Skeleton-based One-shot Action Recognition on Toyota Smart Home Without Occlusion
70.22AccuracyTrans4SOAR
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Trans4SOARModel Scale=Base2022.02 | 70.22 | 33.96 | 37.81 | 35.33 | |
| Trans4SOARModel Scale=Small2022.02 | 66.87 | 28.08 | 31.47 | 34.63 | |
| SL-DMLEncoder=LeViT, Optimization Strategy=LSC2022.02 | 64.46 | 31.91 | 34.07 | 33.58 | |
| SL-DMLEncoder=LeViT2022.02 | 62.22 | 31.98 | 37.56 | 35.16 | |
| SL-DMLEncoder=Original2022.02 | 58.98 | 27.15 | 27.64 | 35 | |
| SL-DMLEncoder=Swin2022.02 | 58.76 | 28.83 | 29.17 | 32.34 | |
| Skeleton-DML2022.02 | 47.31 | 18.45 | 18.58 | 23.8 |