Face shape estimation on Stirling Reconstruction Benchmark NoW Protocol HQ
0.92Non-Metrical Median ErrorMICA
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MICATraining Context=w/o. Stirling2022.04 | 0.92 | 1.16 | 1.04 | 1.06 | 1.35 | 1.2 | |
| Deng et al.Framework=PyTorch2022.04 | 0.99 | 1.27 | 1.15 | 1.31 | 1.71 | 1.57 | |
| DECA2022.04 | 1.03 | 1.32 | 1.18 | 1.22 | 1.58 | 1.42 | |
| Deng et al.Framework=TensorFlow2022.04 | 1.13 | 1.43 | 1.25 | 1.81 | 2.29 | 1.97 | |
| RingNet2022.04 | 1.15 | 1.46 | 1.27 | 1.33 | 1.6 | 1.54 | |
| 3DDFA-V22022.04 | 1.2 | 1.55 | 1.45 | 1.38 | 1.8 | 1.68 | |
| Average Face (FLAME)2022.04 | 1.22 | 1.55 | 1.35 | 1.4 | 1.79 | 1.57 | |
| MICAExclusion=Stirling dataset excluded from training2022.04 | 1.22 | 1.55 | 1.34 | 1.26 | 1.6 | 1.37 | |
| Deng et al.Framework=PyTorch2022.04 | 1.29 | 1.64 | 1.39 | 1.54 | 1.97 | 1.68 | |
| DECA2022.04 | 1.32 | 1.72 | 1.5 | 1.45 | 1.87 | 1.64 | |
| Deng et al.Framework=TensorFlow2022.04 | 1.41 | 1.84 | 1.63 | 2.14 | 2.65 | 2.12 | |
| 3DDFA-V22022.04 | 1.49 | 1.9 | 1.63 | 1.56 | 1.98 | 1.7 | |
| RingNet2022.04 | 1.6 | 2.05 | 1.76 | 1.64 | 2.09 | 1.81 | |
| Average Face (FLAME)Model=FLAME2022.04 | 1.62 | 2.08 | 1.83 | 1.62 | 2.09 | 1.85 |