Action Recognition on BABEL 100% OR
59.65AccuracySASI + ProtoGCN
Evaluation results
| Method | Links | |
|---|---|---|
| SASI + ProtoGCNModality=Skeleton + text, Param.=55.56M, Training Protocol=Separately Trained on each OR (%)2026.04 | 59.65 | |
| SASI + CTR-GCNModality=Skeleton + text, Param.=56.60M, Training Protocol=Separately Trained on each OR (%)2026.04 | 59.1 | |
| SASI + BlockGCNModality=Skeleton + text, Param.=56.99M, Training Protocol=Separately Trained on each OR (%)2026.04 | 59.1 | |
| BlockGCNModality=Skeleton, Publication=CVPR 2024, Param.=1.89M, Training Protocol=Separately Trained on each OR (%)2026.04 | 58.82 | |
| CTR-GCNModality=Skeleton, Publication=ICCV 2021, Param.=1.49M, Training Protocol=Separately Trained on each OR (%)2026.04 | 58.17 | |
| DeGCNModality=Skeleton, Publication=TIP 2024, Param.=1.53M, Training Protocol=Separately Trained on each OR (%)2026.04 | 58.08 | |
| ProtoGCNModality=Skeleton, Publication=CVPR 2025, Param.=4.30M, Training Protocol=Separately Trained on each OR (%)2026.04 | 58.08 | |
| MixformerModality=Skeleton, Publication=MM 2023, Param.=3.85M, Training Protocol=Separately Trained on each OR (%)2026.04 | 57.98 | |
| ST-GCNModality=Skeleton, Publication=AAAI 2018, Param.=3.17M, Training Protocol=Separately Trained on each OR (%)2026.04 | 57.71 | |
| Shift-GCNModality=Skeleton, Publication=CVPR 2020, Param.=0.81M, Training Protocol=Separately Trained on each OR (%)2026.04 | 57.53 | |
| GAP + CTR-GCNModality=Skeleton + text, Publication=ICCV 2023, Param.=2.18M, Training Protocol=Separately Trained on each OR (%)2026.04 | 56.53 | |
| SkateformerModality=Skeleton, Publication=ECCV 2024, Param.=3.33M, Training Protocol=Separately Trained on each OR (%)2026.04 | 56.33 | |
| InfoGCN++Modality=Skeleton, Publication=TPAMI 2024, Param.=0.72M, Training Protocol=Separately Trained on each OR (%)2026.04 | 55.96 | |
| 2s-AGCNModality=Skeleton, Publication=CVPR 2019, Param.=3.54M, Training Protocol=Separately Trained on each OR (%)2026.04 | 52.08 |