Research map / Point clouds / Point cloud classification
Enhancing Spatial Reasoning in Multimodal Large Language Models through Reasoning-based Segmentation
Zhenhua Ning, Zhuotao Tian, Shaoshuai Shi, Guangming Lu, Daojing He, Wenjie Pei, Li Jiang
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 R2S (reasoning prior learning + prior-guided refinement) | baseline + 3D ReasonSeg · ScanRefer | Acc@50: 43.6 -> 48.7 | Improved |
| add R2S | baseline + 3D ReasonSeg · 3D ReasonSeg | gIoU: 29.2 -> 33.1 | Improved |
| add R2S | baseline + 3D ReasonSeg · ScanQA | CIDEr: 75.2 -> 77.3 | Improved |
| add 3D ReasonSeg to the training data | baseline · ScanRefer | Acc@50: 38.7 -> 43.6 | Improved |
| add 3D ReasonSeg to the training data | baseline · ScanQA | CIDEr: 70.5 -> 75.2 | Improved |
| add prior-guided refinement (ablation) | R2S · ScanRefer | Acc@50: 41.3 -> 48.7 | Improved |
| add prior-guided refinement (ablation) | R2S · ScanQA | CIDEr: 66.1 -> 77.3 | Improved |
| add prior-guided refinement (ablation) | R2S · 3D ReasonSeg | gIoU: 27.5 -> 33.1 | Improved |
| use segmentation-mask instance features instead of text-only object hints (ablation) | R2S · ScanRefer | Acc@50: 43.5 -> 48.7 | Improved |
| use segmentation-mask instance features instead of text-only object hints (ablation) | R2S · 3D ReasonSeg | gIoU: 29.8 -> 33.1 | Improved |
| increase Q-Former latent queries from 16 to 32 (ablation) | baseline + 3D ReasonSeg · ScanRefer | Acc@50: 40.5 -> 43.6 | Improved |
| increase Q-Former latent queries from 32 to 48 (ablation) | baseline + 3D ReasonSeg · ScanRefer | Acc@50: 43.6 -> 41.6 | Hurt |
Closest papers by meaning
- SpatialRGPT: Grounded Spatial Reasoning in Vision-Language Models (NeurIPS 2024 · similarity 0.91)
- Actial: Activate Spatial Reasoning Ability of Multimodal Large Language Models (NeurIPS 2025 · similarity 0.91)
- MLLM-For3D: Adapting Multimodal Large Language Model for 3D Reasoning Segmentation (NeurIPS 2025 · similarity 0.91)
- SpatialReasoner: Towards Explicit and Generalizable 3D Spatial Reasoning (NeurIPS 2025 · similarity 0.90)
- 3ViewSense: Spatial and Mental Perspective Reasoning from Orthographic Views in Vision-Language Models (ICML 2026 · similarity 0.90)
- From Flatland to Space: Teaching Vision-Language Models to Perceive and Reason in 3D (NeurIPS 2025 · 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.
