search

microsoft / Deep3DFaceReconstruction

Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set (CVPRW 2019)
MIT License
2.18k stars 443 forks source link

About the channels of the image #142

Closed hanssssssss closed 3 years ago

hanssssssss commented 3 years ago

Hi ! I am confoused about the channels of the face. I found that the channels of the attention_mask and imgs when training are different( I could see that the input channels of the imgs when training are BGR, while the channels of the attention_masks are RGB).

Thanks for your help.

YuDeng commented 3 years ago

The network receives images in BGR channel. attention_mask is actually one channel image but is expanded to three channels. So for attention_mask RGB and BGR have no difference.

hanssssssss commented 3 years ago

The network receives images in BGR channel. attention_mask is actually one channel image but is expanded to three channels. So for attention_mask RGB and BGR have no difference.

Thanks !