Skip to content

Latest commit

 

History

History
46 lines (41 loc) · 2.02 KB

README.md

File metadata and controls

46 lines (41 loc) · 2.02 KB
Raw

Number-Plate-Detection

This repository contains Python scripts for detecting and recognizing number plates using Support Vector Machines (SVM) and Block Binary Pixel Sum descriptors.

Overview

The project consists of two main scripts:

  1. train_simple.py: Trains classifiers for alphabet characters and digits using a dataset of font images.
  2. recognize.py: Uses the trained classifiers to detect and recognize characters in number plates from images.

Features

  • Font Image Processing: Converts font images to grayscale and applies thresholding.
  • Contour Detection: Identifies and sorts contours in the images.
  • Feature Extraction: Uses Block Binary Pixel Sum descriptors to extract features from the regions of interest (ROIs).
  • Model Training: Trains SVM classifiers for both alphabet characters and digits.
  • Model Serialization: Saves the trained models to disk for later use.
  • Number Plate Detection: Detects number plates in images using the trained classifiers.
  • Character Recognition: Recognizes and displays characters from detected number plates.

Installation

  1. Clone the repository: 
    git clone https://github.com/cizodevahm/Number-Plate-Detection.git
  2. Navigate to the project directory: 
    cd Number-Plate-Detection

Usage

Training the Model

  1. Prepare your fonts dataset and place it in a directory.
  2. Run the training script: 
    python train_simple.py --fonts path/to/fonts --char-classifier output/char_classifier.cpickle --digit-classifier output/digit_classifier.cpickle
  3. The trained models will be saved in the specified output paths. Recognizing Number Plates
  4. Place your images in a directory.
  5. Run the recognition script: 
    python recognize.py --images path/to/images --char-classifier output/char_classifier.cpickle --digit-classifier output/digit_classifier.cpickle
  6. The script will display the processed images with detected and recognized number plates.

License

This project is licensed under the MIT License.