Hello author, first of all, thank you very much for your outstanding contribution.
I am currently working on HumanParsing, which involves segmenting multiple parts of the human body such as hair, clothing, skin, etc. If my data consists of original images and grayscale images with label values ranging from 0 to 18, can I use Sam HQ for training?