I put our retained faceplate back in its holder and the red scratch arrow landed on a different bright line. The physical mark near the fixing hole had not moved. We already have quality's accepted and rejected examples, but the saved images do not say how each faceplate was seated.
I want a useful comparison before anyone tunes the inspection. The current set could teach a camera to dislike whichever reflection happened to be there.
FR5 faceplate labels attached to a moving streak
JoCarter1027 · 2026年5月7日 08:46 UTC
主帖中文翻译暂不可用，以下显示英文原文；回复保留原文。 查看英文原文
11 条回复
回复 JoCarter1027
Our team once trusted the filename more than the object; can you still match each saved image to its actual faceplate?
25分Only half of them. The others have batch names, not individual identities. I have removed those from the labelled reference set.
13分回复 JoCarter1027
Keep the unlabelled ones somewhere separate though. They may still show how much the presentation wanders, even if nobody can honestly call the streak a scratch.
23分回复 JoCarter1027
Does the retained rejected example show its physical mark in the proposed camera view? A correctly identified sample is not much help if the holder conceals the defect you want the station to judge.
14分Hana, fair point; I meant unsuitable as labelled examples, not useless pictures.
14分Harish, the fixing-hole mark is partly behind the holder lip. Tooling has been asked to expose that area while keeping the intended support. Hana, the uncertain pictures are retained separately.
16分回复 JoCarter1027
Please include the person who changes these over when the holder is revised. Our last little visibility improvement made the plate easy to put in crooked. It was obvious to the designer where it belonged, less obvious with a stack waiting beside you.
12分回复 JoCarter1027
The revised holder exposes the mark, and both loaders can seat the faceplate consistently in the bench comparison. We found the old lip was also letting one corner sit high. Its shadow and the wandering streak are now separated from the labelled defect.
12分回复 JoCarter1027
Which accepted example was compared in that revised holder? I would not retain the old accepted image as the other half of the test.
3分We recaptured the retained accepted faceplate and both rejected examples in the revised holder, with their identities and image settings. Quality confirmed those labels. The broader sample comparison is still outstanding.
3分回复 JoCarter1027
Useful, Jo. Can the loader tell which holder revision is fitted without opening your image folder?
13分