Research map / Point clouds / Segmentation and detection
Exploiting Local and Global Structure for Point Cloud Semantic Segmentation with Contextual Point Representations
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 |
|---|---|---|---|
| add contextual representation (gated fusion of point and neighbours) (ablation) | Ours · S3DIS Area 5 | mIoU: 56.15 -> 60.06 | Helped |
| add graph pointnet module (ablation) | Ours · S3DIS Area 5 | mIoU: 57.84 -> 60.06 | Helped |
| add spatial and channel attention module (ablation) | Ours · S3DIS Area 5 | mIoU: 58.67 -> 60.06 | Helped |
| use gated fusion instead of concatenation in contextual representation (ablation) | Ours · S3DIS Area 5 | mIoU: 59.14 -> 60.06 | Helped |
| add CR, GPM and AM to DGCNN (ablation) | DGCNN · S3DIS | OA: 84.31 -> 86.07 | Helped |
| add CR alone to DGCNN (ablation) | DGCNN · S3DIS | OA: 84.31 -> 85.35 | Helped |
| add GPM alone to DGCNN (ablation) | DGCNN · S3DIS | OA: 84.31 -> 84.90 | Helped |
| add AM alone to DGCNN (ablation) | DGCNN · S3DIS | OA: 84.31 -> 85.17 | Helped |
| use proposed model instead of PointNet++ under 50% scaling | Ours · S3DIS robustness | OA drop: 3.0% vs 4.5% decrease | Helped |
| use proposed model instead of PointNet++ under rotation pi/10 | Ours · S3DIS robustness | OA drop: 1.7% vs 1.0% decrease | Hurt |
Closest papers by meaning
- Mining Point Cloud Local Structures by Kernel Correlation and Graph Pooling (CVPR 2018 · similarity 0.90)
- Large-Scale Point Cloud Semantic Segmentation With Superpoint Graphs (CVPR 2018 · similarity 0.90)
- Weakly Supervised Point Cloud Semantic Segmentation via Artificial Oracle (CVPR 2024 · similarity 0.89)
- Taylor Series-Inspired Local Structure Fitting Network for Few-shot Point Cloud Semantic Segmentation (AAAI 2025 · similarity 0.88)
- Geometrically-driven Aggregation for Zero-shot 3D Point Cloud Understanding (CVPR 2024 · similarity 0.88)
- Construct to Associate: Cooperative Context Learning for Domain Adaptive Point Cloud Segmentation (CVPR 2024 · similarity 0.88)
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.