SpatiaLab: Can Vision-Language Models Perform Spatial Reasoning in the Wild?
Computational Intelligence and Operations Laboratory(计算智能与运筹实验室);Shahjalal University of Science and Technology(沙赫jalal科技大学);BRAC University(BRAC大学);North South University(北南大学);Monash University(墨尔本大学);Qatar Computing Research Institute(卡塔尔计算研究院)
Spatial reasoning is a fundamental aspect of human cognition, yet it remains a major challenge for contemporary vision-language models (VLMs). Prior work largely relied on synthetic or LLM-generated environments with limited task designs and puzzle-like setups, failing to capture the real-world complexity, visual noise, and diverse spatial relationships that VLMs encounter. To address this, we introduce SpatiaLab, a comprehensive benchmark for evaluating VLMs' spatial reasoning in realistic, unconstrained contexts. SpatiaLab comprises 1,400 visual question-answer pairs across six major categories: Relative Positioning, Depth & Occlusion, Orientation, Size & Scale, Spatial Navigation, and 3D Geometry, each with five subcategories, yielding 30 distinct task types. Each subcategory contains at least 25 questions, and each main category includes at least 200 questions, supporting both multiple-choice and open-ended evaluation. Experiments across diverse state-of-the-art VLMs, including open- and closed-source models, reasoning-focused, and specialized spatial reasoning models, reveal a substantial gap in spatial reasoning capabilities compared with humans. In the multiple-choice setup, InternVL3.5-72B achieves 54.93% accuracy versus 87.57% for humans. In the open-ended setting, all models show a performance drop of around 10-25%, with GPT-5-mini scoring highest at 40.93% versus 64.93% for humans. These results highlight key limitations in handling complex spatial relationships, depth perception, navigation, and 3D geometry. By providing a diverse, real-world evaluation framework, SpatiaLab exposes critical challenges and opportunities for advancing VLMs' spatial reasoning, offering a benchmark to guide future research toward robust, human-aligned spatial understanding. SpatiaLab is available at: https://spatialab-reasoning.github.io/.
WildRoadBench: A Wild Aerial Road-Damage Grounding Benchmark for Vision-Language Models and Autonomous Agents
University of Electronic Science and Technology of China(电子科技大学);National University of Singapore(新加坡国立大学);De Artificial Intelligence Lab(德人工智能实验室);The Chinese University of Hong Kong, Shenzhen(香港中文大学(深圳));University of Science and Technology of China(中国科学技术大学)
Measuring the Unspoken: A Disentanglement Model and Benchmark for Psychological Analysis in the Wild
College of Computer Science, National University of Defense Technology(计算机科学学院,国防科技大学);Shien-Ming Wu School of Intelligent Engineering, South China University of Technology(智能工程学院,华南理工大学)
InteractVLM: 3D Interaction Reasoning from 2D Foundational Models
Max Planck Institute for Intelligent Systems(马克斯·普朗克智能系统研究所);University of Amsterdam (UvA)(阿姆斯特丹大学);Inria(法国国家信息与自动化研究所);École normale supérieure(巴黎高等师范学院);CNRS(法国国家科学研究中心);PSL Research University(巴黎文理研究大学)
GeoArena: Evaluating Open-World Geographic Reasoning in Large Vision-Language Models
Department of Data Science, City University of Hong Kong(香港城市大学数据科学系);Department of Computer Sciences, University of Wisconsin-Madison(威斯康星大学麦迪逊分校计算机科学系)
RAGNet: Large-scale Reasoning-based Affordance Segmentation Benchmark towards General Grasping
The Chinese University of Hong Kong(香港中文大学);Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所);Dexmal;Mohamed bin Zayed University of Artificial Intelligence(穆罕默德·本·扎耶德人工智能大学);SKL-IOTSC, CIS, University of Macau(澳门科学馆-物联网与智能系统研究中心,澳门大学)
ChinaHeritaQA: A Culturally-Grounded Visual Question Answering Dataset for World Heritage Sites in China
LMU Munich(慕尼黑大学);FAU Erlangen-Nuremberg(埃尔朗根-纽伦堡大学);Munich Center for Machine Learning(慕尼黑机器学习中心);University of Tübingen & Tübingen AI Center(图宾根大学与图宾根人工智能中心);Sun Yat-sen University(中山大学);University of Copenhagen(哥本哈根大学);University of Maryland, College Park(马里兰大学帕克分校)
From Segments to Scenes: Temporal Understanding for Agentic Autonomous Driving via Vision-Language Models
University of California, Berkeley(加州大学伯克利分校);University of Cambridge(剑桥大学);University of Toronto(多伦多大学);ETH Zurich(苏黎世联邦理工学院);University of Washington(华盛顿大学);University of Southern California(南加州大学)
DisasterVQA: A Visual Question Answering Benchmark Dataset for Disaster Scenes
Qatar Computing Research Institute(卡塔尔计算研究所);Hamad Bin Khalifa University(哈马德·本·卡伊夫大学);College of Science & Engineering(科学与工程学院);Qatar University(卡塔尔大学)
Generative Human-Object Interaction Detection via Differentiable Cognitive Steering of Multi-modal LLMs
Institute of Image Communication and Network Engineering, Shanghai Jiao Tong University(上海交通大学图像通信与网络工程研究所);Xi’an Jiao Tong University(西安交通大学);Shandong University(山东大学);Tianjin University(天津大学)
Jet Propulsion Laboratory(喷气推进实验室);California Institute of Technology(加州理工学院);Swiss Federal Institute of Technology(瑞士联邦理工学院);ETH Zürich(苏黎世联邦理工学院);FieldAI Inc.(FieldAI公司);Stanford University(斯坦福大学);University of California Berkeley(加州大学伯克利分校)
Reconstructing Close Human Interaction with Appearance and Proxemics Reasoning
Southeast University(东南大学);National University of Singapore(新加坡国立大学);Sichuan University(四川大学);Agency for Science, Technology and Research, Singapore(新加坡科技研究局);CFAR, Agency for Science, Technology and Research, Singapore(新加坡科技研究局)
Unraveling the geometry of visual relational reasoning
Program in Neuroscience, Harvard Medical School(神经科学项目,哈佛医学院);Boston Children’s Hospital, Harvard Medical School(波士顿儿童医院,哈佛医学院);Center for Brains, Minds, and Machines(大脑、心智与机器中心);Edmond and Lily Safra Center for Brain Sciences, Hebrew University(埃德蒙与莉莉·萨弗脑科学中心,希伯来大学)