Text-to-motion generation on OmniZoo 1.0 (test)
1.412DiversityReal
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| RealData Type=Ground Truth2025.12 | 1.412 | — | 1.029 | — | 68.4 | 86 | 92.7 | |
| Generalized Motion Residual VQ-VAEdataset=OmniZoo2025.12 | 1.404 | 0.044 | 1.072 | 0.569 | 62.1 | 80.5 | 87.7 | |
| MMMadaptation=joint padding and binary masking, dataset=OmniZoo2025.12 | 1.395 | 0.084 | 1.141 | 0.032 | 50.3 | 68 | 76.6 | |
| MoMaskadaptation=joint padding and binary masking, dataset=OmniZoo2025.12 | 1.392 | 0.085 | 1.181 | 0.513 | 45.4 | 62.5 | 71.2 | |
| Animoadaptation=padding and retrained, dataset=OmniZoo2025.12 | 1.377 | 0.141 | 1.226 | 0.724 | 33.7 | 47.3 | 54.6 |