Research map / Point clouds / Segmentation and detection
SGPN: Similarity Group Proposal Network for 3D Point Cloud Instance Segmentation
Read the paperCheck 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 |
|---|---|---|---|
| use SGPN similarity-based group proposals instead of semantic clustering | S3DIS instance segmentation · S3DIS | mean AP at IoU 0.5: 17.40 -> 36.30 | Helped |
| use SGPN group proposals instead of semantic clustering | S3DIS instance segmentation · S3DIS | mean AP at IoU 0.25: 34.8 -> 52.6 | Helped |
| use SGPN group proposals instead of semantic clustering | S3DIS instance segmentation · S3DIS | mean AP at IoU 0.75: 11.2 -> 18.8 | Helped |
| derive detections from SGPN instance groups rather than PointNet detections | S3DIS 3D detection · S3DIS | mean AP at IoU 0.5: 24.24 -> 30.20 | Helped |
| add the similarity-matrix branch to the semantic segmentation framework | SGPN · S3DIS | mean IoU (%): 49.76 -> 50.37 | Helped |
| add 2D CNN features from RGB images (ablation) | SGPN · NYUV2 instance segmentation | mean AP at IoU 0.5: 26.5 -> 30.5 | Helped |
| add 2D CNN features from RGB images (ablation) | SGPN · NYUV2 3D detection | mean AP: 36.25 -> 41.30 | Helped |
| add the similarity-matrix branch to the PointNet++ semantic segmentation baseline | SGPN · ShapeNet part segmentation | mean IoU (%): 84.6 -> 85.8 | Helped |
| use SGPN group proposals instead of semantic clustering | SGPN · ScanNet(v1) instance segmentation | mean AP at IoU 0.25: 32.5 -> 35.1 | Helped |
| add 2D CNN features from RGB images (ablation) | SGPN · NYUV2 instance segmentation | AP by category: 55.9 -> 56.4; 59.6 -> 50.6 | Mixed |
Closest papers by meaning
- GSPN: Generative Shape Proposal Network for 3D Instance Segmentation in Point Cloud (CVPR 2019 · similarity 0.91)
- 3D-MPA: Multi-Proposal Aggregation for 3D Semantic Instance Segmentation (CVPR 2020 · similarity 0.89)
- PointGroup: Dual-Set Point Grouping for 3D Instance Segmentation (CVPR 2020 · similarity 0.88)
- OneFormer3D: One Transformer for Unified Point Cloud Segmentation (CVPR 2024 · similarity 0.87)
- FreePoint: Unsupervised Point Cloud Instance Segmentation (CVPR 2024 · similarity 0.87)
- Relation3D : Enhancing Relation Modeling for Point Cloud Instance Segmentation (CVPR 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.