With the continuous development of deep learning, more and more studies are applying deep learning in the field of plant disease classification. This article uses the convolutional neural networks (CNN) and the self-attention mechanism networks (Transformer) as the perspective to sort out the relevant research on deep learning in plant disease classification. First, the basic principles and characteristics of CNN and Transformer are introduced. Secondly, the research results in recent years are reviewed, focusing on the application of CNN and Transformer in plant disease classification, and a comprehensive analysis is conducted. Finally, the challenges and future research directions of deep learning in plant disease classification are discussed, and the importance of detection speed and model lightweight in application is emphasized. This article aims to provide researchers with an in-depth summary and guidance to promote the development and application of plant disease classification techniques.