COVID-19 Screening using Transfer learning model ResNet50

Abstract

The presented people of the world faced a global pandemic for the first time since 2019, which killed and affected a very large population in a short period and continues. In the addition to RT-PCR method of testing the COVID-19 disease, the radiology images can also be very beneficial in the diagnosis of this disease with the involvement of Deep Learning support, which will help in providing fastresults with several advantages through X-ray images. In this paper, we have applied the transfer learning model – ResNet50 for the Chest X-rays images dataset of a total of 42000+ images divided into 3-categories. The outcome of our experiment results is 94.89% and 92.74% accuracy for training and testing respectively. Our model has given a better accuracy on such a large set of data images and reduces the data imbalance problem concerning existing work. In the future, we will mainly work on the enhancement and segmentation of the images and will evaluate the performance of our model on the processed data.