Radiography Images with Transfer Learning on Embedded System

Abstract

A serious public health concern is Novel Coronavirus Disease (COVID-19), which spread quickly over the globe at the end of 2019. This coronavirus is still able to propagate rapidly even after two years. Chest X-rays are crucial for diagnosing infected individuals in the worldwide battle against this illness. Therefore, various COVID-19 quick classification technologies can provide excellent classification accuracy to help medical professionals make the best choices. Here, we propose a trustworthy, compact network that, with the aid of encouraging classification results, can correctly identify COVID-19 from chest X-rays. The experimental findings demonstrated that, in a low-power embedded system, the modified architecture of the proposed model produced excellent performance metrics for four classes. The suggested classification architecture had an overall accuracy speed of 97.67% and an f1-score of 97.64%. This classification model is better than the other classification models used to classify patients with COVID-19 infection.