Research map / Point clouds / Segmentation and detection
Query Refinement Transformer for 3D Instance Segmentation
Jiahao Lu, Jiacheng Deng, Chuxin Wang, Jianfeng He, Tianzhu Zhang
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 Query Initialization Module (ablation) | Mask3D baseline · ScanNetV2 val | mAP: 53.7 -> 55.3 | Improved |
| add cluster-based layer (ablation) | Mask3D baseline · ScanNetV2 val | mAP: 53.7 -> 54.0 | No clear effect |
| add denoising module (ablation) | QIM + cluster-based layer · ScanNetV2 val | mAP: 55.4 -> 56.2 | Improved |
| add DBSCAN post-processing (ablation) | QueryFormer · ScanNetV2 val | mAP: 56.2 -> 56.5 | No clear effect |
| add DBSCAN post-processing | Mask3D · S3DIS Area 5 | mAP: 53.9 -> 56.5 | Improved |
| increase number of queries from 100 to 500 (ablation) | Mask3D · ScanNetV2 val | mAP: 53.7 -> 52.8 | Hurt |
| increase number of queries from 100 to 150 (ablation) | QueryFormer · ScanNetV2 val | mAP: 56.2 -> 55.3 | Hurt |
| replace FPS query sampling with QIM at 100 queries | Mask3D · ScanNetV2 val | coverage rate / repetition rate: CR 0.74 -> 0.89; RR 0.64 -> 0.47 | Improved |
| use cluster-based layer instead of the normal decoder layer | single decoder layer · ScanNetV2 val | FLOPs (G): 2.3 -> 1.3 | Improved |
Closest papers by meaning
- CompetitorFormer: Mitigating Query Conflicts for 3D Instance Segmentation via Competitive Strategy (CVPR 2026 · similarity 0.91)
- OneFormer3D: One Transformer for Unified Point Cloud Segmentation (CVPR 2024 · similarity 0.88)
- LQMFormer: Language-aware Query Mask Transformer for Referring Image Segmentation (CVPR 2024 · similarity 0.86)
- Open3DIS: Open-Vocabulary 3D Instance Segmentation with 2D Mask Guidance (CVPR 2024 · similarity 0.86)
- Insightful Instance Features for 3D Instance Segmentation (CVPR 2025 · similarity 0.86)
- Any3DIS: Class-Agnostic 3D Instance Segmentation by 2D Mask Tracking (CVPR 2025 · 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.
