One shared definition of a defect.
Every defect type has reference images, a severity and a measured limit, written once by your quality team and applied the same way on every line and every shift.
Reference images
Good and bad examples for each defect type, so operators, auditors and the model all see the same thing.
Limits in millimeters
Length, width, area, depth by shading and distance from an edge. Rules read like your control plan.
Severity levels
Critical, major and minor, each with its own action: reject, review or accept.
Approval workflow
Changes are proposed, reviewed and approved by named people before they reach the line.
Full change log
Who changed which rule, when and why. Ready for customer audits and PPAP.
Per customer variants
The same part can carry tighter limits for one customer than another.
From a new defect to a rule on the line.
Three levels, three actions.
Safety or function at risk. Always rejected, never sent to review.
Examples: Cracks, missing clips, missing tablets, wrong part
Visible to the customer or affects fit. Rejected above a measured limit.
Examples: Scratches over 1.5 mm, unreadable lot codes, misaligned labels
Cosmetic. Accepted below the limit, reviewed when borderline.
Examples: Small dents, light burrs, faint marks
Defect Library at a glance
| Defect types per product | Unlimited |
|---|---|
| Rule inputs | Length, width, area, count, position zone, contrast, confidence |
| Actions | Reject, review, accept, stop line |
| Approvals | One or two step sign-off, by role |
| Exports | Defect catalog in PDF for operator training and customer audits |
Asked by quality teams
Can we start from our existing defect catalog?
Yes. Upload your boundary samples and limit sheets and we map them into defect types and rules during the pilot.
Can rules differ between zones of the part?
Yes. Draw zones such as sealing surface or visible face and give each zone its own limits.
Who can change a rule?
Only roles you allow. Every change needs approval and is recorded with the name, time and reason.
Turn the knob to your line speed.
A 5% manual sample checks 5,760 of them. NanoQuantumAI checks all 115,200, and keeps a picture of each one.