Detection of COVID and Viral Pneumonia: A Transfer Learning Model-Based Approach

Abstract

A worldwide epidemic known by the name Covid-19 is said to have posed the greatest threat to public health in the last century and resulted in catastrophic health, social, and economic disasters. Fever and chills were this virus' early symptoms. Although the symptoms of viral pneumonia are extremely similar to the second phase of the virus-caused coughing and acute shortness of breath, the two illnesses require entirely distinct approaches to treatment. Sometimes, because of a lack of appropriate diagnostic tests, patients are unsure whether they have Covid-19 or pneumonia. As a result, a dangerous treatment like death is possible. In this study, MobileNetV2 and RestNet50, two well-known transfer learning approaches, were used and tested to see which model performed the best in properly classifying both of these illnesses from X-ray pictures. After examining the performance, MobileNetV2 had the best accuracy, which was around 95% as opposed to RestNet50's 80%.