License Plate Recognition Software 2026: Free vs Commercial

LPR (License Plate Recognition) and ANPR (Automatic Number Plate Recognition) refer to the same technology: the automatic detection and reading of vehicle number plates from camera images. LPR is the common term in the US market, ANPR in the UK and Europe. The technology combines object detection, optical character recognition and a check of the plate format.

Yes, there are open and free projects that read plates from images and video streams. They suit learning, research and a small single-camera test. For a commercial car park they usually lack an operator panel, integration with a barrier and billing, dependable updates and support, and the documentation that data protection requires.

Free software fits learning, research and a single test setup with no revenue depending on it. Commercial software fits a car park that must run reliably and where processing, barrier control and documentation are required. The decisive difference is less the price than who is responsible for running and updating the system.

Our result comes from our own September 2026 two-camera test, not from a vendor datasheet. The deployed setting confirmed 24 of 24 plates with zero false reads, with a median of 0.6 s from the vehicle appearing to confirmation. The test used 16 clips and 24 plates from an entry and an exit camera; a plate counted as correct only when both cameras agreed. That is a small sample, so we present it as a test result, not an accuracy percentage.

Common patterns are a REST interface, where software sends an image or a video source and receives plates with a confidence value, and webhooks, where the software pushes an event to another system. The recognition can also work directly on an RTSP stream. Which one fits depends on the system you already run and is scoped during the site audit.

Our system runs locally on an Android device on site and handles two 1080p cameras at once. In our tests plate detection took 38-57 ms per frame and OCR took 15-20 ms per plate crop. No separate server on the site is required.

Number plates are personal data. Local processing on site keeps the data in the country and makes the documentation simpler than processing in a foreign cloud. Our system processes locally, retention is configured by the operator, and we help with the record of processing and the information clause.

If a camera is IP-based, supports ONVIF or RTSP, and sees the lane clearly enough, it can often be reused. What matters is framing, distance and light. A site audit confirms which cameras can stay and which should be replaced. No dedicated ANPR camera is required.