One faceplate appears in our FR10 camera trial's accepted and scratched folders because a bright line shifts when it is reseated; who should establish the physical reference before those labels guide threshold tuning?
Quality should judge the identified physical faceplate against the appearance requirement. Keep its images linked to that decision, including any uncertainty while it is being examined.
Keep the awkward accepted examples once quality decides too. A set containing only easy clean pictures won't tell you whether normal finish variation is being rejected. The presentation check matters alongside the labels.
Quality has taken the actual faceplate for assessment; the current images remain separated from the labelled reference set, and we have not adjusted the threshold.
Emma, agreed. Also retain genuine rejected examples; a lighting change that hides everything can look wonderfully consistent on the accepted set alone.
Exactly. I'd want the accepted reflective piece and an identified faint defect in the same comparison. Not necessarily the same surface version, unless that is the family they're actually qualifying.