Talking Head Generation On Voxceleb2 1 Shot
평가 지표
CSIM
LPIPS
Normalized Pose Error
SSIM
inference time (ms)
평가 결과
이 벤치마크에서 각 모델의 성능 결과
모델 이름 | CSIM | LPIPS | Normalized Pose Error | SSIM | inference time (ms) | Paper Title | Repository |
|---|---|---|---|---|---|---|---|
| Fast Bi-layer Avatars (medium size) | 0.653 | 0.358 | 43.3 | 0.508 | 4 | Fast Bi-layer Neural Synthesis of One-Shot Realistic Head Avatars | |
| CainGAN | - | - | - | - | - | Pose Manipulation with Identity Preservation | - |
| Few-shot Adversarial Model | - | - | - | - | - | Few-Shot Adversarial Learning of Realistic Neural Talking Head Models | |
| Few-shot Vid-to-vid (medium size) | 0.604 | 0.368 | 46.1 | 0.419 | 22 | Fast Bi-layer Neural Synthesis of One-Shot Realistic Head Avatars | |
| First Order Motion Model (medium size) | 0.638 | 0.311 | 47.8 | 0.553 | 13 | Fast Bi-layer Neural Synthesis of One-Shot Realistic Head Avatars |
0 of 5 row(s) selected.