mask preprocessing

[soans1994]

hello authors,

please help me with some query. i see your unet model output is shape (batch, num_classes, h,w). i have similar unet model. but i want to know how to prprocess mask. my input image after processing is (b,c=3,h,w) and mask also same since is its color mask for 21 classes. can i train network with 3 channels mask or shd i convert it to some other format.