This repository contains source code for cars' plates detection and recognition, which can be applied for both video and image data. Here is an example:
On the first stage we have detect() function working to put bounding box of the plate and crop the result:
On the next stage we use adjust_perspective() to bring plate back to straight look:
Detected license plate Number is:E600Ey 60.
- YOLO v4 is used as a detector pretrained on Car plates dataset (you can find it in ./data folder).
- We use trained Net to segment number in order to find the nodes.
- 3rd stage includes OpenCV algo to adjust perspective.
- Last stage consits of using any library (like pytesseract) to convert straight plate image into string.
- Done!