The Covid infection 2019 (COVID-19) has gotten a worldwide pandemic since the start of 2020. The disease has been viewed as a Public Health Emergency of International Concern (PHEIC) by the World Health Organization (WHO) and the finish of January 2020. Programmed recognition of lung infections from computed tomography (CT) pictures offers a mind-blowing potential to amplify the traditional clinical benefits framework for dealing with COVID-19. In any case, portioning tainted districts from CT cuts having a few difficulties, remembering high variety for contamination qualities, and low power contrast among diseases and typical tissues. Further, gathering a lot of information is illogical inside a brief timeframe period, repressing the preparation of a deep model. To kill these problems, we propose a convolutional neural network CNN) by separating worldwide and nearby highlights for recognizing and ordering COVID-19 disease from CT pictures. Every pixel from the picture is grouped into the typical and tainted tissues. For improving the characterization precision, we utilized two unique procedures including Squeeze Net for feature extraction and Resnet for classification to represent the input image differently.