4D Facial Expression Generation on CoMA (test)
73.61Classification AccuracyGround Truth Landmarks
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Ground Truth LandmarksLandmark extraction method=Det-LM (Extracted from meshes via S2D-Dec)2021.05 | 73.61 | — | |
| Ground Truth LandmarksLandmark extraction method=Gen-LM (Direct landmarks)2021.05 | 73 | — | |
| Motion3DGANLandmark extraction method=Det-LM (Extracted from meshes via S2D-Dec)2021.05 | 69.44 | 19.01 | |
| Motion3DGANLandmark extraction method=Gen-LM (Generated landmarks directly)2021.05 | 65.28 | 20.45 | |
| LSTMLandmark extraction method=Det-LM (Extracted from meshes via S2D-Dec)2021.05 | 52.08 | 27.96 | |
| LSTMLandmark extraction method=Gen-LM (Generated landmarks directly)2021.05 | 46.53 | 21.76 |