A Comparative Study of CNN, ViT, and ResNet50 for Pneumonia Detection in Medical Imaging

Pneumonia is a common respiratory disease. Early recognition of pneumonia is crucial for treatment and prevention. We use Chest X-ray rays (CXR) to identify pneumonia. With the development of deep learning technology, its application in the field of medical image recognition is more and more extensive. This paper aims to explore the application of convolutional neural network (CNN), Visual Transformer (ViT) and ResNet50 in pneumonia recognition, and compare their performance. By using these methods, the Resnet50 model achieves the highest accuracy of 98.5%, within an average accuracy of 95.4%.