Covid-19 Detection Through Cough Sounds: A Comparative Study Using XGboost and RESNET

Coronavirus disease (COVID-19) which is a rapidly spreading virus disease is considered as a huge crisis that the world is facing today, firstly happened in Wuhan city in China in the year 2019 December and within months spread to other countries. To date, total death in the world due to covid-19 exceeded 6 million and in India over 5 lakhs of lives have already vanished due to the virus attack. The losses are compounding every day. Diagnosis of the virus is most important because most of persons may be asymptomatic. Surprisingly, it has been listed by many studies that cough is the main reason for spreading the virus easily. In this paper, the virus detection is done using neural networks and a machine learning algorithm using recorded cough sounds. A virufy dataset containing recorded cough samples is taken for the research. Deep learning has been done using ResNet50 and XGBOOST is the classification algorithm.