Fig. 14
From: A hybrid approach for facial parsing using transfer learning
Facial parsing results with CelebAMask-HQ dataset and feature activation. (a) Images from CelebAMask-HQ dataset (b) Ground Truth (c) Predicted Image (d) Pre-training parsing results of baseline model (e) Pre-training parsing results with proposed MTL model (f) Encoder feature activation map from proposed MTL model.
