NeRSemble
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Novel Expression Synthesis | NeRSemble | PSNR36.56 | 41 | |
| Novel View Synthesis | NeRSemble | SSIM85.8 | 24 | |
| 3D Head Reconstruction | NeRSemble (test) | PSNR24.22 | 20 | |
| Self-Reenactment | NeRSemble Novel Expression (all view) | PSNR25.32 | 13 | |
| Self-Reenactment | NeRSemble Novel Expression frontal | PSNR26.21 | 13 | |
| Novel View Synthesis | NeRSemble v2 (test) | LPIPS0.152 | 13 | |
| Correspondence Matching Quality | Nersemble (same-person) | MAE3.68 | 10 | |
| 3D Face Reconstruction | NeRSemble-SVFR (Posed) | L1 Error1.37 | 10 | |
| Novel View Synthesis | NeRSemble (test) | PSNR31.5 | 10 | |
| Monocular depth estimation | NeRSemble Video | RMSE0.075 | 8 | |
| Monocular depth estimation | NeRSemble Image | RMSE0.077 | 8 | |
| Monocular 3D Head Avatar Creation | NeRSemble | PSNR20.9 | 8 | |
| Identity Preservation | Nersemble (cross-person) | ArcFace Score38.4 | 7 | |
| Video-driven Talking Head Generation (Cross-Reenactment) | NeRSemble Mono | FID57.82 | 7 | |
| Video-driven Talking Head Generation (Self-Reenactment) | NeRSemble Mono | PSNR31.98 | 7 | |
| Novel-view synthesis | Nersemble frontal-to-mid views v2 (test) | PSNR21.33 | 6 | |
| Self-reenactment | NeRSemble subset v2 | PSNR21.586 | 6 | |
| Talking Head Synthesis | NeRSemble v2 | PSNR21.586 | 6 | |
| Expression kinematic lifting | NeRSemble (test) | PSNR28.685 | 6 | |
| Cross Reenactment | NeRSemble | AED0.1709 | 6 | |
| Cross Reenactment | Nersemble (test) | CSIM85.49 | 6 | |
| Self Reenactment | Nersemble (test) | PSNR22.6313 | 6 | |
| Motion-conditioned Image-to-Video Animation | NeRSemble (test) | Hair SSIM0.967 | 6 | |
| Novel View Synthesis | NeRSemble (val) | PSNR22.76 | 5 | |
| Self-Reenactment | NeRSemble (val) | PSNR22.21 | 5 |
Showing 25 of 55 rows