Research on crop disease and pest detection by combining an improved residual network and transfer learning

Timely detection and diagnosis of crop diseases and pests are crucial to agricultural production. Modern computer vision technology based on deep learning provides an efficient and automated solution for the identification of these issues. To enhance the generalization capability of deep learning models, this paper improves the ResNet50 network and incorporates transfer learning techniques to train and test on a publicly available dataset of crop diseases and pests. This approach achieves high-precision recognition of crop species, disease types, and severity levels. Experimental comparisons demonstrate that the improved network exhibits superior performance in training accuracy, feature extraction, and model generalization.