I am rewriting 3D V-Net to 2.5D V-Net to segment livers.
It means I have 2 labels - background, livers.
This specific 2.5D networks means that we use 5 slices as 5 channels.
Input to the network has shape (1,5,300,300).
But the output has a shape (1,2,300,300) where 2 represents number of classes.
Why the output shape is not like (1,2,5,300,300)? I need different output for each channel since they are different right? Do you have any idea how it should be correctly?
Thanks for the answers.