Glossary · Machine vision
Camera-based inspection
Also known as: Vision inspection, Automated optical inspection
German: Kamerabasierte Inspektion
In manufacturing, camera-based inspection is the automated checking of products or process results from camera images, for example for presence, completeness, dimensions, surface defects or correct labeling, with the result used to accept, reject or sort parts.
- Machine vision
In one sentence
Camera-based inspection checks products automatically from images for presence, dimensions, defects or labels, and triggers accept or reject.
Example
Camera-based inspection at the end of the capping line rejects bottles with missing or skewed caps before they reach the case packer.
How it applies
- Engineering: An inspection task starts with a clear specification of what counts as a defect, with sample parts. Lighting and optics are designed to make that defect visible before choosing algorithms.
- Commissioning: Inspection performance is measured with good and bad parts: false rejects cost yield, missed defects cost quality. The accepted rates are agreed as an Acceptance criterion.
- Documentation: The documentation team should document the inspection criteria, the sample sets, the test results and the procedure for changing limits, and state which checks are relevant to product quality or safety. For AI-based inspection, record how the model was evaluated, see Evaluation (evals).
Camera-based inspection vs. manual visual inspection
Camera-based inspection is repeatable and fast but only finds what it was designed and tested for. Manual inspection is flexible but varies between people and shifts. Many lines combine both for rare defect types.