Our trim plate changes verdict when the bright streak moves

LeoBaker0449 · 1 Mar 2026, 16:02 UTC

Reply to discussion
LE
LeoBaker0449
One stainless trim plate has two scratch labels in our FR5 image folder. The bright streak moves between captures. The physical plate does not have two matching marks. I want those labels checked before they become training examples.

10 replies

SO
SofiaArcher0369
Replying to LeoBaker0449

Keep the plate tied to both captures. Who assigned the labels, and were they looking at the actual plate or only the images?

17 points
LE
LeoBaker0449
Replying to SofiaArcher0369

Two people labelled separate image folders. Neither had the plate. Quality has now examined it and found no corresponding scratch at either labelled position.

8 points
JA
JackAbbott0080
Replying to LeoBaker0449

Can you withdraw both labels without deleting the images? I would keep them as examples of the confusion, provided their corrected status is obvious.

22 points
IS
IsaacAdams0172
Replying to LeoBaker0449

And get an actual marked plate beside it. We improved seating on a cover job and still lost the real scratch under the lamp. A clean image wasn't enough.

4 points
SO
SofiaArcher0369
Replying to IsaacAdams0172

Isaac's point is useful. Leo, do you have an agreed reject available for the imaging comparison?

11 points
LE
LeoBaker0449
Replying to SofiaArcher0369

Yes. A separate identified plate with an accepted rejection decision. We have kept its physical mark location with its captures. The two false labels on the first plate are withdrawn.

5 points
JA
JackAbbott0080
Replying to LeoBaker0449

Who updates the copies already sent to the programmer? I do not know your folder arrangement, but correcting the source alone may leave them using the old version.

10 points
LE
LeoBaker0449
Replying to JackAbbott0080

I sent the corrected index and asked them to discard the old labelled set. They confirmed the active set now uses the corrected status. Thank you, Jack; there was a local copy I had forgotten.

14 points
IS
IsaacAdams0172
Replying to LeoBaker0449

How does the actual reject look in the proposed presentation?

10 points
LE
LeoBaker0449
Replying to IsaacAdams0172

Its mark is clear in one capture and weak in another. The labelling error is closed. Imaging still needs work before we tune acceptance.

9 points

Add to the discussion

Welcome to Application Robot

Everyone can read the forum. Sign in or create an account to start a discussion, reply, or upload photos.

Forgot your password?

By creating an account, you agree to our Terms and Conditions and community guidelines. Read our Privacy Policy for how your information is handled.