Research map / Video / Action recognition
Self-Supervised 3D Skeleton Action Representation Learning With Motion Consistency and Continuity
Yukun Su, Guosheng Lin, Qingyao Wu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 triplet loss pre-training (ablation) | ST-GCN · NTU60 X-sub | accuracy: 81.5 -> 82.3 | Improved |
| add InfoNCE loss only pre-training (ablation) | ST-GCN · NTU60 X-sub | accuracy: 81.5 -> 81.8 | Improved |
| add interpolation loss (ablation) | ST-GCN · NTU60 X-sub | accuracy: 81.5 -> 82.5 | Improved |
| combine triplet and InfoNCE losses (ablation) | ST-GCN · NTU60 X-sub | accuracy: 82.3 -> 82.6 | Improved |
| add interpolation loss to triplet + InfoNCE (ablation) | ST-GCN · NTU60 X-sub | accuracy: 82.6 -> 83.0 | Improved |
| add interpolation loss to triplet + InfoNCE (ablation) | 2S-AGCN · NTU60 X-view | accuracy: 96.0 -> 96.3 | Improved |
| use motion-shuffled negatives with speed-changed positives instead of motion-jittered negatives (ablation) | ST-GCN · NTU60 X-sub | accuracy: 81.9 -> 82.3 | Improved |
| use motion-shuffled instead of view-transformed negatives (ablation) | ST-GCN · NTU60 X-sub | accuracy: 81.7 -> 82.3 | Improved |
| add speed-changed positives to motion-shuffled negatives (ablation) | ST-GCN · NTU60 X-view | accuracy: 88.6 -> 89.2 | Improved |
| use specific rate interpolation instead of direct reconstruction (ablation) | ST-GCN · NTU60 X-sub | accuracy: 82.0 -> 82.5 | Improved |
| use specific rate interpolation instead of direct reconstruction (ablation) | ST-GCN · NTU60 X-view | accuracy: 88.8 -> 89.4 | Improved |
| add MCC pre-training with 5% labels | ST-GCN · NTU60 X-sub 5% labels | accuracy: 38.2 -> 42.4 | Improved |
Closest papers by meaning
- Self-Supervised Skeleton-Based Action Representation Learning: A Benchmark and Beyond (IJCV 2026 · similarity 0.94)
- View-Invariant Skeleton Action Representation Learning via Motion Retargeting (IJCV 2024 · similarity 0.91)
- Self-Supervised 3D Action Representation Learning With Skeleton Cloud Colorization (TPAMI 2023 · similarity 0.90)
- Self-Supervised Action Representation Learning from Partial Spatio-Temporal Skeleton Sequences (AAAI 2023 · similarity 0.90)
- Heterogeneous Skeleton-Based Action Representation Learning (CVPR 2025 · similarity 0.89)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · 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.
