Table 4 per-class IoU (%) comparison on small categories of the CelebAMask-HQ test set.
From: A hybrid approach for facial parsing using transfer learning
Method | Necklace | L-Eyebrow | R-Eyebrow | U-Lip | L-Lip | Mean (All Classes) |
|---|---|---|---|---|---|---|
Baseline (A) | 0.00 | 80.11 | 80.05 | 82.34 | 81.99 | 78.34 |
Han et al56. | 8.91 | 85.10 | 84.95 | 87.22 | 86.88 | 80.40 |
Proposed MTL (D) | 15.32 | 87.45 | 87.31 | 89.50 | 89.15 | 81.95 |
