Face Shape Estimation on Stirling Reconstruction Benchmark NoW Protocol (LQ)
0.96Non-Metrical Median ErrorMICA
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MICATraining Context=w/o. Stirling2022.04 | 0.96 | 1.22 | 1.11 | 1.15 | 1.46 | 1.3 | |
| DECA2022.04 | 1.09 | 1.39 | 1.26 | 1.32 | 1.71 | 1.54 | |
| Deng et al.Framework=PyTorch2022.04 | 1.12 | 1.44 | 1.31 | 1.47 | 1.93 | 1.77 | |
| RingNet2022.04 | 1.17 | 1.49 | 1.31 | 1.37 | 1.72 | 1.6 | |
| Deng et al.Framework=TensorFlow2022.04 | 1.22 | 1.57 | 1.4 | 1.85 | 2.41 | 2.16 | |
| Average Face (FLAME)2022.04 | 1.23 | 1.56 | 1.38 | 1.44 | 1.84 | 1.64 | |
| 3DDFA-V22022.04 | 1.26 | 1.63 | 1.52 | 1.49 | 1.93 | 1.78 | |
| MICAExclusion=Stirling dataset excluded from training2022.04 | 1.26 | 1.62 | 1.41 | 1.36 | 1.73 | 1.48 | |
| DECA2022.04 | 1.4 | 1.81 | 1.59 | 1.56 | 2.03 | 1.81 | |
| Deng et al.Framework=PyTorch2022.04 | 1.51 | 1.95 | 1.71 | 1.78 | 2.28 | 1.97 | |
| RingNet2022.04 | 1.56 | 2.01 | 1.75 | 1.67 | 2.16 | 1.9 | |
| Deng et al.Framework=TensorFlow2022.04 | 1.56 | 2.02 | 1.77 | 2.13 | 2.71 | 2.33 | |
| Average Face (FLAME)Model=FLAME2022.04 | 1.58 | 2.06 | 1.82 | 1.7 | 2.19 | 1.96 | |
| 3DDFA-V22022.04 | 1.58 | 2.03 | 1.74 | 1.7 | 2.16 | 1.88 |