SimpleLPR

I implemented this LPR (license plate recognition) system as a assignment of a Digital Image Processing course I took in college.

I wrote a small paper with a colleague explaining the system and presenting some results. It's in portuguese, but fell free to email me or open a issue if you have any questions.

Dependencies

Use python >=3

pip install matplotlib

pip install numpy

pip install scikit-learn

pip install scikit-image

Usage

First you need to train the models:

• Extract the dataset inside imgs
• Then run python train.py (you can modify the list of models you want to train inside the file. The way it is, it will train all models one after another).

To predict, run python predict.py and follow the instructions.

Basic Functioning

When a license plate image is entered, the firt thing to do is segmente the characters inside the image. This is done by combining a series of image processing tecniques until the number of segmented regions (that oby some constaints) match the number of characters at the image label (or the number of your domain). Then, each segmented charecter is passed thought all the models that try to predict them.