Inputs with M>3 channels to a CNN


I have input objects with size (10, 1024, 1024). I wish to pass these objects to a CNN to generate 1-dimensional embeddings. Is it ok to pass these objects as they are or does something need to be changed in the CNN architecture?

Actually, it’s relevant to know what these “objects” are.
It’s not like you can pass tensors with whichever meaning and it will work :sweat_smile:
Perhaps you should take a foundations course in ML or some pre-existing tool for the object type you are using.