CNG483 Dataset Processing

Description

This project processes a dataset for license plate recognition, extracting features and labels from images organized in different plate configurations (1x1, 1x2, 2x2). The main script generates histogram features and proposed features for training, validation, and testing sets.

Installation

1. Ensure you have Python installed (version 3.6 or higher recommended).
2. Install the required dependencies:

   pip install -r requirements.txt
   

   Or if using pyproject.toml:

   pip install .
   

Usage

Run the main script to process the dataset and generate features:

python main.py

This will create the feature files in the outputs/ directory.

Dataset Structure

• Plate_1x1/: Contains test, train, and validation folders for 1x1 plate images.
• Plate_1x2/: Contains test, train, and validation folders for 1x2 plate images.
• Plate_2x2/: Contains test, train, and validation folders for 2x2 plate images.

Outputs

The outputs/ directory contains the following NumPy arrays:

• hist_features_training.npy: Histogram features for training data.
• hist_features_validation.npy: Histogram features for validation data.
• hist_features_testing.npy: Histogram features for testing data.
• proposed_features_training.npy: Proposed features for training data.
• proposed_features_validation.npy: Proposed features for validation data.
• proposed_features_testing.npy: Proposed features for testing data.
• labels_training.npy: Labels for training data.
• labels_validation.npy: Labels for validation data.
• labels_testing.npy: Labels for testing data.

Requirements

See requirements.txt for the list of Python packages needed.

Troubleshooting

• Ensure all dataset folders are present and contain the expected images.
• If you encounter import errors, verify that all dependencies are installed.