Machine Vision and Industrial Quality Inspection
SCENE · AOI · Workstation re-inspection · End-side collaboration
Stable delivery focused on three capabilities: defect detection / dimensional measurement / positioning. Built for industrial operations: thresholds and gray‑zone policies route low‑confidence cases to HITL. Every decision ships with traceable evidence (raw image, parameters, model version and results), supporting regression comparison and acceptance.
competency matrix
- Defect and foreign object detection; texture and micro defect contrast learning
- Sub-pixel size and geometric relationship measurements
- Workpiece pose and grab point output
- Compliance comparison (barcode, nameplate, assembly in place)
Training efficiency (acceptance criteria)
In a typical AOI scenario, approx.30The first round of available models can be started with just one annotated image; under streamlined annotation, the fastest time from collection to edge deployment is approximately10Minutes (subject to GPU resource queuing).
Specification excerpt
| item | content |
|---|---|
| camera protocol | GigE/ USB3 / Mainstream domestic industrial camera SDK |
| cluster | Training task queues are parallelized with multiple cards; quotas are isolated from tenants |
| deploy | Edge box + industrial computer + computer room server hybrid |
Manufacturing quality inspection plan white paper Book a demo