Diseases and pests seriously affect the sustainable development of agriculture and the environment， leading to crop yield loss and quality decline. Deep learning technology provides new methods for identifying and controlling pests and diseases， with unique advantages in recognition accuracy and efficiency. On the basis of exploring and learning the characteristics of deep learning technology and the advantages and disadvantages of algorithms， this article explores its application in the research of pests and diseases in three economic crops： tomatoes， grapes， and apples. It mainly analyzes the accuracy of ResNet network model in recognizing and classifying pest and disease images of these crops， and analyzes the training loss， validation loss， and validation accuracy of ResNet network. The experimental results demonstrate that the ResNet network model has a high recognition accuracy of 94.54% for pest and disease images.