Arabic Car License Plate Detection and Recognition

This project implements a deep learning-based system for detecting and recognizing Arabic car license plates. The system can handle both images and video inputs, making it versatile for real-world applications. It is built using YOLOv11 (You Only Look Once) for object detection and OpenCV for image processing, with a focus on detecting license plates and recognizing Arabic numbers and alphabets with both YOLOv11 and EasyOCR.

Features

• Car Plate Detection: Detects license plates from images and videos using a trained YOLO model.
• Arabic Characters Recognition: Recognizes Arabic letters and numbers on detected license plates.
• Video Processing: Annotates video frames with bounding boxes for plates, letters, and numbers, and displays the processed video with these annotations.
• Image Processing: Handles both full images of cars and cropped license plate images.
• Streamlit Deployment: The system can be deployed as a web application using Streamlit for real-time plate detection and recognition.

Project Structure

The project is divided into two main sections:

1. Detection: Detects the car plate in images or video frames. If the input is an image of a car, it will crop the plate using YOLOv11 and pass it to the next stage.
2. Recognition: If the input is a cropped car plate image or the plate detected from the first stage, it recognizes the Arabic characters (numbers and letters) using YOLOv11 & EasyOCR.

Installation

Prerequisites

• Python
• Streamlit
• OpenCV (cv2)
• YOLO
• EasyOCR

Install Dependencies

1. Clone the repository:

   git clone https://github.com/AhmAshraf1/sic-dl-Arabic-Car-License.git
   cd sic-dl-Arabic-Car-License

2. Install the required packages:

   pip install -r requirements.txt

Usage

Running the App Locally

1. To start the Streamlit web application, run the following command:

   streamlit run ocr_app.py

2. The web interface will allow you to upload either an image or a video for detection and recognition.Streamlit Web App

Dataset

This project utilizes an Egyptian Arabic License Plate (EALPR) dataset. If you are using a different dataset, make sure the labels are in the appropriate format for YOLO.

1. first dataset on Kaggle for detecting license car plates: Egyptian Car Plates
2. Second dataset on Roboflow for Recognition of Arabic numbers and letters: egyptian car plates Dataset

Notebook

This is out Kaggle Notebook, you can have a look for more detailed work: Kaggle Notebook

Model

The project uses the YOLOv11 Detection to detect license car plates then YOLOv11 OCR model for recognition of Arabic letters and numbers. Fine-tuning of the model was done on the EALPR dataset. The recognized characters are stored in a dictionary and drawn on the output image or video.

Deployment

The project is deployed using Streamlit for an interactive user experience, allowing users to upload images and videos for processing.

Troubleshooting

• Library Version Issues: Ensure that the correct version of cv2 and other dependencies are installed by using the requirements.txt file. If you’re encountering issues with cv2==4.10.0, consider updating or reinstalling the package.

Future Work

• Improve the accuracy of character recognition, especially for challenging cases like occluded plates.
• Add support for more types of license plates and alphabets.
• Extend the project to detect plates from different countries.