One-shot Skeleton-based Action Recognition on NTU-120 SOAR
57.05AccuracyTrans4SOAR
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Trans4SOARScale=Base2022.02 | 57.05 | 55.9 | 57.26 | 57.12 | |
| Trans4SOARScale=Small2022.02 | 56.27 | 56.43 | 58.59 | 56.32 | |
| SL-DML (LeViT) + LSCEncoder=LeViT, Extension=LSC loss2022.02 | 55.94 | 54.29 | 55.8 | 56.04 | |
| Skeleton-DML2022.02 | 54.2 | — | — | — | |
| SL-DMLEncoder=LeViT, Optimization=DML2022.02 | 53.19 | 52.22 | 53.85 | 53.29 | |
| SL-DMLEncoder=Swin, Optimization=DML2022.02 | 53.13 | 52.09 | 53.48 | 53.16 | |
| SL-DMLEncoder=ResT, Optimization=DML2022.02 | 52.58 | 51.86 | 53.99 | 52.61 | |
| SL-DML2022.02 | 50.9 | — | — | — | |
| SL-DMLEncoder=ResNet18, Optimization=DML2022.02 | 49.19 | 47.54 | 49.8 | 49.23 | |
| SL-DMLEncoder=Twins, Optimization=DML2022.02 | 49 | 48.04 | 49.3 | 49.06 | |
| SL-DMLEncoder=ViT, Optimization=DML2022.02 | 48.45 | 47.4 | 48.59 | 48.52 | |
| SL-DMLEncoder=CaiT, Optimization=DML2022.02 | 47.86 | 47.53 | 50.06 | 47.94 | |
| TCN-OneShot2022.02 | 46.3 | — | — | — | |
| APSR2022.02 | 45.3 | — | — | — | |
| AP2022.02 | 42.9 | — | — | — | |
| SL-DMLEncoder=SqueezeNet, Optimization=DML2022.02 | 42.55 | 40.52 | 41.88 | 42.51 | |
| FC†2022.02 | 42.1 | — | — | — | |
| AN2022.02 | 41 | — | — | — | |
| SL-DMLEncoder=AlexNet, Optimization=DML2022.02 | 40.33 | 39.14 | 42.42 | 40.35 |