Potato plant leaf disease classification using an Enhanced Deep Learning Model

Abstract

Potato is one of the most important vegetable crops cultivated worldwide, but its production is often threatened by various leaf diseases, particularly early and late blight caused by Alternaria solani and Phytophthora infestans. These diseases can significantly reduce crop yield and quality, and farmers often rely on visually detecting changes in potato leaf color, which can be time-consuming and unreliable. To address this problem, there is a need for computer-aided techniques that can accurately and quickly identify these diseases, even in their early stages. In this paper, we propose a deep learning approach called Enhanced EfficientNET, which uses the Enhanced EfficientNET network to recognize various types of potato leaf disorders. To enhance the model's recognition ability, we introduce a spatial-channel attention method that focuses on the damaged areas. We also address the issue of class-imbalanced samples and improve the model's generalization ability by tuning the EANet model using transfer learning and adding dense layers to enhance its feature selection power. We test the model on a challenging dataset called PlantVillage, which contains images taken in diverse and complicated background conditions and achieve an accuracy of 98.12% for classifying various potato plant leaf diseases. Our experiments demonstrate the effectiveness of our approach in robustly tackling distorted samples and classifying potato plant leaf diseases.