture of Experts (MoE) paradigm, which dynamically routes inputs to balance emotion-specific and general-purpose processing. The model is pre-trained in a three-stage pipeline on large-scale general and emotional image datasets to support robust multimodal representations. Furthermore, we construct the Video Emotion Reasoning (VER) dataset, comprising more than 40K bilingual video clips with fine-grained descriptive annotations, to further enrich Emotion-Qwen's emotional reasoning capability. Experimental results demonstrate that Emotion-Qwen achieves state-of-the-art performance on multiple emotion recognition benchmarks, while maintaining competitive results on general VL tasks. Code and models are available at https://github.com/24DavidHuang/Emotion-Qwen.
Authors: Kevin Lu, Nicky Kriplani, Rohit Gandikota, Minh Pham, David Bau, Chinmay Hegde, Niv Cohen
Abstract: Concept erasure, the ability to selectively prevent a model from generating specific concepts, has attracted growing interest, with various approaches emerging to address the challenge. However, it remains unclear how thoroughly these methods erase the target concept. We begin by proposing two conceptual models for the erasure mechanism in diffusion models: (i) reducing the likelihood of generating the target concept, and (ii) interfering with the model's internal guidance mechanisms. To thoroughly assess whether a concept has been truly erased from the model, we introduce a suite of independent evaluations. Our evaluation framework includes adversarial attacks, novel probing techniques, and analysis of the model's alternative generations in place of the erased concept. Our results shed light on the tension between minimizing side effects and maintaining robustness to adversarial prompts. Broadly, our work underlines the importance of comprehensive evaluation for erasure in diffusion models.
Authors: Alex L. Zhang, Thomas L. Griffiths, Karthik R. Narasimhan, Ofir Press
Abstract: Vision-language models (VLMs) have achieved strong results on coding and math benchmarks that are challenging for humans, yet their ability to perform tasks that come naturally to humans--such as perception, spatial navigation, and memory management--remains understudied. Real video games are crafted to be intuitive for humans to learn and master by leveraging innate inductive biases, making them an ideal testbed for evaluating such capabilities in VLMs. To this end, we introduce VideoGameBench, a benchmark consisting of 10 popular video games from the 1990s that VLMs directly interact with in real-time. VideoGameBench challenges models to complete entire games with access to only raw visual inputs and a high-level description of objectives and controls, a significant departure from existing setups that rely on game-specific scaffolding and auxiliary information. We keep three of the games secret to encourage solutions that generalize to unseen environments. Our experiments show that frontier vision-language models struggle to progress beyond the beginning of each game. We find inference latency to be a major limitation of frontier models in the real-time setting; therefore, we introduce VideoGameBench Lite, a setting where the game pauses while waiting for the LM's next action. The best performing model, Gemini 2.5 Pro, completes only 0.48% of VideoGameBench and 1.6% of VideoGameBench Lite. We hope that the formalization of the human skills mentioned above into this benchmark motivates progress in these research directions.
Authors: Yubo Peng, Luping Xiang, Bingxin Zhang, Kun Yang
Abstract: Traditional single-modal sensing systems-based solely on either radio frequency (RF) or visual data-struggle to cope with the demands of complex and dynamic environments. Furthermore, single-device systems are constrained by limited perspectives and insufficient spatial coverage, which impairs their effectiveness in urban or non-line-of-sight scenarios. To overcome these challenges, we propose a novel large language model (LLM)-driven distributed integrated multimodal sensing and semantic communication (LLM-DiSAC) framework. Specifically, our system consists of multiple collaborative sensing devices equipped with RF and camera modules, working together with an aggregation center to enhance sensing accuracy. First, on sensing devices, LLM-DiSAC develops an RF-vision fusion network (RVFN), which employs specialized feature extractors for RF and visual data, followed by a cross-attention module for effective multimodal integration. Second, a LLM-based semantic transmission network (LSTN) is proposed to enhance communication efficiency, where the LLM-based decoder leverages known channel parameters, such as transceiver distance and signal-to-noise ratio (SNR), to mitigate semantic distortion. Third, at the aggregation center, a transformer-based aggregation model (TRAM) with an adaptive aggregation attention mechanism is developed to fuse distributed features and enhance sensing accuracy. To preserve data privacy, a two-stage distributed learning strategy is introduced, allowing local model training at the device level and centralized aggregation model training using intermediate features. Finally, evaluations on a synthetic multi-view RF-visual dataset generated by the Genesis simulation engine show that LLM-DiSAC achieves a good performance.
Authors: Atsunori Moteki, Shoichi Masui, Fan Yang, Yueqi Song, Yonatan Bisk, Graham Neubig, Ikuo Kusajima, Yasuto Watanabe, Hiroyuki Ishida, Jun Takahashi, Shan Jiang
Abstract: This paper proposes FieldWorkArena, a benchmark for agentic AI targeting real-world field work. With the recent increase in demand for agentic AI, they are required to monitor and report safety and health incidents, as well as manufacturing-related incidents, that may occur in real-world work environments. Existing agentic AI benchmarks have been limited to evaluating web tasks and are insufficient for evaluating agents in real-world work environments, where complexity increases significantly. In this paper, we define a new action space that agentic AI should possess for real world work environment benchmarks and improve the evaluation function from previous methods to assess the performance of agentic AI in diverse real-world tasks. The dataset consists of videos captured on-site and documents actually used in factories and warehouses, and tasks were created based on interviews with on-site workers and managers. Evaluation results confirmed that performance evaluation considering the characteristics of Multimodal LLM (MLLM) such as GPT-4o is feasible. Additionally, the effectiveness and limitations of the proposed new evaluation method were identified. The complete dataset (HuggingFace) and evaluation program (GitHub) can be downloaded from the following website: https://en-documents.research.global.fujitsu.com/fieldworkarena/.
URLs: https://en-documents.research.global.fujitsu.com/fieldworkarena/.
Authors: Abhijit Chunduru, Majid Morafah, Mahdi Morafah, Vishnu Pandi Chellapandi, Ang Li
Abstract: The inevitable presence of data heterogeneity has made federated learning very challenging. There are numerous methods to deal with this issue, such as local regularization, better model fusion techniques, and data sharing. Though effective, they lack a deep understanding of how data heterogeneity can affect the global decision boundary. In this paper, we bridge this gap by performing an experimental analysis of the learned decision boundary using a toy example. Our observations are surprising: (1) we find that the existing methods suffer from forgetting and clients forget the global decision boundary and only learn the perfect local one, and (2) this happens regardless of the initial weights, and clients forget the global decision boundary even starting from pre-trained optimal weights. In this paper, we present FedProj, a federated learning framework that robustly learns the global decision boundary and avoids its forgetting during local training. To achieve better ensemble knowledge fusion, we design a novel server-side ensemble knowledge transfer loss to further calibrate the learned global decision boundary. To alleviate the issue of learned global decision boundary forgetting, we further propose leveraging an episodic memory of average ensemble logits on a public unlabeled dataset to regulate the gradient updates at each step of local training. Experimental results demonstrate that FedProj outperforms state-of-the-art methods by a large margin.
Authors: Zhongyi Zhou, Yichen Zhu, Junjie Wen, Chaomin Shen, Yi Xu
Abstract: Vision-language-action (VLA) models have emerged as the next generation of models in robotics. However, despite leveraging powerful pre-trained Vision-Language Models (VLMs), existing end-to-end VLA systems often lose key capabilities during fine-tuning as the model adapts to specific robotic tasks. We argue that a generalizable VLA model should retain and expand upon the VLM's core competencies: 1) Open-world embodied reasoning - the VLA should inherit the knowledge from VLM, i.e., recognize anything that the VLM can recognize, be capable of solving math problems, and possess visual-spatial intelligence, 2) Reasoning following - effectively translating the open-world reasoning into actionable steps for the robot. In this work, we introduce ChatVLA-2, a novel mixture-of-expert VLA model coupled with a specialized two-stage training pipeline designed to preserve the VLM's original strengths while enabling actionable reasoning. To validate our approach, we design a math-matching task wherein a robot interprets math problems written on a whiteboard and picks corresponding number cards from a table to solve equations. Remarkably, our method exhibits exceptional mathematical reasoning and OCR capabilities, despite these abilities not being explicitly trained within the VLA. Furthermore, we demonstrate that the VLA possesses strong spatial reasoning skills, enabling it to interpret novel directional instructions involving previously unseen objects. Overall, our method showcases reasoning and comprehension abilities that significantly surpass state-of-the-art imitation learning methods such as OpenVLA, DexVLA, and pi-zero. This work represents a substantial advancement toward developing truly generalizable robotic foundation models endowed with robust reasoning capacities.