Regarding the testing phase, does the model process the entire image at once, or is it necessary to crop the input into small patches and reconstruct them individually? I am asking to understand if there are potential blocking artifacts at boundaries that need post-processing.
Regarding the testing phase, does the model process the entire image at once, or is it necessary to crop the input into small patches and reconstruct them individually? I am asking to understand if there are potential blocking artifacts at boundaries that need post-processing.