Action Recognition on CMU Mocap
98.8Top-1 AccSym-GNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Sym-GNNvariant=Complete model, architecture=light version2019.10 | 98.8 | 100 | |
| 2s-AGCN2019.10 | 97.1 | 99.8 | |
| Sym-GNNvariant=No prediction for multitasking, architecture=light version2019.10 | 96.6 | 99.4 | |
| Sym-GNNvariant=Only joint-scale structural graphs, architecture=light version2019.10 | 96.5 | 99.4 | |
| HCN2019.10 | 95.4 | 99.2 | |
| Sym-GNNvariant=Only part-scale graph, architecture=light version2019.10 | 94.9 | 98 | |
| Sym-GNNvariant=No bone-based dual graphs, architecture=light version2019.10 | 93.5 | 95.8 | |
| ST-GCN2019.10 | 87.5 | 96.9 |