About model input - Githubissues

NVIDIA / partialconv

A New Padding Scheme: Partial Convolution based Padding

Other

1.22k stars 213 forks source link

Closed XiaoYangon closed 3 years ago

XiaoYangon commented 4 years ago

hello, I have a question, Is the model input = image mask or input = imagemask +1-mask？ Is the hole white or black in image？

liuguilin1225 commented 4 years ago

input = image * mask. The holes are not used so the value should not matter.