Deep Learning and Transfer Learning-based Detection of COVID-19 using Radiography Images

Abstract

Covid-19 has been fought on several fronts and in various forms. Many aspects of coping with the pandemic may be dealt with using computerized methods; Physical sampling, vital sign monitoring, and disinfection are just a few examples of robotic solutions. Furthermore, image recognition and artificial intelligence are being utilized to confirm cases. In this work, we present a deep learning and transfer learning method for automatically diagnosing COVID-19-related pneumonia from X-ray pictures. We used these approaches in our work, Convolutional neural network (CNN), MobileNetV2, and Residual Network(ResNet50) are three deep learning architectures we presented . We also look at which one is the most accurate. This research focuses on the Covid-19 viral identification work, which is concerned with finding the virus in chest X-ray pictures. The CNN model(96.58%) outperformed the MobileNetV2 (94%) and ResNet50 (92.83%) models in every performance metric