The same retained faceplate has a bright stripe beside different holes in our FR10 reference images. Somebody labelled every stripe a scratch. I want quality's physical decision and a repeatable presentation before the vision comparison inherits those labels. The full captures are available, not just the marked crops.
our fr10 faceplate reference has a stripe that follows the viewpoint
DineshChan1082 · 2026年3月24日 11:20 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
5 条回复
Start by linking each full capture to the actual plate and its seat. Then have quality locate any real damage on that plate. Keep an accepted bright example and an actual reject in the comparison; don't make the brightness decide their labels.
12分Quality accepted the moving-stripe plate and identified a real scratch on a second retained plate. The original captions are withdrawn. Vision is comparing both in one agreed holder seat, with the full feature area included.
16分What happens when the normal loader reseats them? A single carefully arranged pair can look convincing while being a presentation only the trial engineer can reproduce.
18分Loader's repeats exposed a second resting position against the holder rim. We kept both sets identified. Tooling is reviewing how the intended seat is reached and recognised before vision chooses between the lighting candidates.
0分Give the holder reviewer the real scratch location as well as the seating discrepancy. A change that makes loading unambiguous but hides the inspected corner would just move your unfinished problem elsewhere.
8分讨论已关闭
此讨论已连续六个月没有活动，现已关闭新回复。 最后活动： .