Research map / Point clouds / Segmentation and detection
MR-COSMO: Visual-Text Memory Recall and Direct CrOSs-MOdal Alignment Method for Query-Driven 3D Segmentation
Chade Li, Pengju Zhang, Yihong Wu
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 Memory Module for Feature Pairs (ablation) | MR-COSMO · Instruct3D | Acc: 31.9 -> 33.8 | Improved |
| add Memory Module for Feature Pairs (ablation) | MR-COSMO · Instruct3D | mIoU: 27.4 -> 28.5 | Improved |
| add Direct Cross-Modal Alignment to baseline (ablation) | MR-COSMO · Instruct3D | mIoU: 26.4 -> 27.4 | Improved |
| add Memory Module to baseline alone (ablation) | MR-COSMO · Instruct3D | mIoU: 26.4 -> 27.5 | Improved |
| combine DCMA and Memory Module (ablation) | MR-COSMO · Instruct3D | mIoU: 26.4 -> 28.5 | Improved |
| add MLP point features to voxel encoder (ablation) | MR-COSMO · Instruct3D | mIoU: 27.7 -> 28.5 | Improved |
| use image-voxel and text-point matching instead of reversed (ablation) | MR-COSMO · Instruct3D | mIoU: 27.9 -> 28.5 | Improved |
| add Alignment Constrains Block (contrastive) (ablation) | MR-COSMO · Instruct3D | mIoU: 28.0 -> 28.5 | Improved |
| add loss on bounding boxes (ablation) | MR-COSMO · Instruct3D | mIoU: 28.4 -> 28.5 | No clear effect |
| replace Mamba with Transformer (ablation) | MR-COSMO · Instruct3D | mIoU: 28.5 -> 28.4 | Hurt |
| replace ResNet50 with ViT (ablation) | MR-COSMO · Instruct3D | mIoU: 28.5 -> 28.7 | Mixed |
| scale LLaMA2 from 7B to 13B (ablation) | MR-COSMO · Instruct3D | mIoU: 28.5 -> 28.6 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports MR-COSMO as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- MORE-STEM: Long-Short MemOry REcall and Spatio-TEmporal Consistency Model for Query-Driven 3D/4D Point Cloud Segmentation (CVPR 2026 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SegPoint — SegPoint: Segment Any Point Cloud via Large Language Model (ECCV 2024) · better on 3 of 3 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
- LESS: Label-Efficient and Single-Stage Referring 3D Segmentation (NeurIPS 2024 · similarity 0.88)
- More Text, Less Point: Towards 3D Data-Efficient Point-Language Understanding (AAAI 2025 · similarity 0.87)
- COS3D: Collaborative Open-Vocabulary 3D Segmentation (NeurIPS 2025 · similarity 0.86)
- Visual-Text Cross Alignment: Refining the Similarity Score in Vision-Language Models (ICML 2024 · similarity 0.86)
- Tango3D: Towards Alignment for Global and Local 2D-3D Correspondence (NeurIPS 2026 · similarity 0.86)
- XMask3D: Cross-modal Mask Reasoning for Open Vocabulary 3D Semantic Segmentation (NeurIPS 2024 · similarity 0.86)
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.
