RFQ Checklists

Machine Vision Acceptance Test Checklist

Define machine vision acceptance tests with good/bad samples, repeatability checks, false-pass risk, read-rate targets, station timing and integration handoff notes.

Machine vision acceptance test bench with industrial camera lighting sample parts and validation checklist

Direct answer

Machine Vision Acceptance Test Checklist

A machine vision acceptance test should prove repeatable production evidence, not only a successful demo image. Define sample set, pass/fail criteria, repeatability run, read rate or measurement tolerance, station timing, I/O output and documentation before shipment.

Where this matters

Start with the inspection condition.

Acceptance should be measurable: sample count, false-pass limit, false-reject target, read rate, measurement repeatability, timing and handoff files.

Why projects fail

Confirm the limits that change hardware.

Check repeatability across part position, surface variation and speed.

RFQ preparation

Send enough context for a real review.

Acceptance criteria should be measurable before shipment.

What engineering should check

What this page should help teams decide.

Use real good, bad and borderline samples.
Check repeatability across part position, surface variation and speed.
Acceptance criteria should be measurable before shipment.

Related Deyi routes

Product families commonly involved.

Practical note

Build the sample set before judging the system.

A useful acceptance set includes good parts, clear failures, borderline defects, surface variation and position variation. If the sample set is too clean, the system can pass the demo and still fail at production handoff.

Practical note

Measure repeatability, not only detection once.

Run the same samples repeatedly across expected fixture positions, lighting warm-up, speed changes and operator loading patterns. The result should state false pass, false reject, read rate or measurement spread.

Practical note

Confirm station timing and output behavior.

Acceptance should include trigger timing, exposure, processing time, PLC output, reject delay, image saving, alarm behavior and recovery from no-read or no-detect events.

Practical note

Document assumptions for future maintenance.

Delivery notes should record camera model, lens, working distance, light angle, exposure, fixture references, sample images and acceptance criteria so later maintenance does not change the route accidentally.

How to test before buying

Use this guide as a pre-RFQ decision filter, not as a part-number shortcut.

Machine vision selection is usually stable when the project starts from the inspection condition instead of a catalog model. Before requesting a quote, define what must be detected or measured, how the part moves, what surface behavior affects contrast and which factory constraint cannot change.

Use this guide to translate the requirement into testable inputs: sample images, target tolerance, line speed, field of view, working distance, mounting envelope and the current failure mode. That gives the factory enough evidence to map the request to camera, lighting, optics, reader or 3D routes.

Decision checks

Three checks before locking the route.

Sample evidence

Use good, bad, borderline and variation samples, not only one clean target.

Repeatability run

Repeat the same samples across position, speed and surface variation.

Output timing

Check trigger, processing, reject delay and PLC/database handoff.

Decision table

Use these data points to turn the concept into an RFQ-ready decision.

Factor	Practical rule	RFQ impact
Sample evidence	Use good, bad, borderline and variation samples, not only one clean target.	Send sample list and defect definition before testing.
Repeatability run	Repeat the same samples across position, speed and surface variation.	Define sample count, repeat count and acceptable result spread.
Output timing	Check trigger, processing, reject delay and PLC/database handoff.	Prevents a vision pass that cannot operate in the real line.
Handoff notes	Record optics, lighting, fixture, settings and acceptance limits.	Makes commissioning and maintenance safer.

Application proof

Related delivery routes that make this selection decision concrete.

View all cases

Electronics inspection

PCB Connector Pin Presence Inspection

Small connector features require stable contrast, low distortion and repeatable part positioning before a pass/fail threshold can be trusted.

Battery inspection

Battery Tab Weld Height Inspection

Battery tab weld inspection needs geometry confidence, stable position reference and traceability under high station throughput.

Automotive parts inspection

Automotive Metal Surface Defect Inspection

Reflective metal surfaces can hide scratches and burrs unless the lighting geometry is matched to the defect direction and part movement.

Common mistakes

Problems that slow down selection.

Selecting by model number before the inspection target is measurable.
Treating lighting as an accessory instead of the main contrast-control tool.
Ignoring fixture stability, part variation and operator maintenance workflow.

Factory handoff

What Deyi Vision reviews after receiving the project details.

The factory route review starts by checking whether the image can be made stable with lighting and fixture control. Then the camera, lens, reader or 3D sensor route is sized against speed, resolution, interface and installation constraints.

If you already have a Keyence, Cognex, Basler, OPT, LMI, Hikrobot or barcode-reader reference, include it as a reference model. Deyi Vision uses it to understand the application class; final selection still depends on real samples and production limits.

Guide to RFQ

Have a real part, sample image or production constraint?

Use the guide to frame the question, then send the details so engineering can recommend a route.

Request engineering RFQ

Guide FAQ

Questions related to machine vision acceptance test checklist.

Ask engineering

What is included in a machine vision acceptance test?

It should include sample set, pass/fail definition, repeatability run, false-pass or false-reject target, read rate or measurement tolerance, station timing, I/O output and handoff documentation.

Why is one successful demo image not enough?

A single demo image does not prove repeatability across part variation, fixture movement, lighting changes, speed, surface finish or operator loading.

What should I prepare before factory acceptance?

Prepare good, bad and borderline samples, defect definitions, tolerance targets, line speed, trigger and output requirements, current failure examples and the final acceptance threshold.