Can warping tokens, rather than pixels, help multimodal large language models (MLLMs) understand how a scene appears from a nearby viewpoint? While MLLMs perform well on visual reasoning, they remain fragile to viewpoint changes, as pixel-wise warping is highly sensitive to small depth errors and often introduces geometric distortions. Drawing on theories of mental imagery that posit part-level structural representations as the basis for human perspective transformation, we examine whether image tokens in ViT-based MLLMs serve as an effective substrate for viewpoint changes. We compare forward and backward warping, finding that backward token warping, which defines a dense grid on the target view and retrieves a corresponding source-view token for each grid point, achieves greater stability and better preserves semantic coherence under viewpoint shifts. Experiments on our proposed ViewBench benchmark demonstrate that token-level warping enables MLLMs to reason reliably from nearby viewpoints, consistently outperforming all baselines including pixel-wise warping approaches, spatially fine-tuned MLLMs, and a generative warping method.
Actial: Activate Spatial Reasoning Ability of Multimodal Large Language Models
Nanjing University(南京大学);Xiaohongshu Inc.(小红书公司);East China Normal University(华东师范大学);The Chinese University of Hong Kong(香港中文大学);Shanghai Jiao Tong University(上海交通大学);University of Oxford(牛津大学)
Baidu Inc.(百度公司);Tianjin University(天津大学);The University of Sydney(悉尼大学);University of Technology Sydney(悉尼科技大学);Tsinghua University(清华大学);Chinese Academy of Science(中国科学院)
DynTrace: Tracking Dynamic Object Evidence for 4D Spatio-Temporal Reasoning in MLLMs
Xiamen University(厦门大学);Shenzhen International Graduate School, Tsinghua University(清华大学深圳国际研究生院);China University of Mining and Technology(中国矿业大学);Shenzhen Technology University(深圳技术大学)
State Beyond Appearance: Diagnosing and Improving State Consistency in Dial-Based Measurement Reading
Beijing Advanced Innovation Center for Future Blockchain and Privacy Computing(未来区块链与隐私计算北京先进创新中心);Beihang University(北航);Fudan University(复旦大学);Peking University(北京大学)
Agentic Collaborative Cognition for Zero-Shot 3D Understanding
Sichuan University(四川大学);Adelaide University(阿德莱德大学);University of Electronic Science and Technology of China(电子科技大学);Beijing Institute of Technology(北京理工大学)
STAR-R1: Spatial TrAnsformation Reasoning by Reinforcing Multimodal LLMs
Gaoling School of Artificial Intelligence, Renmin University of China(中国人民大学北京校区人工智能学院);MAIS, Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所MAIS);DAMO Academy, Alibaba Group, Hangzhou, China(阿里云达摩院);Hupan Lab, Hangzhou, China(杭州华普实验室)
Coarse Correspondences Boost Spatial-Temporal Reasoning in Multimodal Language Model
University of Washington(华盛顿大学);Tsinghua University(清华大学);Tencent(腾讯);Google Deepmind(谷歌DeepMind);Allen Institute for AI(艾伦人工智能研究所);Cornell University(康奈尔大学)
Earth-OneVision: Extending Remote Sensing Multimodal Large Language Models to More Sensor Modalities and Tasks
National Key Laboratory of Science and Technology on Space-Born Intelligent Information Processing (SBIIP), Beijing Institute of Technology(北京理工大学空间智能信息处理国家重点实验室);Aerospace Information Research Institute, Chinese Academy of Sciences(中国科学院空天信息创新研究院);Key Laboratory of Technology in Geo-Spatial Information Processing and Application System, Chinese Academy of Sciences(中国科学院地理空间信息处理与应用系统技术重点实验室);Advanced Research Institute of Multidisciplinary Sciences, Beijing Institute of Technology(北京理工大学前沿交叉科学研究院);School of Mechatronical Engineering, Beijing Institute of Technology(北京理工大学机电学院);School of Earth and Space Sciences, Peking University(北京大学地球与空间科学学院);School of Electronics, Peking University(北京大学电子学院);School of Computer Science and Hubei Key Laboratory of Intelligent Geo-Information Processing(华中科技大学计算机科学与技术学院&湖北省智能地理信息处理重点实验室)