I am using the emnist data set via the PyTorch datasets together with a neural network that expects a 3 Channel input.
I would like to use PyTorch transforms to copy my 1D greyscale into 3D so I can use the same net for 1D and 3D data.
Which transform can I use? Or how would I extend PyTorch transforms as suggested here: https://stackoverflow.com/a/50530923/18895809