Action-Conditioned Video Generation on EgoDex (test)
39.58FVDLOME
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| LOMEResolution=256 x 256, Condition=3D action maps2026.03 | 39.58 | 90.3 | 28.4 | 0.051 | 66.85 | |
| GwtFResolution=256 x 256, Condition=optical flow2026.03 | 40.33 | 87.8 | 25.6 | 0.071 | 47.06 | |
| CoSHANDResolution=256 x 256, Condition=hand masks2026.03 | 59.83 | 84.6 | 26.7 | 0.171 | 51.33 | |
| Wan-I2VResolution=256 x 256, Condition=text and image2026.03 | 68.77 | 88.4 | 29.4 | 0.057 | 10.71 |