Research map / Point clouds / Point cloud classification
Clustering based Point Cloud Representation Learning for 3D Analysis
Tuo Feng, Wenguan Wang, Xiaohan Wang, Yi Yang, Qinghua Zheng
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 point-point contrast (ablation) | KPConv · S3DIS Area-5 | mIoU: 67.1 -> 68.0 | Improved |
| add point-center contrast (ablation) | KPConv · S3DIS Area-5 | mIoU: 67.1 -> 68.4 | Improved |
| combine point-point and point-center contrast (ablation) | KPConv · S3DIS Area-5 | mIoU: 67.1 -> 69.0 | Improved |
| combine point-point and point-center contrast (ablation) | KPConv · SemanticKITTI multi-scan val | mIoU: 53.3 -> 55.7 | Improved |
| add point-point contrast (ablation) | KPConv · SemanticKITTI multi-scan val | mIoU: 53.3 -> 54.4 | Improved |
| add point-center contrast (ablation) | KPConv · SemanticKITTI multi-scan val | mIoU: 53.3 -> 54.7 | Improved |
| add clustering-based training (training cost) (ablation) | KPConv · S3DIS Area-5 | training time per epoch: 281.46 -> 311.71 | Hurt |
| split each class into M=40 sub-clusters instead of M=1 (ablation) | KPConv · S3DIS Area-5 | mIoU: 67.5 -> 69.0 | Improved |
| split each class into M=40 sub-clusters instead of M=1 (ablation) | KPConv · SemanticKITTI multi-scan val | mIoU: 53.7 -> 55.7 | Improved |
| increase cluster number M above 40 (ablation) | KPConv · S3DIS Area-5 | mIoU | No clear effect |
| add class-wise memory bank (10 points per scene per cluster) over mini-batch-only contrast (ablation) | KPConv · S3DIS Area-5 | mIoU: 68.0 -> 69.0 | Improved |
| add class-wise memory bank (10 points per scene per cluster) over mini-batch-only contrast (ablation) | KPConv · SemanticKITTI multi-scan val | mIoU: 54.4 -> 55.7 | Improved |
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.
- RPVNet — RPVNet: A Deep and Efficient Range-Point-Voxel Fusion Network for LiDAR Point Cloud Segmentation (ICCV 2021) · better on 1 of 1 shared result
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
- Clustering-based Supervised Training for 3D Perception (TPAMI 2026 · similarity 0.95)
- X-3D: Explicit 3D Structure Modeling for Point Cloud Recognition (CVPR 2024 · similarity 0.90)
- PointGLR: Unsupervised Structural Representation Learning of 3D Point Clouds (TPAMI 2022 · similarity 0.90)
- Unsupervised Point Cloud Representation Learning by Clustering and Neural Rendering (IJCV 2024 · similarity 0.90)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.89)
- FreePoint: Unsupervised Point Cloud Instance 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.
