Publication

Abstract : This work proposes the Covid-DWNet deep learning-based architecture for the quick identification of Covid-19 and other symptoms from chest CT and X-ray images. Depth-wise dilated convolutions (DDC) units and feature reuse residual block (FRB) units form the foundation of the architecture, which effectively extracts a variety of features from the chest scan pictures. The proposed architecture greatly enhances the ability of CT images and X-ray images to recognize Covid-19 and other pulmonary diseases. In addition, Skip connections were introduced from the first Feature Residual Block layer to the last Feature Residual Block layer for the retention of features in the tensors. An accuracy of 98.44% is achieved on X-ray images as compared to 96.8% in the Covid-DWNet architecture. In addition, Skip connections were introduced from the first Feature Residual Block layer to the last Feature Residual Block layer for the retention of features in the tensors. An accuracy of 98.44% is achieved on X-ray images as compared to 96.8% in the Covid-DWNet architecture.