In the past, Automatic Number Plate Recognition (ANPR) has been costly, unreliable, and only relevant for a limited number of applications. Errors were common and the available hardware could not read every licence plate reliably, and so could not operate without human confirmation. Additionally, low-light areas, obstructed vantage points, and non-standard characters caused issues in detecting and capturing license plates.

Today’s ANPR cameras, engineered with deep learning algorithms and GPU processors, are extremely accurate – more than sufficient to overcome the previous challenges.

ANPR systems can “learn” number plate information within a scene, recognising a much higher number of vehicle license plates and with much higher efficiency than conventional ANPR systems. They also no longer require the frequent and often cumbersome camera positioning adjustments.