1hemmem/Algerian-License-Plates-Recegnition-System

An end-to-end car detection & license plates recegnition system with OCR integration # Algerian License Plate Recegnition System Watch the demo video ## Project Description The goal of this project is to develop a comprehensive end-to-end system that detects cars and their license plates, converts the license plate information into text using Optical Character Recognition (OCR), and stores both the high-quality images of the cars and license plates along with the extracted text and a timestamp indicating when the car passed the camera. ### Challenges - The object detection model (in this case YOLO) works per frame and do not care if the same (car/license plate) is detected twice, **where in our case we care so much!!** - Working with the object tracking and re-identification (DeepSORT) - Making the OCR work well required us to ensure that we feed it the best quality image that have a readable license plate during it's path in the frame. ### Lessons - Simplicity is key: keeping solutions as simple as possible reduces complexity and improves maintainability. - Embracing Object-Oriented Programming (OOP) can save significant time and effort when implemented correctly. - RTFM: reading the flipping manual also will save much time, WALLAH !! ### Dataset You can find the dataset I have worked with in this link: mouadb0101/License_Plates_o… ## Setup clone the repository: ``` git clone git@github.com:1hemmem/Algerian-License-Plates-Recegnition-System.git ``` install the required packages: ``` cd src/ pip install -r requirements.txt ``` Run the application on your video: ``` cd prod/ && python app.py --path="your/path/to/video.mp4" ```