HART: High-Resolution Annotation-Free Reasoning Technique through a Closed-loop Framework
State Key Laboratory of Novel Software Technology(新型软件技术国家重点实验室);Institute of Brain-inspired Intelligence(脑启发式智能研究院);Shenzhen Research Institute of Nanjing University(南京大学深圳研究院)
Current Large Multimodal Models (LMMs) struggle with high-resolution visual inputs during the reasoning process, as the number of image tokens increases quadratically with resolution, introducing substantial redundancy and irrelevant information. A common practice is to identify key image regions and refer to their high-resolution counterparts during reasoning, typically trained with external visual supervision. However, such visual supervision cues require costly grounding labels from human annotators. Meanwhile, it remains an open question how to enhance a model's grounding abilities to support reasoning without relying on additional annotations. In this paper, we propose High-resolution Annotation-free Reasoning Technique (HART), a closed-loop framework that enables LMMs to focus on and self-verify key regions of high-resolution visual inputs. HART incorporates a post-training paradigm in which we design Advantage Preference Group Relative Policy Optimization (AP-GRPO) to encourage accurate localization of key regions without external visual annotations. Notably, HART provides explainable reasoning pathways and enables efficient optimization of localization. Extensive experiments on MME-RealWorld-Lite, TreeBench, V* Bench, HR-Bench-4K/8K, and MMStar demonstrate that HART improves performance across a wide range of high-resolution visual tasks, consistently outperforming strong baselines.
StreamingCoT: A Dataset for Temporal Dynamics and Multimodal Chain-of-Thought Reasoning in Streaming VideoQA
Henan Institute of Advanced Technology, Zhengzhou University(河南高级技术研究所,郑州大学);Institute of Automation, CAS(自动化研究所,中国科学院);UCAS(中国科学院大学);Peng Cheng Laboratory(鹏城实验室)
CoTZero: Annotation-Free Human-Like Vision Reasoning via Hierarchical Synthetic CoT
University of Electronic Science and Technology of China(电子科技大学);Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);The Chinese University of Hong Kong MMLab(香港中文大学 MMLab);The College of Computer Science and Technology(计算机科学与技术学院);Nanjing University of Aeronautics and Astronautics(南京航空航天大学)
GenCAMO: Scene-Graph Contextual Decoupling for Environment-aware and Mask-free Camouflage Image-Dense Annotation Generation
China University of Petroleum (East China)(中国石油大学(华东));Southwest Jiaotong University(西南交通大学);Beijing Information Science and Technology University(北京信息科技大学);Beihang University(北航)
Aligning Vision to Language: Annotation-Free Multimodal Knowledge Graph Construction for Enhanced LLMs Reasoning
Tongji University(同济大学);Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);East China Normal University(华东师范大学);Stanford University(斯坦福大学);New York University(纽约大学)
本文提出 VaLiK 方法,通过跨模态信息补充构建无需注释的多模态知识图谱,提升大语言模型推理能力。
ReVSI: Rebuilding Visual Spatial Intelligence Evaluation for Accurate Assessment of VLM 3D Reasoning
Simon Fraser University(西蒙弗雷泽大学);University of Waterloo(滑铁卢大学);Hong Kong University of Science(香港科学大学);Alberta Machine Intelligence Institute (Amii)(阿尔伯塔人工智能研究所(Amii))
Reasoning as Intersection: Consensus-Frame Alignment for Visual Focus in Video-MLLMs
School of Information Science and Engineering, Lanzhou University(兰州大学信息科学与工程学院);Beijing University of Posts and Telecommunications(北京邮电大学);Cloud and AI BU, Huawei(华为云与AI业务部);School of Computing, National University of Singapore(新加坡国立大学计算机学院)
Mags-RL: Wearing Multimodal LLMs a Magnifying Glass via Agentic Reinforcement Learning For Complex Scene Reasoning
Arizona State University(亚利桑那州立大学);Clemson University(克莱姆森大学);Washington University in St. Louis(圣路易斯华盛顿大学);Halmstad University(哈姆斯塔德大学);Florida State University(佛罗里达州立大学);Rice University(里士满大学)
MammoExpert: Benchmarking Chain-of-Thought Reasoning in Mammography Diagnosis
State Key Laboratory of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University(北京大学智能科学与技术学院通用人工智能国家重点实验室);School of Computer Science and Engineer, Beijing University of Aeronautics and Astronautics(北京航空航天大学计算机科学与工程学院);Center for Data Science, Peking University(北京大学数据科学中心);Yizhun co. ltd(医准有限公司);International School, Beijing University of Post and Telecommunications(北京邮电大学国际学院);School of Public Health, University of Michigan, Ann Arbor(密歇根大学安娜堡分校公共卫生学院);Future Technology College, Xi'an Jiaotong University(西安交通大学未来技术学院);Peking University(北京大学)
AMPLIFAI: A Multiphase CT Dataset for Benchmarking Clinical Reasoning in LI-RADS Assessment of Liver Lesions
University of Maryland, College Park(马里兰大学帕克分校);University of Maryland School of Medicine(马里兰大学医学院);University of Maryland Institute for Health Computing(马里兰大学健康计算研究所);University of Maryland Medical System(马里兰大学医学系统)
IBM Almaden Research Center(IBM阿尔马登研究中心);Rensselaer Polytechnic Institute(伦斯勒理工学院);Virginia Polytechnic Institute and State University(弗吉尼亚理工学院暨州立大学);MIT Critical Data(麻省理工学院关键数据实验室);Albert Einstein Healthcare Network-Philadelphia Campus(阿尔伯特·爱因斯坦医疗网络费城院区);Harvard Medical School(哈佛医学院)
Perception-R1: Advancing Multimodal Reasoning Capabilities of MLLMs via Visual Perception Reward
University of Science and Technology of China(中国科学技术大学);State Key Laboratory of Cognitive Intelligence(认知智能国家重点实验室);iFLYTEK AI Research(iFLYTEK人工智能研究院);Hong Kong University of Science and Technology(香港科学与技术大学)