Research map / Point clouds / Point cloud classification
Self-Supervised Few-Shot Learning on 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 |
|---|---|---|---|
| add cover-tree self-supervised pre-training | DGCNN · ModelNet40 10-way 10-shot | accuracy (%): 19.85 -> 48.50 | Helped |
| add cover-tree self-supervised pre-training | DGCNN · Sydney 10-way 20-shot | accuracy (%): 76.10 -> 81.50 | Helped |
| add cover-tree self-supervised pre-training | PointNet · ModelNet40 10-way 10-shot | accuracy (%): 46.60 -> 49.15 | Helped |
| add cover-tree self-supervised pre-training | PointNet · Sydney 10-way 20-shot | accuracy (%): 58.30 -> 64.0 | Helped |
| add cover-tree self-supervised pre-training | PointNet · ShapeNet part segmentation | mean mIoU (%): 27.23 -> 36.81 | Helped |
| add cover-tree pre-training | DGCNN · S3DIS 5-way 10-shot | mIoU (%): 61.36 -> 62.63 | Helped |
| pretrain with cover-tree tasks instead of VoxelSSL | DGCNN · ModelNet40 10-way 10-shot | accuracy (%): 26.05 -> 48.50 | Helped |
| add regression task R to classification task C (ablation) | DGCNN pretraining · Sydney 10-way 10-shot | accuracy (%): 60.15 -> 66.15 | Helped |
| add classification task C to regression task R (ablation) | DGCNN pretraining · ModelNet40 10-way 10-shot | accuracy (%): 47.0 -> 48.50 | Helped |
| add only classification task C (vs none) (ablation) | DGCNN pretraining · ModelNet40 10-way 10-shot | accuracy (%): 19.85 -> 46.5 | Helped |
| use only regression R instead of only classification C (ablation) | DGCNN pretraining · Sydney 10-way 20-shot | accuracy (%): 76.80 -> 80.60 | Helped |
Closest papers by meaning
- Self-Supervised Deep Learning on Point Clouds by Reconstructing Space (NeurIPS 2019 · similarity 0.91)
- Sonata: Self-Supervised Learning of Reliable Point Representations (CVPR 2025 · similarity 0.87)
- A Unified Approach to Interpreting Self-supervised Pre-training Methods for 3D Point Clouds via Interactions (CVPR 2025 · similarity 0.86)
- FreePoint: Unsupervised Point Cloud Instance Segmentation (CVPR 2024 · similarity 0.86)
- Self-Supervised Class-Agnostic Motion Prediction with Spatial and Temporal Consistency Regularizations (CVPR 2024 · similarity 0.86)
- Mitigating Object Dependencies: Improving Point Cloud Self-Supervised Learning through Object Exchange (CVPR 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.