Classification of COVID-19 Patients Using Deep Learning Architecture of InceptionV3 and ResNet50

Abstract

This paper aims to compare the deep learning Convolutional Neural Network (CNN) model for a case study of 3 classes chest x-ray classification of patients with "COVID-19", "pneumonia", and "normal people" using 2 architectures, namely InceptionV3 and ResNet50. This model was created using the GoogleColab platform with the Python programming language. This comparison aims to find the best results using 4 evaluation metrics and several scenarios for dividing the number of datasets used for training and validation. The evaluation metrics used include accuracy, precision, recall, and F1-score. The best accuracy is generated on a model with the ResNet50 architecture with a training accuracy value of 98.62% and accuracy validation of 96.53%. While in the InceptionV3 architecture, the resulting value for training accuracy is 96.13% and accuracy validation is 91.52%.