X-Ray Image Classification of Covid-19 Using Transfer Learning with Vgg16 and Vgg19

COVID-19 is a contagious disease that continues to be a scourge in the developing world. Chest radiology has also been proven useful for detecting abnormalities in patients’ lungs. Motivated by this, our methods propose a quick and automated detection system based on deep learning as a secondary COVID-19 diagnosis option. An imbalanced chest X-ray dataset called the COVID-Xray-5k dataset, containing X-ray scans of subjects diagnosed with COVID-19 and healthy individuals, has been used in the investigation. Transfer learning was used with VGG16 and VGG19, and then CNN model structures were proposed and parameters were tuned. One of the proposed CNN models classifies the test dataset with an F1 score of 0.91 and an accuracy of 99.45%. Proposed methods could also help with basic COVID-19 variant identification as long as a large enough dataset is given, even if it isn't balanced.  To diagnose COVID-19, a proposal of two novel deep-learning sequential architectures, both of which are based on the conventional methods of convolutional neural networks.Employing machine learning to investigate subjective feature engineering and transfer learning was used for VGG16 and VGG19, with and without weighted sampling, to reliably classify COVID-19.The whole investigation supports dealing with an imbalanced dataset.