IVSGround introduces a lightweight view selector that learns to choose the most informative camera views for vision‑language model (VLM) based 3D visual grounding, replacing heuristic view selection. The selector is trained via a two‑stage rejection sampling process that uses feedback from a reasoning VLM to generate supervision signals. Experiments on ScanRefer and NR3D demonstrate that IVSGround consistently improves grounding accuracy over existing zero‑shot pipelines, underscoring the importance of selecting where to look for effective 3D visual grounding.
By Tsung-Chih Chiang, Hsuan-Kung Yang, Jou-Min Liu, Ting-Ru Liu, Chun-Wei Huang, Quan Kong, Chun-Yi Lee
arXiv:2609.18058v1 Announce Type: new
Abstract: Finding the object referred to by language in a partially observed 3D scene is a core capability for embodied agents. Existing approaches either couple...
By Shixiong Xu, Zhiyuan Chen, Song Ding, Rui Luo, Xiaowei Liang, Dongxu Miao, Zhiying Du
arXiv:2610.28018v1 Announce Type: new
Abstract: Examining existing benchmarks for grounding in 3D VLMs reveals a pervasive "semantic leakage" in the evaluation and prompting protocols: a heavy relian...
By Anna-Maria Halacheva, Jan-Nico Zaech, Sombit Dey, Luc Van Gool, Danda Pani Paudel
UrbanGround is a sandbox that tests how well multimodal large language model agents can translate local street‑view perception into reliable action within a physically realistic replica of Hong Kong. The platform offers closed‑loop first‑person interaction and an interactive map, allowing agents to navigate the 3D city and answer spatial questions. The study evaluates agents across three research questions—scene grounding, navigation over increasing distances, and robustness to route changes—revealing that while agents excel at visual recognition and short‑range reasoning, they struggle with sustained goal‑directed behavior and pedestrian‑aware movement.
By Tianjie Ju, Zheng Wu, Yueqing Sun, Yuhan Cui, Bobo Li, Shengqiong Wu, Pengzhou Cheng, Haodong Zhao, Zongru Wu, Xinbei Ma, Doris Zhang, Kunling Li, Mong-Li Lee, Wynne Hsu, Hao Fei, Qi Gu, Gongshen Liu, Zhuosheng Zhang
arXiv:2512. 23020v3 Announce Type: replace-cross Abstract: 3D visual grounding aims to locate objects based on natural language descriptions in 3D scenes.
By Wenyuan Huang, Zhenyu Zhang, Zhao Wang, Zhou Wei, Ting Huang, Fang Zhao, Jian Yang
arXiv:2509.19571v2 Announce Type: replace-cross
Abstract: Designing or learning robot policies that generalize zero-shot across a range of language instructions and objects is a core problem in robot...
By Sacha Morin, Kumaraditya Gupta, Mahtab Sandhu, Charlie Gauthier, Francesco Argenziano, Kirsty Ellis, Liam Paull
arXiv:2506.17545v2 Announce Type: replace
Abstract: Large language models are increasingly being applied to 3D scene understanding. However, existing 3D-aware LLMs largely operate as black boxes: the...
By Junjie Wang, Zhihao Yuan, Shuyi Jiang, Chun-Mei Feng, Yaolun Zhang, Shuguang Cui, Zhen Li, Na Zhao
arXiv:2606. 31148v1 Announce Type: cross Abstract: 3D Visual Grounding (3DVG) aims to localize target objects in 3D scenes given natural language descriptions.
By Duc Cao Dinh, Khai Le-Duc, Florent Draye, Chris Ngo, Terry Jingchen Zhang, Bernhard Sch\"olkopf, Zhijing Jin
AgentVQA is an offline benchmark designed to evaluate vision‑language models on realistic agentic scenarios across five domains: Web Agents, Robotics, Egocentric Videos, Games, and Spatial Planning. It comprises 1,000 carefully selected real‑world examples drawn from 23 source datasets, enabling reliable comparisons without costly online simulations. The benchmark reveals that current models achieve only 58.5% accuracy, with most errors stemming from difficulties in grounding and tracking evolving visual states rather than from task misunderstanding or reasoning.
By Advait Gupta, Sachin Konan, Tom Yin, Gabriel Sarch, Zhuang Liu
SpatialScout is a benchmark that tests vision‑language model agents on active, object‑level 3D mapping. It includes 80 indoor scenes with 3,140 target objects, where agents must control a camera, gather RGB observations, and estimate object centers in a shared coordinate frame. Performance is measured by localization accuracy and target coverage using object‑scale overlap scores, revealing that current VLMs, including Claude Opus 5.5, struggle to build precise metric maps.
By Zhaoxi Zhang, Hao Liang, Linxuan Wu, Yiming Wu, Yifu Yuan, Kun Liang, Chenming Tang, Meiyi Qiang, Wentao Zhang, Weikang Li
arXiv:2608.29483v1 Announce Type: cross
Abstract: Modern Vision-Language Models (VLMs) perform well above the human baseline in image geolocalization, a task critically important in disaster response...
By Arka Mukherjee, Soham Roy, Kartikeya Trivedi, Shreya Ghosh
arXiv:2512.03284v2 Announce Type: replace
Abstract: Spatial reasoning in large-scale 3D environments remains challenging for current vision--language models, which are typically constrained to room-s...
By Hongpei Zheng, Shijie Li, Lin Qian, Zhenghao Li, Qijun Yang, Yanran Li, Hujun Yin