AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Make Geometry Matter for Spatial Reasoning

Shihua Zhang, Qiuhong Shen, Shizun Wang, Tianbo Pan, Xinchao Wang

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Spatial reasoning in VLMs · 7 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use Geometry-Guided Fusion instead of original fusion (ablation)GeoSR · VSI-BenchAvg accuracy: 50.0 -> 51.9Improved
use Geometry-Guided Fusion instead of original fusion (ablation)GeoSR · DSR-BenchAvg accuracy: 64.7 -> 66.1Improved
add Geometry-Unleashing Masking to Geo. Fus. (ablation)GeoSR · DSR-BenchAvg accuracy: 64.9 -> 66.1Improved
add Geometry-Unleashing Masking to Geo. Fus. (ablation)GeoSR · VSI-BenchAvg accuracy: 50.9 -> 51.9Improved
masking without any fusion module (ablation)GeoSR · DSR-BenchAvg accuracy: 66.1 -> 58.1Hurt
naive geometry injection (original fusion) vs no geometry (ablation)Qwen2.5-VL-7B · DSR-BenchAvg accuracy: 64.0 -> 62.8Hurt
naive geometry injection (original fusion) vs no geometry (ablation)Qwen2.5-VL-7B · VSI-BenchAvg accuracy: 49.8 -> 50.2No clear effect
full GeoSR vs no geometry modules (ablation)Qwen2.5-VL-7B · DSR-BenchAvg accuracy: 64.0 -> 66.1Improved
full GeoSR vs no geometry modules (ablation)Qwen2.5-VL-7B · VSI-BenchAvg accuracy: 49.8 -> 51.9Improved
lower masking ratio gamma (0.8 to 0.4) (ablation)GeoSR · DSR-BenchAvg accuracy: 66.1 -> 65.0Hurt
higher masking probability beta (0.5 to 0.7) (ablation)GeoSR · DSR-BenchAvg accuracy: 66.1 -> 64.5Hurt
lower masking probability beta (0.5 to 0.3) (ablation)GeoSR · DSR-BenchAvg accuracy: 66.1 -> 64.9Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy