Research map / Point clouds / Segmentation and detection
PointGroup: Dual-Set Point Grouping for 3D Instance Segmentation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| combine clustering on original and shifted coordinates instead of using either set alone (ablation) | PointGroup · ScanNet v2 validation | AP: 0.283 / 0.328 -> 0.348 | Helped |
| combine clustering on original and shifted coordinates instead of using either set alone (ablation) | PointGroup · ScanNet v2 validation | AP50: 0.507 / 0.529 -> 0.569 | Helped |
| set clustering radius to 3cm rather than 2cm, 4cm, or 5cm (ablation) | PointGroup · ScanNet v2 validation | AP: 0.285 / 0.337 / 0.342 -> 0.348 | Helped |
| set clustering radius to 3cm rather than 2cm, 4cm, or 5cm (ablation) | PointGroup · ScanNet v2 validation | AP50: 0.501 / 0.552 / 0.552 -> 0.569 | Helped |
| use ScoreNet cluster-quality scores instead of averaged semantic probability as confidence (ablation) | PointGroup · ScanNet v2 validation | AP: 30.2% -> 34.8% | Helped |
| use ScoreNet cluster-quality scores instead of averaged semantic probability as confidence (ablation) | PointGroup · ScanNet v2 validation | AP50: 51.9% -> 56.9% | Helped |
| use ScoreNet cluster-quality scores instead of averaged semantic probability as confidence (ablation) | PointGroup · ScanNet v2 validation | AP25: 68.9% -> 71.3% | Helped |
Closest papers by meaning
- FreePoint: Unsupervised Point Cloud Instance Segmentation (CVPR 2024 · similarity 0.90)
- Open3DIS: Open-Vocabulary 3D Instance Segmentation with 2D Mask Guidance (CVPR 2024 · similarity 0.88)
- OneFormer3D: One Transformer for Unified Point Cloud Segmentation (CVPR 2024 · similarity 0.88)
- Relation3D : Enhancing Relation Modeling for Point Cloud Instance Segmentation (CVPR 2025 · similarity 0.87)
- SAI3D: Segment Any Instance in 3D Scenes (CVPR 2024 · similarity 0.87)
- Point-SAM: Promptable 3D Segmentation Model for Point Clouds (ICLR 2025 · similarity 0.87)
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.