I am trying to save an LAB image using torchvision.utils.save_image. The image dimensions are number_of_channel x L channel x A channel x B channel. Can the image be stored directly, or is there any change that needs to be made in order to store it correctly?
When I try to store the image directly using save_image, it distorts the image colors (the saved image and image plotted using pyplot imshow are completely different).
You cannot directly display images with values in Lab space: the packages you are using expects pixels to have values in RGB space. You need to explicitly convert the Lab values to RGB before saving/displaying the image.
You can find this post on SO useful for converting pixels between different color spaces.