Research map / Video / Action recognition
Hierarchical-aware Orthogonal Disentanglement Framework for Fine-grained Skeleton-based Action Recognition
Haochen Chang, Pengfei Ren, Haoyang Zhang, Liang Xie, Hongbo Chen, Erwei Yin
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 spatial granularity-aware bases to baseline (ablation) | Baseline · FineGYM (joint, ablation) | accuracy (%): 92.4 -> 93.0 | Improved |
| add temporal granularity-aware bases to baseline (ablation) | Baseline · FineGYM (joint, ablation) | accuracy (%): 92.4 -> 92.9 | Improved |
| combine spatial and temporal bases (ablation) | Baseline · FineGYM (joint, ablation) | accuracy (%): 92.4 -> 94.1 | Improved |
| add shared orthogonal bases for cross-granularity interaction (ablation) | HiOD · FineGYM (joint, ablation) | accuracy (%): 94.1 -> 94.4 | Improved |
| use granularity-aware bases instead of channel split (ablation) | HiOD · FineGYM (joint, ablation) | accuracy (%): 92.7 -> 94.1 | Improved |
| use granularity-aware bases instead of FC heads (ablation) | HiOD · FineGYM (joint, ablation) | accuracy (%): 92.9 -> 94.1 | Improved |
| use granularity-aware bases instead of wavelet transform (ablation) | HiOD · FineGYM (joint, ablation) | accuracy (%): 92.8 -> 94.1 | Improved |
| hard prototype contrastive loss instead of cross-entropy for both granularities (ablation) | HiOD · FineGYM (joint, ablation) | accuracy (%): 93.3 -> 93.6 | Improved |
| add soft prototypes for coarse-grained features (ablation) | HiOD · FineGYM (joint, ablation) | accuracy (%): 93.6 -> 93.8 | Improved |
| add soft prototype weighted contrastive loss (ablation) | HiOD · FineGYM (joint, ablation) | accuracy (%): 93.8 -> 94.4 | Improved |
| use HiOD over HD-GCN on NTU60 hard subset | HiOD · NTU60-Hard subset | accuracy (%): 66.3 -> 70.2 | Improved |
| use HiOD over HD-GCN on NTU60 easy subset | HiOD · NTU60-Easy subset | accuracy (%): 95.2 -> 95.5 | Improved |
Closest papers by meaning
- Multi-Granularity Anchor-Contrastive Representation Learning for Semi-Supervised Skeleton-Based Action Recognition (TPAMI 2022 · similarity 0.90)
- Hierarchical Consistent Contrastive Learning for Skeleton-Based Action Recognition with Growing Augmentations (AAAI 2023 · similarity 0.89)
- SCD-Net: Spatiotemporal Clues Disentanglement Network for Self-Supervised Skeleton-Based Action Recognition (AAAI 2024 · similarity 0.89)
- Skeleton-based Action Recognition with Non-linear Dependency Modeling and Hilbert-Schmidt Independence Criterion (AAAI 2025 · similarity 0.87)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.86)
- USDRL: Unified Skeleton-Based Dense Representation Learning with Multi-Grained Feature Decorrelation (AAAI 2025 · similarity 0.86)
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.
