AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

SEERBench: A Spatial Ego-Exo Reasoning Benchmark for MLLMs with a Simple Yet Effective Baseline

Fengyuan Lu, Jiahe Feng, Zhengyang Zhou, Shaofeng Zhang, Wenbin Li, Qi Fan, Yang Gao

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Spatial reasoning in VLMs

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
add SEER-Map BEV topologyGPT-5.2 · SEERBenchaverage accuracy: 38.2 -> 51.1Improved
add SEER-Map BEV topologyGemini-3-Flash · SEERBenchaverage accuracy: 42.2 -> 52.8Improved
add SEER-Map BEV topologyQwen3-VL-Plus · SEERBenchaverage accuracy: 35.8 -> 43.9Improved
add SEER-Map BEV topologyQwen3-VL-8B-Instruct · SEERBenchaverage accuracy: 32.5 -> 40.1Improved
add SEER-Map BEV topologySenseNova-SI-8B · SEERBenchaverage accuracy: 27.5 -> 36.1Improved
add SEER-Map to boost camera layout inferenceGPT-5.2 · SEERBench Camera Layout Inferenceaccuracy: 39.1 -> 82.2Improved
add SEER-Map on robust spatial logicQwen3-VL-8B-Instruct · SEERBench Robust Spatial Logicaccuracy: 21.8 -> 12.7Hurt
add SEER-Map on active viewpoint selectionGemini-3-Flash · SEERBench Active Viewpoint Selectionaccuracy: 60.6 -> 59.2Hurt
add SEER-Map on robust spatial logicGPT-5.2 · SEERBench Robust Spatial Logicaccuracy: 32.7 -> 32.7No clear effect
render geometry as visual BEV instead of numeric text (global prior) (ablation)Gemini-3-Flash · SEERBenchaverage accuracy: 44.6 -> 52.3 (+7.7)Improved
render geometry as visual BEV instead of numeric text (global prior) (ablation)Qwen3-VL-8B-Instruct · SEERBenchaverage accuracy: 33.1 -> 40.1 (+7.0)Improved
render geometry as visual BEV instead of numeric text (global prior) (ablation)SenseNova-SI-8B · SEERBenchaverage accuracy: 28.4 -> 36.1 (+7.7)Improved

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