Yes, quality owns that decision. We have the faceplates, though the image folder doesn't identify which is which. I can arrange fresh identified captures; I can't reconstruct those identities reliably from the thumbnails.

Keep the old folder separate, then; fresh labels should not pretend to repair its missing identities.

And let quality mark the actual region they examined, because an accepted faceplate can still have a visible harmless feature and a rejected one can still contain glare; the image needs more than a whole-part verdict.