Enhancing the Tomato and Potato Crop Health: A Novel Approach for Ensemble Deep Learning based Disease Classification

The primary goal of this chapter is to study the current research in order to detect diseases in tomato and potato crops through the analysis of leaf images. We propose a novel methodology that involves a comprehensive data preprocessing stage, including image resizing and augmentation to enhance the dataset. We study the various pre-trained transfer learning architectures that can be used for classification. In this chapter, we discuss three ensemble models based on the best-performing architectures. Model 1 is a generalized model for both tomato and potato plants, Model 2 is specialized for potato plant diseases, and Model 3 focuses on tomato plant diseases. These ensemble models utilize a weighted average approach to calculate the results. The combination of these models is based on the highest-performing individual base models, ensuring optimal performance. The highest accuracy achieved for potato plant disease identification is 99.06%, while the best accuracy for tomato plant disease identification is 98.12%. The chapter studies and explores the efficacy of ensemble deep learning models in classifying plant diseases with accuracy, offering a robust tool for agricultural disease management.