Depth-Wise Convolution with Attention Neural Network (DWA) for Pneumonia Detection

Pneumonia can lead to necrosis of lung tissue and lung abscess. The further expansion of pneumonia infection can also cause a variety of specific complications. Pneumonia infection can spread to the pleural cavity and cause empyema. It can also lead to pleurisy, endocarditis, pericarditis, arthritis and so on. Accurate detection of pneumonia is very important. This paper proposes Depth-Wise Convolution with Attention neural network (DWA) and a set of automatic pneumonia recognition algorithm based on the deep learning method and the morphology of the lungs. Vgg16, DenseNet, ResNet are excellent models in the field of target detection, they can be trained to achieve classification task. This article uses the same data set and sets the same training parameters during training. Under a unified evaluation criteria, the VGG16, DenseNet, ResNet, and DWA models were evaluated respectively. The experimental results show that DWA has the best performance of the classification (training accuracy of 97.5%, testing accuracy of 96% and validation accuracy of 79%).