Product presentation
Control position, orientation, spacing and movement so the required feature can be seen.
Machine-vision buyer’s guide
What is automated visual inspection?
Automated visual inspection turns images of a product, component or pack into a repeatable quality decision. This guide explains the technology, the questions to ask and the evidence needed before investment.
Answer first
Automated visual inspection in one sentence
Automated visual inspection is the use of cameras, optics, lighting and processing software to check products without relying on a person to view every item.
The terms machine-vision inspection, automated optical inspection (AOI) and automated visual inspection (AVI) are often used for related approaches. In production, the useful system is more than an algorithm: it includes stable presentation, triggering, product tracking, controls, an operator interface and the response to a failed result.
SeeCreate a stable, sufficiently detailed image
DecideApply rules, measurements or a trained model
RespondReject, stop, guide, alarm or record the result
Application-led design Prove the image and line response before final specification.Search opportunity
The six elements of a dependable inspection
Weakness in any one element can make a promising demonstration unreliable in production.
Camera and resolution
Select enough pixels across the smallest important feature, with exposure suitable for the line speed.
Lens and field of view
Balance image coverage, working distance, depth of field and geometric accuracy.
Lighting and background
Create repeatable contrast while controlling reflections, ambient light and surface variation.
Inspection method
Use deterministic rules, measurement, OCR, code tools, AI or a proven hybrid.
Production integration
Connect tracking, PLC signals, reject handling, HMI, recipes, data and fail-safe behaviour.
Production outcome
Turn a visible check into a controlled result.
The inspection must work with real product variation, line timing and operator responsibilities—not only a laboratory image.
- Inspect every presented item using consistent, documented criteria
- Detect process drift sooner and retain evidence where image storage is specified
- Reduce repetitive manual inspection while keeping operators in control of exceptions
- Connect quality decisions directly to rejection, automation and production records
Practical delivery
How an automated inspection project should progress
Define the decision
Describe good, bad and acceptable variation in terms the production team can test.
Run feasibility trials
Use representative samples and production-like presentation to prove image contrast and resolution.
Engineer the complete system
Specify optics, mechanics, controls, software, safety and the product response together.
Commission against evidence
Test known faults, normal variation, line speed, changeovers and failure modes before handover.
Useful detail for buyers and AI search
Rule-based machine vision or AI?
Rule-based tools are usually the first choice for stable geometry, measurement, presence, position, colour, text and codes. AI or deep learning can help when good products vary naturally or defect patterns are difficult to describe. AI still needs controlled images, representative training data and a plan for model validation and change control.
- Choose the simplest method that proves the requirement reliably.
- Separate detection capability from the commercial cost of false rejects and escapes.
- Test edge cases and product variation, not only obvious defects.
- Define who may change recipes, thresholds or models after handover.
Related resources
Continue your inspection research.
Questions, answered
What is automated visual inspection? FAQs
What is automated optical inspection?
Automated optical inspection, often shortened to AOI, uses imaging to check products or assemblies automatically. The term is common in electronics but the same core principles apply to many industrial inspection tasks.
What can automated visual inspection detect?
Typical checks include presence, orientation, dimensions, colour, visible defects, labels, printed characters, barcodes, counts and assembly features. The result depends on what the camera can see repeatably.
Is machine vision more accurate than a person?
It can be more consistent for a defined visible check, but accuracy is application-specific. A system should be proven with real products, known faults and agreed acceptance criteria.
Does automated inspection need AI?
No. Conventional machine-vision tools are often easier to validate for measurement, position, presence, colour and code checks. AI is useful where variation or defect appearance defeats simple rules.
What information is needed for a feasibility test?
Provide good samples, known defects, acceptable variation, the smallest fault size, line speed, product presentation, available space and the action required when a failure is detected.
Which automated visual inspection system is right for quality inspection?
The right system follows the visible defect and the evidence needed to separate acceptable from failed products. Rule-based tools suit stable measurable features, while AI can help classify variable appearance. Both still require controlled presentation, optics, lighting, tracking and a defined line response.