Pay Attention For Covid-19 Detection Using Efficient Convolution

AI techniques, especially Deep Learning, are efficient and reliable for COVID-19 screening. During diagnosis, other lung infections can overlap with COVID-19 lung infections. It makes the human-centered diagnosis of COVID-19 challenging. Convolutional Neural Networks (CNNs), can facilitate the diagnosis of positive COVID-19 cases extremely easy and swift. Our research presents an alternative modeling framework, a new deep learning model(COVIDAT-Net), for faster and reliable screening from x-ray images. For building a robust, lightweight deep learning model, we implemented a Convolutional Block Attention Module(CBAM)\cite{woo2018cbam} with Separable convolution \cite{chollet2017xception}. Finally, we compared two-state of art model performance with our model for detection COVID-19.