Research map / Point clouds / Segmentation and detection
Weakly Supervised Semantic Point Cloud Segmentation: Towards 10x Fewer Labels
Check 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 the combined MIL, Siamese, smoothness, and label-propagation objectives | DGCNN weakly supervised segmentation · ShapeNet 1pt | CatAvg mIoU: 72.2 -> 74.4 | Helped |
| Add the combined proposed objectives and label propagation | DGCNN weakly supervised segmentation · S3DIS Area 5, 10% | CatAvg mIoU: 45.7 -> 48.0 | Helped |
| Add the combined proposed objectives and label propagation | DGCNN weakly supervised segmentation · PartNet level 1, 1pt | CatAvg mIoU: 50.2 -> 54.6 | Helped |
| Add Siamese consistency on augmented point clouds (ablation) | PointNet weak supervision · S3DIS 1pt | CatAvg mIoU: 36.8 -> 41.9 | Helped |
| Add Siamese consistency on augmented point clouds (ablation) | PointNet weak supervision · ShapeNet 1pt | CatAvg mIoU: 65.2 -> 66.0 | Helped |
| Add the MIL inexact-supervision loss (ablation) | PointNet with Siamese consistency · ShapeNet 1pt | CatAvg mIoU: 66.0 -> 69.0 | Helped |
| Add training spatial and color smoothness (ablation) | PointNet with MIL and Siamese losses · PartNet 1pt | CatAvg mIoU: 52.1 -> 52.5 | Helped |
| Add TeLP label propagation at inference (ablation) | PointNet with MIL, Siamese, and smoothness · ShapeNet 1pt | CatAvg mIoU: 69.6 -> 70.2 | Helped |
| Use data augmentation without consistency constraints (ablation) | PointNet weak supervision · S3DIS 1pt | CatAvg mIoU: 36.8 -> 38.9 | Helped |
| Distribute the fixed label budget across more samples, with fewer labeled points per sample | PointNet segmentation · ShapeNet, fixed 10% label budget | CatAvg mIoU: 70.37 -> 77.71 | Helped |
| Increase labeled points from 1pt to 10% | DGCNN weakly supervised segmentation · ShapeNet | CatAvg mIoU: 74.4 -> 81.7 | Helped |
Closest papers by meaning
- Less Is More: Label Recommendation for Weakly Supervised Point Cloud Semantic Segmentation (AAAI 2024 · similarity 0.92)
- Weakly Supervised Point Cloud Semantic Segmentation via Artificial Oracle (CVPR 2024 · similarity 0.92)
- Multi-Modality Affinity Inference for Weakly Supervised 3D Semantic Segmentation (AAAI 2024 · similarity 0.92)
- Distribution Guidance Network for Weakly Supervised Point Cloud Semantic Segmentation (NeurIPS 2024 · similarity 0.90)
- Point Cloud Semantic Segmentation with Sparse and Inhomogeneous Annotations (AAAI 2025 · similarity 0.89)
- Rethinking Few-shot 3D Point Cloud Semantic 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.