Research map / Point clouds / Point cloud classification
Learning Generalizable Part-based Feature Representation for 3D Point Clouds
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 |
|---|---|---|---|
| replace optimal transport with cross-attention for part alignment (ablation) | PDG (PointNet) · ModelNet to ScanObjectNN M->SO (PointNet backbone, last five) | accuracy (%): 65.2 -> 67.6 | Helped |
| replace max-pooling with confidence-based part aggregation (ablation) | PDG (PointNet) · ModelNet to ScanObjectNN M->SO (PointNet backbone, last five) | accuracy (%): 65.3 -> 67.6 | Helped |
| replace average-pooling with confidence-based part aggregation (ablation) | PDG (PointNet) · ModelNet to ScanObjectNN M->SO (PointNet backbone, last five) | accuracy (%): 65.4 -> 67.6 | Helped |
| add contrastive learning framework (ablation) | PDG (PointNet) · ModelNet to ScanObjectNN M->SO (PointNet backbone, last five) | accuracy (%): 66.4 -> 67.6 | Helped |
| add shape-level contrastive loss (ablation) | PDG (PointNet) · ModelNet to ScanObjectNN M->SO (PointNet backbone, last five) | accuracy (%): 67.1 -> 67.6 | Helped |
| use 8 parts of 512 points instead of 16 parts of 256 points (ablation) | PDG (PointNet) · ModelNet to ScanObjectNN M->SO (PointNet backbone, last five) | accuracy (%): 65.4 -> 67.6 | Helped |
| add part-based representation to PointNet baseline (ablation) | PointNet · ModelNet to ScanObjectNN M->SO (PointNet backbone, last five) | accuracy (%): 59.8 -> 67.6 | Helped |
| PDG with PointNet backbone vs plain PointNet | PDG (P) · Real-to-Sim average | accuracy (%): 52.9 -> 57.1 | Helped |
| PDG with DGCNN backbone vs plain DGCNN | PDG (D) · Real-to-Sim average | accuracy (%): 53.0 -> 59.2 | Helped |
| PDG vs MetaSets with PointNet backbone | PDG (P) · Real-to-Sim average | accuracy (%): 50.3 -> 57.1 | Helped |
| PDG vs MetaSets with DGCNN backbone | PDG (D) · Real-to-Sim average | accuracy (%): 51.0 -> 59.2 | Helped |
| PDG vs MetaSets on PointDA-10 | PDG · PointDA-10 average | accuracy (%): 64.5 -> 67.2 | Helped |
Closest papers by meaning
- MetaSets: Meta-Learning on Point Sets for Generalizable Representations (CVPR 2021 · similarity 0.90)
- Point-PRC: A Prompt Learning Based Regulation Framework for Generalizable Point Cloud Analysis (NeurIPS 2024 · similarity 0.88)
- PointDGMamba: Domain Generalization of Point Cloud Classification via Generalized State Space Model (AAAI 2025 · similarity 0.87)
- Geometrically-driven Aggregation for Zero-shot 3D Point Cloud Understanding (CVPR 2024 · similarity 0.87)
- Multi-View Representation is What You Need for Point-Cloud Pre-Training (ICLR 2024 · similarity 0.87)
- Interpretable3D: An Ad-Hoc Interpretable Classifier for 3D Point Clouds (AAAI 2024 · similarity 0.86)
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.