ANPR/ALPR with OpenMV Camera

OpenMV Boards
1

I would like to open the gate based on a recognized license plate.
So I want to recognize the license plate in an image, determine the size of the license plate and determine the text (OCR).
The size of the marking allows me to determine whether a car is moving towards the gate.
What is the general approach?
I tried to follow the tutorial [Coding Robin](http://coding-robin.de/2013/07/22/train-your-own-opencv-haar-
However, opencv_traincascade is disabled in OpenCV 4.x.
I can’t find anything ready-made for German license plates.

Best regards Dirk

2

Hi, this feature is not really possible on the current version of the system. We’ve done some tests though that show License Plate detection is possible via YOLOV3 running on the camera at 2.5 FPS. However, this feature is only available through edge impulse’s enterprise system.

The text detection though is not part of the YOLO network, so, that would have to be done as a second step.

New versions of the camera will have much stronger ML support where this will be easy with a network that can do everything at once.

3

Many thanks for the quick response.