Text-to-motion generation on Compound Action Dataset (test)
99.1R-Precision Top-1Ground Truth
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Ground Truth2026.04 | 99.1 | 100 | 0.413 | 0.014 | 7.134 | 1.872 | |
| Motion-AdapterBackbone=MotionDiffuse2026.04 | 16.2 | 35.9 | 14.988 | 3.719 | 8.246 | 1.381 | |
| Motion-AdapterBackbone=MDM2026.04 | 15.8 | 36 | 14.952 | 3.592 | 7.744 | 1.936 | |
| SALAD2026.04 | 15.4 | 35.7 | 15.018 | 14.363 | 9.572 | 0.366 | |
| MotionLab2026.04 | 14.1 | 33.8 | 15.238 | 8.342 | 8.866 | 0.017 | |
| MoGenTS2026.04 | 13.7 | 33.4 | 14.991 | 9.51 | 9.306 | 1.322 | |
| MDM2026.04 | 12.1 | 28.2 | 16.132 | 8.019 | 8.47 | 2.248 | |
| MotionDiffuse2026.04 | 10.9 | 25.3 | 16.38 | 4.733 | 8.556 | 0.25 | |
| SALADMode=Motion Editing2026.04 | 7.8 | 21.7 | 17.723 | 27.131 | 5.938 | 1.096 | |
| PriorMDM2026.04 | 7.6 | 17.4 | 22.35 | 15.712 | 8.344 | 1.194 | |
| STMC2026.04 | 5.1 | 13.2 | 19.056 | 52.563 | 1.003 | 0.012 |