ET-NET_Covid-Detection

Based on our paper "ET-NET: An Ensemble of Transfer Learning Models for Prediction of COVID-19 Infection through Chest CT-scan Images" published in Springer- Multimedia Tools and Applications.

Required Dependencies

To install the dependencies, run the following using the command prompt:

pip install -r requirements.txt

Running the code on the COVID data

Download the dataset from Kaggle and split it into 5-fold cross-validation train and validation sets.

Required Directory Structure:

+-- data
|   +-- .
|   +-- train
|   +-- val
+-- utils
|   +-- .
|   +-- utils_cnn
|   +-- utils_ensemble.py
+-- main.py

To run the ensemble model on the base learners run the following:

python main.py --root "path/"

Available arguments:

• --epochs: Number of epochs of training. Default = 100
• --batch_size: Batch Size. Default = 4
• --num_workers: Number of Worker processes. Default = 2
• --learning_rate: Learning Rate. Default = 0.001
• --momentum: Momentum value. Default = 0.99

Citation

If this repository helps you in any way, please consider citing our paper:

Kundu, Rohit, et al. "ET-NET: an ensemble of transfer learning models for prediction of COVID-19 infection through chest CT-scan images." Multimedia Tools and Applications (2021): 1-20.

Bibtex:

@article{kundu2021net,
  title={ET-NET: an ensemble of transfer learning models for prediction of COVID-19 infection through chest CT-scan images},
  author={Kundu, Rohit and Singh, Pawan Kumar and Ferrara, Massimiliano and Ahmadian, Ali and Sarkar, Ram},
  journal={Multimedia Tools and Applications},
  pages={1--20},
  year={2021},
  publisher={Springer}
}