Research map / Video / Action recognition
Learning Discriminative Representations for Skeleton Based Action Recognition
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 the FR Head | CTR-GCN · NTU RGB+D 120 X-Sub | accuracy (%): 84.5 -> 85.5 | Helped |
| add the FR Head | ST-GCN · NTU RGB+D 120 X-Set | accuracy (%): 85.1 -> 86.5 | Helped |
| add the FR Head | 2s-AGCN · NTU RGB+D 120 X-Sub | accuracy (%): 84.3 -> 84.6 | Helped |
| add the FR Head | TCA-GCN · NTU RGB+D 120 X-Sub | accuracy (%): 85.0 -> 85.2 | Helped |
| add the FR Head | HD-GCN · NTU RGB+D 120 X-Set | accuracy (%): 87.2 -> 87.7 | Helped |
| add CL loss to final-layer features (ablation) | CTR-GCN · NTU RGB+D 120 X-Sub, joint | accuracy (%): 84.5 -> 85.0 | Helped |
| decouple features into spatial and temporal components before refinement (ablation) | CTR-GCN · NTU RGB+D 120 X-Sub, joint | accuracy (%): 84.5 -> 85.3 | Helped |
| apply refinement at multiple stages (ablation) | CTR-GCN · NTU RGB+D 120 X-Sub, joint | accuracy (%): 84.5 -> 84.7 | Helped |
| combine CL loss, spatial-temporal decoupling, and multi-level refinement (ablation) | CTR-GCN · NTU RGB+D 120 X-Sub, joint | accuracy (%): 84.5 -> 85.5 | Helped |
| set w_cl to 0.1 instead of 1 with only the final stage weighted (ablation) | FR Head on CTR-GCN · NTU RGB+D 120 X-Sub, joint | accuracy (%): 84.4 -> 85.3 | Helped |
| reduce w_cl from 0.1 to 0.01 with only the final stage weighted (ablation) | FR Head on CTR-GCN · NTU RGB+D 120 X-Sub, joint | accuracy (%): 85.3 -> 85.0 | Hurt |
| use equal weights for all refinement stages (lambda_i = 1) (ablation) | FR Head on CTR-GCN · NTU RGB+D 120 X-Sub, joint | accuracy (%): 85.3 -> 84.5 | Hurt |
Closest papers by meaning
- InfoGCN: Representation Learning for Human Skeleton-Based Action Recognition (CVPR 2022 · similarity 0.91)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.89)
- Heterogeneous Skeleton-Based Action Representation Learning (CVPR 2025 · similarity 0.88)
- Stitch, Contrast, and Segment: Learning a Human Action Segmentation Model Using Trimmed Skeleton Videos (AAAI 2025 · similarity 0.88)
- Recovering Complete Actions for Cross-dataset Skeleton Action Recognition (NeurIPS 2024 · similarity 0.87)
- Skeleton-based Action Recognition with Non-linear Dependency Modeling and Hilbert-Schmidt Independence Criterion (AAAI 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.