Skip to content
Snippets Groups Projects
Select Git revision
  • main default protected
1 result
Obeid Hanna's avatar
Update README.md
Obeid Hanna authored
1473fd6c
History
Obeid Hanna's avatar 1473fd6c
History
Name Last commit Last update
Helpers
Main
Presentation_Projet.pdf
README.md
Rapport_Projet.pdf
README.md

Automatic Number Plate Recognition (ANPR) using YOLOv7 and pytesseract

Overview

This project implements an Automatic Number Plate Recognition (ANPR) system using YOLOv7 and pytesseract on a Jupyter Notebook. YOLOv7 is used for object detection and pytesseract is used for text recognition. The system is trained on a dataset of number plates and can detect and recognize the number plate.

Repository structure

  1. Helpers directory: containing an anotation conversion script and data augmentation notebook
  2. Main directory: containing the main ANPR notebook, the images, as well as the YOLOv7 necessary files

Main Dependencies

The following libraries and packages are the main ones used in this project:

  • OpenCV
  • pytesseract
  • torch
  • numpy

Usage

To run the ANPR system, follow these steps:

  1. Clone this repository and open the ANPR_TrainYOLOv7.ipynb file in Jupyter Notebook, or use ANPR_TrainYOLOv7-colab.ipynb on Google Colab
  2. Make sure all dependencies are installed.
  3. Run all cells in the notebook to load the trained model and start the ANPR system.
  4. The ANPR system can be tested by providing an image from the /Main/taken_test_car_images directory

Results

The ANPR system is able to detect and recognize number plates with a high accuracy.

Helpers

  1. convert_voc_to_yolo.py:

    This is a Python script that converts annotated images in PASCAL VOC format into the format required by YOLO object detection. It iterates over a list of directories containing annotated images, converts the annotations for each image, and saves the converted annotations in a specified output directory.

  2. images_aug.ipynb:

    This notebook is used to read images and its corresponding label files and performs data augmentation (such as Scale, Translate, Rotate, Shear) using the data_aug library.

Useful Documents

Contact

If you have any questions or feedback, please contact: