Research map / Point clouds / Point cloud classification
Self-Supervised Deep Learning on Point Clouds by Reconstructing Space
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 |
|---|---|---|---|
| Pre-train DGCNN by reconstructing randomly rearranged point-cloud parts | DGCNN · ModelNet40 classification | accuracy: 92.2% -> 92.4% | Helped |
| Pre-train DGCNN by reconstructing randomly rearranged point-cloud parts | DGCNN embeddings with linear SVM · ModelNet40 | accuracy: 90.64% vs 90.19% | Helped |
| Pre-train DGCNN by reconstructing randomly rearranged point-cloud parts | DGCNN embeddings with linear SVM · ModelNet10 | accuracy: 94.52% vs 95.30% | Mixed |
| Pre-train DGCNN before supervised segmentation | DGCNN · ShapeNet Part | mean IoU: 85.1 -> 85.3 | Helped |
| Pre-train DGCNN on unlabelled S3DIS areas | DGCNN · S3DIS area 6 test, supervised train area 3 | mIoU%: 39.9% -> 42.4% | Helped |
| Pre-train DGCNN on unlabelled S3DIS areas | DGCNN · S3DIS area 6 test, supervised train area 6 | mIoU%: 43.9% -> 43.9% | No clear effect |
| Use a 3-by-3-by-3 voxel grid rather than k=2 or k=4 (ablation) | Point-cloud self-supervised pre-training · Evaluated experiments | downstream performance | Helped |
| Add random rotations of 15% of voxels and replace one voxel with a voxel from another input (ablation) | Point-cloud self-supervised pre-training · Object classification | SVM accuracy: around 0.2% | Helped |
Closest papers by meaning
- Self-Supervised Few-Shot Learning on Point Clouds (NeurIPS 2020 · similarity 0.91)
- Self-Supervised Global-Local Structure Modeling for Point Cloud Domain Adaptation With Reliable Voted Pseudo Labels (CVPR 2022 · similarity 0.90)
- Sonata: Self-Supervised Learning of Reliable Point Representations (CVPR 2025 · similarity 0.88)
- Mitigating Object Dependencies: Improving Point Cloud Self-Supervised Learning through Object Exchange (CVPR 2024 · similarity 0.87)
- FreePoint: Unsupervised Point Cloud Instance Segmentation (CVPR 2024 · similarity 0.87)
- DeepLA-Net: Very Deep Local Aggregation Networks for Point Cloud Analysis (CVPR 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.