The machine-vision buyer’s map: camera, system or AI platform?

The hardest vision problems are often not the model: optics, lighting, line speed, product variation and what happens after a detection all matter.

There is no single machine-vision product category

A camera component and a turnkey inspection application solve different procurement problems. Basler and Teledyne DALSA describe imaging components and supporting software; Cognex and KEYENCE sell integrated vision systems; Oxipital AI, 36ZERO Vision and Maneva emphasize AI-supported application software or vision-to-action workflows. These are public product positions, not statements of technical superiority.

Map the integration boundary

Where will illumination be engineered? Who chooses optics? Who owns inference hardware? Which partner connects alarms to PLCs and robot safety logic? Who handles production changes? A buyer should compare the scope of deliverables and ongoing integration responsibilities rather than putting all suppliers into one arbitrary scoring grid.

Compare the evidence each proposal should supply

For any approach, request a representative acceptance dataset, defect taxonomy, latency histogram, bypass procedures, maintenance ownership, cybersecurity and traceability story. Ask how model updates are controlled when production conditions change. The best proposal is the one that can demonstrate fitness for your actual line, within your risk and staffing constraints.

Avoid the false precision of online rankings

This publication has not purchased or laboratory-tested the listed systems. Source links are supplied so teams can verify vendor statements. It does not assign star ratings or claim a definitive winner. Supplier corrections are welcome, but they do not determine editorial conclusions.

APPLY THIS THINKING

Make the next vendor conversation more productive.

Use the free project fit check to generate an initial list of questions for your team.

Open planning tool ↗

Primary sources and limitations

These links support the public product or research context. They do not substantiate comparative performance conclusions. Source content may change; verify directly with the supplier.