kaylan21k/South-African-Bank-Coin-Recognition-Project

# South African Bank Coin Recognition Project A computer vision system for accurate recognition and classification of South African bank coins using deep learning techniques. ## Model Training Notebook **The model training notebook is a crucial component of this project. It contains the complete pipeline for developing the coin recognition system, including:** - Processing 714 original coin images - Performing multiple augmentations to increase dataset size to 7,854 images - Image preprocessing and enhancement - Image segmentation - Feature extraction - Coin classification (both denomination and side) - Model comparison and selection **Access the notebook here:** COMP702_PROJECT.ipynb **Original images required by notebook(COMP702_PROJECT.ipynb) here:** original_coins.zip ## Important Note The application requires model files located in the `models_final.zip` archive. You must extract these files before running the application. **1. Unzip the Archive** Open a terminal or command prompt in the project's root directory and use the appropriate command for your operating system. * **On macOS or Linux:** ```bash unzip models_final.zip ``` * **On Windows (using PowerShell):** The `tar` command is generally more reliable for avoiding nested directories. ```powershell tar -xf models_final.zip ``` **2. Verify the Directory Structure** After unzipping, you must have a single `models_final` directory containing the model files. Please verify that your folder structure looks like the Project Structure Below. ## Project Structure ``` South-African-Bank-Coin-Recognition-Project/ ├── README.md ├── image_processing_pipeline.py ├── main.py ├── models_final.py (After Unzipped) │ ├── feature_names.json │ ├── random_forest_side_model.joblib │ ├── random_forest_type_model.joblib │ ├── scaler.joblib ├── model_loader.py ├── models_final.zip (must be unzipped before running) └── requirements.txt ``` ## Installation Instructions ### Prerequisites - Python 3.7 or hi …