OmarChan1100 · 28 Jul 2025, 12:43 UTC
Quality, with the physical plates identified. We've had long arguments about a bright line that nobody had ever looked for on the actual metal. Record the real feature's position and judgement before asking the camera to agree with an annotation.
2 pointsQuality can inspect the retained plates. We haven't matched all the old images to individual pieces.
21 pointsLeave the untraceable images outside the approved examples. Rephotograph identified samples instead of guessing which plate resembles each old picture
9 pointsWho decides what happens to a borderline plate? We had operators handed a bad box with no explanation of whether bad meant scratched, dirty or bent. Not much use as a teaching reference.
15 pointsQuality lead. Borderline pieces get an unresolved label until reviewed, not automatically rejected to tidy the folder.
8 pointsAnd agree the viewing conditions with them. Then compare the camera's presentation using the same identified pieces. A physical acceptance decision doesn't guarantee the proposed image makes the relevant feature visible.
16 pointsCheck seating while you're there. A plate catching on the nest edge can make small presentation changes look like random lighting behaviour. I'm not saying that's yours, just worth inspecting
3 pointsQuality found a real edge scuff on one plate. Viewing conditions agreed. Fixture check still outstanding.
21 pointsWas that scuff accepted or rejected, Omar? Found is not quite enough for the training sheet, even if its location is now clear.
11 pointsAccepted under the agreed requirement, thanks Jo. I've put that judgement beside its location. Imaging checks remain open.
4 pointsThis discussion is closed to new replies after six months without activity. Last activity: .