Research map / Graph learning / Knowledge and temporal graphs
Adaptive Hyper-Graph Convolution Network for Skeleton-based Human Action Recognition with Virtual Connections
Youwei Zhou, Tianyang Xu, Cong Wu, Xiaojun Wu, Josef Kittler
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 multi-head hyper-graph convolution (non-uniform, K=9) (ablation) | baseline GCN · NTU-120 X-Sub (joint stream) | accuracy: 84.7 -> 86.7 | Improved |
| use non-uniform instead of uniform hyper-graph at K=9 (ablation) | M-HGC · NTU-120 X-Sub (joint stream) | accuracy: 86.0 -> 86.7 | Improved |
| use uniform hyper-graph at K=5 versus non-uniform K=5 (ablation) | M-HGC · NTU-120 X-Sub (joint stream) | accuracy: 86.5 vs 86.2 | Mixed |
| increase K from 9 to 11 (non-uniform) (ablation) | M-HGC · NTU-120 X-Sub (joint stream) | accuracy: 86.7 -> 86.4 | Hurt |
| decrease K from 9 to 3 (non-uniform) (ablation) | M-HGC · NTU-120 X-Sub (joint stream) | accuracy: 86.7 -> 86.0 | Hurt |
| add 3 hyper joints with Divergence Loss without M-HGC (ablation) | baseline GCN · NTU-120 X-Sub (joint stream) | accuracy: 84.7 -> 85.2 | Improved |
| add Divergence Loss to 3 hyper joints with M-HGC (ablation) | Hyper-GCN · NTU-120 X-Sub (joint stream) | accuracy: 86.6 -> 86.9 | Improved |
| add Divergence Loss to 5 hyper joints without M-HGC (ablation) | baseline GCN · NTU-120 X-Sub (joint stream) | accuracy: 84.7 -> 85.0 | Improved |
| increase hyper joints from 3 to 5 per layer (ablation) | Hyper-GCN with L_h · NTU-120 X-Sub (joint stream) | accuracy: 86.9 -> 86.8 | Hurt |
| use 1 hyper joint instead of 3 with Divergence Loss (ablation) | Hyper-GCN · NTU-120 X-Sub (joint stream) | accuracy: 86.9 -> 86.7 | Hurt |
| scale base to large version (channels 512 in last stage) | Hyper-GCN · NTU-120 X-Sub (4-stream ensemble) | accuracy: 90.5 -> 90.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- BlockGCN — BlockGCN: Redefine Topology Awareness for Skeleton-Based Action Recognition (CVPR 2024) · better on 5 of 5 shared results
- SkateFormer — SkateFormer: Skeletal-Temporal Transformer for Human Action Recognition (ECCV 2024) · better on 3 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Context Aware Graph Convolution for Skeleton-Based Action Recognition (CVPR 2020 · similarity 0.94)
- Learning Graph Convolutional Network for Skeleton-Based Human Action Recognition by Neural Searching (AAAI 2020 · similarity 0.93)
- Dynamic Semantic-Based Spatial Graph Convolution Network for Skeleton-Based Human Action Recognition (AAAI 2024 · similarity 0.93)
- BlockGCN: Redefine Topology Awareness for Skeleton-Based Action Recognition (CVPR 2024 · similarity 0.92)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.90)
- Are Spatial-Temporal Graph Convolution Networks for Human Action Recognition Over-Parameterized? (CVPR 2025 · similarity 0.90)
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.
