I may not know the correct term, but what is it called when the input/output dataset are in 2 different folders, and each input output pair is linked by name, and how do I load it into pytorch? I plan to load point clouds or images, for either both input and output, so is the usage the same for both point clouds or images?
If the file names of the input and output files use the same file pattern, you could try to sort them and load the pair in a Dataset.
On the other hand, if that’s not possible, I would recommend to create a lookup table (e.g. using a list of tuples, dict or any other container) to get the right file paths for the current index in the __getitem__ of your Dataset.