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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add SEER-Map BEV topology | GPT-5.2 · SEERBench | average accuracy: 38.2 -> 51.1 | Improved |
| add SEER-Map BEV topology | Gemini-3-Flash · SEERBench | average accuracy: 42.2 -> 52.8 | Improved |
| add SEER-Map BEV topology | Qwen3-VL-Plus · SEERBench | average accuracy: 35.8 -> 43.9 | Improved |
| add SEER-Map BEV topology | Qwen3-VL-8B-Instruct · SEERBench | average accuracy: 32.5 -> 40.1 | Improved |
| add SEER-Map BEV topology | SenseNova-SI-8B · SEERBench | average accuracy: 27.5 -> 36.1 | Improved |
| add SEER-Map to boost camera layout inference | GPT-5.2 · SEERBench Camera Layout Inference | accuracy: 39.1 -> 82.2 | Improved |
| add SEER-Map on robust spatial logic | Qwen3-VL-8B-Instruct · SEERBench Robust Spatial Logic | accuracy: 21.8 -> 12.7 | Hurt |
| add SEER-Map on active viewpoint selection | Gemini-3-Flash · SEERBench Active Viewpoint Selection | accuracy: 60.6 -> 59.2 | Hurt |
| add SEER-Map on robust spatial logic | GPT-5.2 · SEERBench Robust Spatial Logic | accuracy: 32.7 -> 32.7 | No clear effect |
| render geometry as visual BEV instead of numeric text (global prior) (ablation) | Gemini-3-Flash · SEERBench | average accuracy: 44.6 -> 52.3 (+7.7) | Improved |
| render geometry as visual BEV instead of numeric text (global prior) (ablation) | Qwen3-VL-8B-Instruct · SEERBench | average accuracy: 33.1 -> 40.1 (+7.0) | Improved |
| render geometry as visual BEV instead of numeric text (global prior) (ablation) | SenseNova-SI-8B · SEERBench | average accuracy: 28.4 -> 36.1 (+7.7) | Improved |
Closest papers by meaning
- EgoProx: Evaluating MLLMs on Egocentric 3D Proximity Reasoning Across a Cognitive Hierarchy (CVPR 2026 · similarity 0.92)
- EgoMind: Activating Spatial Cognition through Linguistic Reasoning in MLLMs (CVPR 2026 · similarity 0.91)
- Spatial Reasoning with Vision-Language Models in Ego-Centric Multi-View Scenes (ICLR 2026 · similarity 0.91)
- From Where Things Are to What They Are For: Benchmarking Spatial-Functional Intelligence in Multimodal LLMs (CVPR 2026 · similarity 0.91)
- ERGeoBench: A Comprehensive Benchmark for Embodied Reasoning and Geo-localization in Multimodal Large Language Models (ICML 2026 · similarity 0.90)
- Unfolding Spatial Cognition: Evaluating Multimodal Models on Visual Simulations (ICLR 2026 · similarity 0.90)
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.
