Convolutional Neural Network based on ResNet152-V2 to predict COVID-19 disease

This repository contains the Convolutional Neural Network (CNN) model created to predict the COVID-19 disease based only in the x-ray chest scans.

Table of Contents

• covid-19-transfer-learning-resnet152v2.ipynb is the jupyter notebook with the data treatment, model creation and model validation.

Data

---

• The data comes from the following Kaggle repository : https://www.kaggle.com/datasets/pranavraikokte/covid19-image-dataset/code?datasetId=627146&sortBy=dateRun&tab=profile
• The data has two folders training and test. The training folder was split into 80% for training and 20% for validation.

Model results

---

The best model was the one who used transfer learning with a ResNet152-V2 architecture. The model was able to get 0.65 of loss and 0.92 of accuracy.

FlowChart of the Model Construction

---

  graph TD
    A[Load data] --> B(Split into train, validation, and test)
    B --> C{Create a big CNN }
    C --> D[Validate with test data]
    D -->|Yes| E[Export the model]
    D -->|No| F[Deal with overfitting]
    F --> G[Create a simpler model]
    G --> H[Validate the simpler model]
    H -->|Yes| I[Export the model]
    H -->|No| J[Deal with underfitting]
    J --> K[Use transfer learning]
    K --> L[Add new layers]
    L-->|Yes| M[Export the model]
    L-->|No| N[Repeat all again]
    

Contacts:

---

• Twitter
• LinkedIn
• ResearchGate