I created a model that will look at cars, detect where's the license plate, segment that out and 'read it', return what it read in string format using Tesseract and Opencv in Python .
The main method comprises of 2 parts. The objective of the first part is to be able to detect where's the license plate on the car. Using OpenCV, I retrieved the coordinates for where the model detects rectangular boxes. I had to filter these out by the size of the license plate. After sieving through all the non-license plate coordinates, i created a new image of the zoomed license plate. This will make it easier for the next step, which is to read the numbers off the picture. We zoom into the license plate so as to reduce noise and non-essential information.
Now that we have a zoomed image of the license plate, we implement a module that is able to segment the individual characters in the picture and churn out a string. Now that the bulk of the project is done and dusted, all i had to do was to clean the string for the license plate numbers only.
Submitted by Thimmareddy Venkata Hemanth Reddy (hemanthreddy)
Download packets of source code on Coders Packet