I have a special dataset and images in the dataset are with channels more than 3.
So I want to chage the code in torchvision.transforms. I do not know how many images will be produced after the transform.
For example, I input one image with the shape 256 256 30, and I want to crop the input image to size 224 224 30. How many images will produced normally for RGB images?
I also try to write a new crop function for numpy array, then combine it with transforms.Lambda, but when I use the transforms.ToTensor():
ValueError: some of the strides of a given numpy array are negative. This is currently not supported, but will be added in future releases.
Thanks!!