Begin with the entrance
Describe the number of lanes, direction of travel and where vehicles normally slow or stop. Identify cars, motorcycles, trucks or other vehicle types expected at the site. Test the view at the intended capture point, rather than assuming an existing overview camera can read every plate.
Make capture quality a separate check
Camera alignment, vehicle speed, shutter settings and lighting affect plate capture. Axis’s plate-capture guidance explains why these need to be considered together. The correct configuration depends on the camera and scene; a setting copied from another site is not an acceptance test.
Build a representative test set
Include the plate formats and vehicle conditions you expect during normal operation. Review day and night periods and keep a record of unreadable plates, incorrect text and successful reads. Confirm OCR output against source imagery. Do not calculate a headline success rate after silently excluding the difficult vehicles.
Define exceptions before automation
An unreadable or uncertain plate needs a fallback route. Agree how a guard confirms entry, how visitors and deliveries are handled and who maintains any authorization list. Recognition alone is not permission to open a barrier. Keep a manual override and test the behavior when the camera or network is unavailable.
Document the decision to expand
Summarize the observed results, known limitations and remaining configuration work. Confirm logging, retention, support and access-control integration separately. Expand the pilot only after the capture and operational workflow meet the agreed criteria.
Further reading
Axis license plate capture guidanceThis guide helps structure a project conversation. Confirm hardware, configuration and commercial scope for your own site.