Paper Detail
Shihua Zhang, Qiuhong Shen, Shizun Wang, Tianbo Pan, Xinchao Wang
Empowered by large-scale training, vision-language models (VLMs) achieve strong image and video understanding, yet their ability to perform spatial reasoning in both static scenes and dynamic videos remains limited. Recent advances try to handle this limitation by injecting geometry tokens from pretrained 3D foundation models into VLMs. Nevertheless, we observe that naive token fusion followed by standard fine-tuning in this line of work often leaves such geometric cues underutilized for spatial reasoning, as VLMs tend to rely heavily on 2D visual cues. In this paper, we propose GeoSR, a framework designed to make geometry matter by encouraging VLMs to actively reason with geometry tokens. GeoSR introduces two key components: (1) Geometry-Unleashing Masking, which strategically masks portions of 2D vision tokens during training to weaken non-geometric shortcuts and force the model to consult geometry tokens for spatial reasoning; and (2) Geometry-Guided Fusion, a gated routing mechanism that adaptively amplifies geometry token contributions in regions where geometric evidence is critical. Together, these designs unleash the potential of geometry tokens for spatial reasoning tasks. Extensive experiments on both static and dynamic spatial reasoning benchmarks demonstrate that GeoSR consistently outperforms prior methods and establishes new state-of-the-art performance by effectively leveraging geometric information. The project page is available at https://suhzhang.github.io/GeoSR/.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{zhang2026make,
title = {Make Geometry Matter for Spatial Reasoning},
author = {Shihua Zhang and Qiuhong Shen and Shizun Wang and Tianbo Pan and Xinchao Wang},
year = {2026},
abstract = {Empowered by large-scale training, vision-language models (VLMs) achieve strong image and video understanding, yet their ability to perform spatial reasoning in both static scenes and dynamic videos remains limited. Recent advances try to handle this limitation by injecting geometry tokens from pretrained 3D foundation models into VLMs. Nevertheless, we observe that naive token fusion followed by standard fine-tuning in this line of work often leaves such geometric cues underutilized for spatial},
url = {https://huggingface.co/papers/2603.26639},
keywords = {vision-language models, spatial reasoning, geometry tokens, pretrained 3D foundation models, Geometry-Unleashing Masking, Geometry-Guided Fusion, gated routing mechanism, 2D visual cues, geometric shortcuts, code available, huggingface daily},
eprint = {2603.26639},
archiveprefix = {arXiv},
}
{}