QLRO
Manufacturingml

Vision-based defect detection

Classify production-line camera tiles as defective vs. acceptable after CNN feature extraction. Semiconductor wafer, PCB, and automotive-paint inspection.

Configure your run
Data scale

How big is the problem you're trying to solve?

Accuracy you need

What error tolerance can your downstream decision absorb?

Total project budget (USD)

We use this to filter candidate devices. We divide by the typical run count for this workload — you can override that under Advanced.

Advanced— response time, audit level, label
Response time
Audit level

'Regulated' triggers automatic decision-record + bilingual PDF + ZIP bundle.

Free text — rides into the decision record so audits can grep your reference (e.g. "AAPL Dec 2026 $200 call").

Sign in to run — free tier is 500/month, shared with the qlro CLI on the same key.

The same circuit shape and ranking are produced by qlro.recommend_workload("industry.manufacturing.vision_defect_detection") in the Python SDK — useful if you want to automate this in CI / CD.