face-parsing.PyTorch
face-parsing.PyTorch copied to clipboard
train classes mis-leading..
Since there are 17 classes1 in the prepare stags, while there are 19 classes in pretrained model from test.py
.2
So,my concern is what is the differences between those two models?
they said that it contains last background part, but I test result found the class name make more confused
and this is my label order: ['skin', 'l_brow', 'r_brow', 'l_eye', 'r_eye', 'eye_g', 'l_ear', 'r_ear', 'ear_r', 'nose', 'mouth', 'u_lip', 'l_lip', 'neck', 'neck_l', 'cloth', 'hair', 'hat', 'bg']
https://github.com/switchablenorms/CelebAMask-HQ/tree/master/face_parsing#well-trained-model
Background is first class. I tested this model (the one that is mentioned in README.md) and found this labeling:
['bg', 'skin', 'l_brow', 'r_brow', 'l_eye', 'r_eye', 'eye_g', 'l_ear', 'r_ear', 'ear_r', 'nose', 'mouth', 'u_lip', 'l_lip', 'neck', 'neck_l', 'cloth', 'hair', 'hat']
https://github.com/switchablenorms/CelebAMask-HQ/tree/master/face_parsing#well-trained-model
This label is not suitable for this model