Research map / Graph learning / Knowledge and temporal graphs
Multi-Scale Spatial Temporal Graph Convolutional Network 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 |
|---|---|---|---|
| Replace local spatial convolution with MS-GC using 4 subsets (ablation) | Adaptive ST-GCN · NTU RGB+D X-sub | Top-1 accuracy: 87.2% -> 88.2% | Helped |
| Replace local temporal convolution with MT-GC using 4 subsets (ablation) | Adaptive ST-GCN · NTU RGB+D X-sub | Top-1 accuracy: 87.2% -> 88.4% | Helped |
| Use 2 instead of 4 spatial subsets in MS-GCN (ablation) | MS-GCN · NTU RGB+D X-sub | Top-1 accuracy: 87.8% -> 88.2% | Hurt |
| Use 3 instead of 4 spatial subsets in MS-GCN (ablation) | MS-GCN · NTU RGB+D X-sub | Top-1 accuracy: 88.1% -> 88.2% | Hurt |
| Use a lighter 16c × 4s channel setting (ablation) | MS-GCN · NTU RGB+D X-sub | Top-1 accuracy: 88.2% -> 88.0% | Hurt |
| Use 2 instead of 4 temporal subsets in MT-GCN (ablation) | MT-GCN · NTU RGB+D X-sub | Top-1 accuracy: 87.7% -> 88.4% | Hurt |
| Use 3 instead of 4 temporal subsets in MT-GCN (ablation) | MT-GCN · NTU RGB+D X-sub | Top-1 accuracy: 88.1% -> 88.4% | Hurt |
| Use a lighter 16c × 4s channel setting (ablation) | MT-GCN · NTU RGB+D X-sub | Top-1 accuracy: 88.4% -> 87.9% | Hurt |
| Replace only spatial convolution with MS-GC (ablation) | Adaptive ST-GCN · NTU RGB+D X-sub | Top-1 accuracy: 87.2% -> 88.2% | Helped |
| Replace only temporal convolution with MT-GC (ablation) | Adaptive ST-GCN · NTU RGB+D X-sub | Top-1 accuracy: 87.2% -> 88.4% | Helped |
| Use MS-GC and MT-GC together (ablation) | Adaptive ST-GCN · NTU RGB+D X-sub | Top-1 accuracy: 87.2% -> 89.0% | Helped |
| Use both MS-GC and MT-GC with a lightweight channel setting (ablation) | Adaptive ST-GCN · NTU RGB+D X-sub | Top-1 accuracy: 87.2% -> 88.1% | Helped |
Closest papers by meaning
- Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition (AAAI 2018 · similarity 0.95)
- Dynamic Semantic-Based Spatial Graph Convolution Network for Skeleton-Based Human Action Recognition (AAAI 2024 · similarity 0.94)
- Are Spatial-Temporal Graph Convolution Networks for Human Action Recognition Over-Parameterized? (CVPR 2025 · similarity 0.92)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.91)
- Spatio-Temporal Fusion for Human Action Recognition via Joint Trajectory Graph (AAAI 2024 · similarity 0.91)
- BlockGCN: Redefine Topology Awareness for Skeleton-Based Action Recognition (CVPR 2024 · similarity 0.91)
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.