COVID-19 Radiography Classification and Visualization

This project aims to build a deep learning model for classifying and visualizing COVID-19 radiography images. It utilizes the VGG19 model with a Convolutional Block Attention Module (CBAM) to extract features from images and employs Grad-CAM for visualizing model predictions.

Key Features

• Classification of COVID-19 radiography images
• Visualization of model predictions using Grad-CAM
• Application of various image filtering techniques
• Model evaluation through K-fold cross-validation
• Experiment tracking with Weights & Biases

Installation and Setup

Requirements

• Python 3.x
• PyTorch
• torchvision
• wandb
• numpy
• matplotlib
• opencv-python

1. Clone this repository:

   git clone https://github.com/yourusername/COVID19-Radiography-Project.git
   cd COVID19-Radiography-Project

2. Set up and activate a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows, use `venv\Scripts\activate`

3. Install the required packages:

   pip install -r requirements.txt

Usage

1. Run train_test_main.py to train the model:

   python train_test_main.py

2. Run grad_cam_plot_main.py to perform Grad-CAM visualization:

   python grad_cam_plot_main.py

3. Run kfold_main.py to perform K-fold cross-validation:

   python kfold_main.py

Contributors

• JUN-SU PARK
  • Email: [email protected]

License

This project is licensed under the MIT License. See the LICENSE file for details.