License Plate Detector & Recognizer 🚗🔍

A robust end-to-end pipeline for automatic vehicle and license plate detection, tracking, and recognition in videos. Built with YOLOv8, custom-trained models, SORT tracking, and OCR, this project enables accurate extraction and export of license plate data from traffic footage.

---

Features

• Vehicle Detection: Uses YOLOv8 COCO model for robust vehicle localization.
• License Plate Detection: Employs a custom YOLOv8 model for precise license plate detection.
• Vehicle Tracking: Integrates SORT for multi-object tracking across frames.
• OCR Recognition: Utilizes EasyOCR for reading license plate text from detected plates.
• CSV Export: Outputs bounding boxes, recognized text, and scores to CSV for further analysis.

---

Tech Stack

• Python
• YOLOv8 (via Ultralytics)
• OpenCV
• NumPy
• EasyOCR
• SORT (repo)

---

Dataset & Resources

• Custom License Plate Detector Model: Trained on Roboflow License Plate Recognition Dataset
• SORT Tracking Module: Clone from https://github.com/abewley/sort
• Sample Video for Testing: Pexels Traffic Flow Video

---

Project Structure

main.py                # Main pipeline: detection, tracking, OCR, export
utils.py               # Utility functions (OCR, formatting, helpers)
visualise.py           # Visualization and video output
models/
  license_plate_detector.pt  # Custom-trained YOLOv8 license plate model
  yolov8n.pt                # YOLOv8 COCO model
outputs/
  main.csv             # Output CSV with results
  test.csv             # Additional output
sort/                  # SORT tracking module (clone from repo)
requirements.txt       # Python dependencies
sample.mp4             # Sample input video
out.mp4                # Output visualization video

---

Installation

1. Clone this repository:

   git clone github.com/AdamNgazzou/Automated-License-Plate-Recognition-YOLOv8
   cd licence_plate_detection

2. Install Python dependencies:

   pip install -r requirements.txt

3. Clone the SORT tracking module:

   git clone https://github.com/abewley/sort.git

---

Usage

Run inference on the sample video:

python main.py

• Output CSV: Results are saved in outputs/main.csv
• Output Video: Visualization saved as out.mp4

You can adjust input/output paths in main.py and visualise.py as needed.

---

Results

• CSV Output:
  All detected vehicles, license plates, bounding boxes, and recognized text are exported to outputs/main.csv.

• Visualization:
  Annotated video with bounding boxes and recognized plates is saved as out.mp4.

• Sample Output:
  

---

Future Improvements

• Support for multi-language license plates
• Real-time inference and streaming support

---

Contributing

Contributions, bug reports, and feature requests are welcome! Please open an issue or submit a pull request.

---