Detection of COVID-19 based on attention mechanism and residual network

Corona Virus Disease 2019(COVID-19) is a respiratory disease whose chest X-ray (CXR) or computer tomography (CT) images are different from those of ordinary people which providing a basis for in-depth learning method detection. The article proposes a Residual Network (ResNet) pneumonia detection model based on the ResNet50 network which integrates Hierarchical Split Block (HS-Block) and Convolutional Block Attention Module (CBAM) during the feature extraction process to enhance lesion features. The datasets are collected and integrated which containing 31883 CXR and CT images tor verify the effectiveness of the model. The experimental results show that the average accuracy of the model reaches 98.8% which indicate that incorporating HS-Block and CBAM modules into the residual network can improve the performance of pneumonia detection.