关键词: CNN, 视觉Transformer, 混合模型, 图像处理, 深度学习

Abstract: Convolutional Neural Network (CNN) and vision Transformer are two important deep learning models in the field of image processing. After years of continuous research and progress, both of them have made extraordinary achievements in this field. In recent years, the hybrid model of CNN and vision Transformer is gradually rising. Extensive research has continuously overcome the weaknesses of the two models, efficiently display their respective highlights, and show excellent results in image processing tasks. This paper elaborates in depth based on the hybrid model of CNN and Vision Transformer. Firstly, the architecture, advantages and disadvantages of CNN and Vision Transformer models are generally overviewed, and the concept and advantages of hybrid models are summarized. Secondly, the classification and main representative models of hybrid models are comprehensively sorted out, and their applications in specific image processing tasks are described from multiple perspectives. Finally, the future research directions of hybrid models were deeply analyzed, and the forward-looking outlook was put forward.

Key words: CNN, visual Transformer, hybrid model, image processing, deep learning