/Automatic-Number-Plate-Recognition

Automatic License/Number Plate Recognition using YOLOv3 and LeNet Architecture

Primary LanguagePythonBSD 3-Clause "New" or "Revised" LicenseBSD-3-Clause

Maintenance Status ⚠️

This project is no longer actively maintained.

Automatic-Number-Plate-Recognition

This project is to detect and recognize the Indian License plate in images.

MODEL FILE:

Download model from the google drive https://drive.google.com/drive/folders/14e051ocMTDwH7EHP4Y7l-RkFa-LUe-vR?usp=sharing and move to the model folder.

Technologies

This project is created with

  • python 3.7

SETUP

To run this project install the requirements from the requirements.txt file.

Then run main.py file

DATASET

We created our own dataset by capturing images of vechiles in differnt viewpoint and lighting condition. We collected about over 2000 images. Then these images has been augumented.

Then all these images are manually annotated.

MODULES

Plate detection

YOLOv3 is used to detect and segment the number plate in the images. The YOLOv3 model is trained with over 3000 images in google colab.

Preprocessing

The segmented plate will be preprocessed using OpenCV

Character detection

Another YOLOv3 model is used to detect the character in the preprocessed image which has been trained by annotating the segmented images.

Character Classification

Finally the LeNet architecture is used to classify the letter and numbers in the image.