Research map / Point clouds / Segmentation and detection
Spatiotemporal Self-Supervised Learning for Point Clouds in the Wild
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 point-to-cluster learning to pre-training (ablation) | STSSL · SemanticKITTI, 0.1% labels | mIoU: 34.01 -> 35.48 | Helped |
| Add inter-frame learning to SegContrast (ablation) | SegContrast · SemanticKITTI, 0.1% labels | mIoU: 34.01 -> 36.03 | Helped |
| Replace MoCo with BYOL (ablation) | SegContrast · SemanticKITTI, 0.1% labels | mIoU: 34.01 -> 33.94 | No clear effect |
| Add feature similarity to location similarity for tracking (ablation) | STSSL · SemanticKITTI, 0.1% labels | mIoU: 36.01 -> 37.71 | Helped |
| Add feature similarity to location similarity for tracking (ablation) | STSSL · SemanticKITTI, 1% labels | mIoU: 51.06 -> 52.60 | Helped |
| Use a fixed interval of 5 between frames (ablation) | STSSL · SemanticKITTI | mIoU: 36.05 | Helped |
| Increase frame interval beyond the optimum (ablation) | STSSL · SemanticKITTI | mIoU: 36.05% | Mixed |
| Pre-train with STSSL rather than SegContrast | MinkUnet · SemanticKITTI, 0.1% labels, KITTI pre-training | mIoU: 34.01 -> 37.71 | Helped |
| Pre-train on nuScenes seq1-2 instead of seq1 | STSSL · SemanticKITTI, 0.1% labels | mIoU: 34.43 -> 35.08 | Helped |
Closest papers by meaning
- PSA-SSL: Pose and Size-aware Self-Supervised Learning on LiDAR Point Clouds (CVPR 2025 · similarity 0.90)
- Mitigating Object Dependencies: Improving Point Cloud Self-Supervised Learning through Object Exchange (CVPR 2024 · similarity 0.89)
- FreePoint: Unsupervised Point Cloud Instance Segmentation (CVPR 2024 · similarity 0.88)
- Construct to Associate: Cooperative Context Learning for Domain Adaptive Point Cloud Segmentation (CVPR 2024 · similarity 0.87)
- Sonata: Self-Supervised Learning of Reliable Point Representations (CVPR 2025 · similarity 0.87)
- Self-Supervised Class-Agnostic Motion Prediction with Spatial and Temporal Consistency Regularizations (CVPR 2024 · 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.