NumberPlate_project

A Python implementation of vehicle number plate detection using the Haar Cascade algorithm in OpenCV. Built as a 2nd-year engineering project, this repo detects license/number plates in vehicle images and logs the cropped plate images along with detection data.

Overview

The project uses a pre-trained Haar Cascade classifier to locate number plates within an input image or video frame. Once a plate region is detected, it is cropped, saved, and logged for further processing (e.g. OCR, record-keeping).

Project Structure

NumberPlate_project/
├── Scripts/             # Core Python scripts for detection
├── model/                # Haar Cascade classifier (.xml) file(s)
├── plates/               # Output folder for cropped/detected plate images
├── number_plates.xlsx    # Log of detected plates (e.g. image name, timestamp)
├── requirements.txt      # Python dependencies
└── README.md

How It Works

1. An input image or video frame is loaded.
2. The frame is converted to grayscale (Haar Cascades work on grayscale intensity gradients).
3. The pre-trained Haar Cascade classifier (stored in model/) scans the frame for plate-like rectangular regions.
4. Detected regions are cropped and saved to the plates/ folder.
5. Metadata about each detection is recorded in number_plates.xlsx.

Requirements

Install dependencies with:

pip install -r requirements.txt

Typical dependencies for this kind of project include:

• opencv-python
• numpy
• openpyxl / pandas (for writing to the Excel log)

Check requirements.txt for the exact pinned versions used in this project.

Usage

1. Clone the repository:

   git clone https://github.com/ilovevampire/NumberPlate_project.git
   cd NumberPlate_project

2. Install dependencies:

   pip install -r requirements.txt

3. Run the detection script (from the Scripts/ folder):

   python Scripts/<script_name>.py

4. Detected plates will be saved in the plates/ directory, and detection records will be appended to number_plates.xlsx.

Replace <script_name>.py with the actual entry-point script inside the Scripts/ folder.

Model

The model/ directory contains the Haar Cascade XML file used for plate detection (e.g. haarcascade_russian_plate_number.xml or a custom-trained cascade). Haar Cascades are lightweight, fast object detectors well-suited for real-time or low-resource applications like this one, though they are less robust than modern deep learning detectors (e.g. YOLO) under varying lighting, angles, or plate formats.

Output

• plates/ — cropped images of each detected number plate.
• number_plates.xlsx — a spreadsheet log of detections (filenames, counts, and/or timestamps depending on script implementation).

Future Improvements

• Integrate OCR (e.g. Tesseract or EasyOCR) to read the actual plate text rather than just detecting the region.
• Replace the Haar Cascade with a deep learning-based detector (YOLOv8/SSD) for improved accuracy across angles, lighting, and plate styles.
• Add support for real-time detection via webcam/video stream.

Author

ilovevampire

License

No license specified. Feel free to reach out to the author regarding usage and reuse.