A method for tomato leaf disease identification based on an improved ConvNeXt model

This paper describes a tomato leaf disease recognition method based on an improved ConvNeXt model. The method aims to improve the model’s ability to recognize key features and feature extraction performance by introducing the local attention mechanism (ELA) and PReLU activation function. The dataset used in the study contains images of 10 tomato pests and diseases from 40 plant leaf pest categories. The local attention mechanism enhances the model’s ability to capture local features, while the PReLU activation function helps to solve the gradient vanishing problem and avoid neuron death. The experimental results show that the model achieves a high accuracy of ${9 9. 3 1 \%}$ on the tomato leaf disease classification task, which is significantly better than the traditional methods, and can effectively deal with the challenges of complex background interference and subtle and variable disease features. This research result is of great practical value for improving tomato disease identification and ensuring high yield and quality of tomato.