Research map / Point clouds / Point cloud classification
Shape Self-Correction for Unsupervised Point Cloud Understanding
Ye Chen, Jinxian Liu, Bingbing Ni, Hang Wang, Jiancheng Yang, Ning Liu, Teng Li, Qi Tian
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 template (original shape) features (ablation) | Shape Self-Correction · ModelNet40 | accuracy: 87.8 -> 92.4 | Improved |
| distort with exchange, replace, rotate, translate, scale instead of noise only (ablation) | Shape Self-Correction · ModelNet40 | accuracy: 87.2 -> 92.4 | Improved |
| pre-train with Shape Self-Correction before fine-tuning | PointNet · ShapeNetPart 1% data | Ins. mIoU: 68.6 -> 72.9 | Improved |
| pre-train with Shape Self-Correction before fine-tuning | RSCNN · ShapeNetPart 1% data | Ins. mIoU: 71.6 -> 74.3 | Improved |
| pre-train with Shape Self-Correction before fine-tuning | RSCNN · ShapeNetPart 100% data | Ins. mIoU: 84.3 -> 85.2 | Improved |
| pre-train with Shape Self-Correction before fine-tuning | RSCNN · ModelNet40 | accuracy: 91.7 -> 93.0 | Improved |
| pre-train with Shape Self-Correction before fine-tuning | RSCNN · ScanNet | accuracy: 90.1 -> 92.9 | Improved |
| pre-train with Shape Self-Correction before fine-tuning | PointNet · ScanNet | accuracy: 87.8 -> 89.7 | Improved |
| pre-train with PointGLR instead of random init | RSCNN · ModelNet10 | accuracy: 94.8 -> 94.8 | No clear effect |
| use RSCNN instead of PointNet as encoder | Shape Self-Correction · ModelNet40 (unsupervised transfer) | accuracy: 89.9 -> 92.4 | Improved |
| increase number of incorrect points K beyond 256 (ablation) | Shape Self-Correction · ModelNet40 | accuracy | No clear effect |
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.
- FoldingNet — FoldingNet: Point Cloud Auto-Encoder via Deep Grid Deformation (CVPR 2018) · 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
- Unsupervised Multi-Task Feature Learning on Point Clouds (ICCV 2019 · similarity 0.90)
- Learning by Restoring Broken 3D Geometry (TPAMI 2023 · similarity 0.89)
- Mapping in a Cycle: Sinkhorn Regularized Unsupervised Learning for Point Cloud Shapes (ECCV 2020 · similarity 0.89)
- Unsupervised 3D Learning for Shape Analysis via Multiresolution Instance Discrimination (AAAI 2021 · similarity 0.89)
- FreePoint: Unsupervised Point Cloud Instance Segmentation (CVPR 2024 · similarity 0.88)
- SPU-IMR: Self-supervised Arbitrary-scale Point Cloud Upsampling via Iterative Mask-recovery Network (AAAI 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.
