Automatic disease identification has turned into a crucial area of study in medical science as a result of the current rapid population growth. A method for automatically recognizing diseases aids in medical diagnosis, generates reliable data quickly, and lowers the death rate. As a result of its recent global expansion, the COVID-19 is currently one of the utmost severe and critical infections. The quickest means of diagnosis should be developed in the form of an automated detection system to stop the spread of COVID-19. The international community works to identify and treat COVID-19 patients as soon as feasible by looking for, implementing, or developing novel strategies. In order to effectively diagnose and identify sick tissues in Covid-19 patients based on pulmonary medical imaging, we apply deep learning algorithms. The suggested technique predicts output labels using a Convolutional Neural Network (CNN) architecture. We try to use classification and feature extraction based on CNN techniques in order to assess the sensitivity and accuracy of the strategy. We try to use classification and feature extraction based on CNN techniques in order to assess the sensitivity and accuracy of the strategy. In order to achieve high efficiency and accuracy, we implement various pre-existing models and suggest an application using DenseNet201, VGG16, and Xception architectures that uses radiographic pictures to detect and classify the potential existence of SARS-CoV-2.