The input to the network is 4 channel tensor. For that, I am using bitmap to read two images (one is gray scale and the other is rgb). Any idea on how to concatenate in a tensor form?
Once you’ve loaded each image and transformed it to a tensor in the shape
[channels, height, width], you could concatenate it with
torch.cat as seen here:
gray = torch.randn(1, 24, 24) rgb = torch.randn(3, 24, 24) x = torch.cat((gray, rgb), dim=0) print(x.shape) > torch.Size([4, 24, 24])