This project is based on License Plate Recognition on this github: https://github.com/MicrocontrollersAndMore/OpenCV_3_License_Plate_Recognition_Python
The following video explains general process: https://www.youtube.com/watch?v=fJcl6Gw1D8k
What has been added:
- Char generation from real plate and char augmentation
- Specific model to recognize Vietnam's motorbike plate
- Use CNN model instead of KNN model
- Add more constraints to increase results
You can also see DocsAndPresentation folder for more information
To recognize character in an image
python Main.py detect ./test/train_data/plate_2.jpg --steps True --save True
--save=True
save image with recognized chars in outputs
--steps=True
will show step by step recognition process
To generate chars from a font. All fonts contained in fonts
folder will be converted to template
python GenChars.py --font True
To get trained chars from real plate
python Main.py gen TRAIN_PLATE_FOLDER REAL_CHARS_FOLDER
To augment chars from real chars. Check required folders in GenChars.py
python GenChars.py --gen_char 10
If you want to retrain model by yourself, you can check Train_CNN_Model.ipynb