Comparison of Machine Learning and Deep Learning Classification Models for Apple Leaf Disease Detection

Abstract

Diameter variations in the leaf's visual characteristics enable the differentiation of diseased conditions; thus, leaves function as distinguishing indicators. To facilitate disease detection, the ability to identify the distinct patterns produced by these pathogens on the foliage is crucial. Historically, the physical examination of plants has been delegated to experts or cultivators. Nevertheless, this methodology may require a substantial financial investment in addition to a considerable labor force. Given the circumstances, it is critical to prioritize the implementation of automated techniques for detecting agricultural maladies, particularly in regions with limited access to specialists. By employing five distinct classification algorithms, this study aims to develop a model capable of discerning the presence of diseases on apple leaves. The following are the methodologies: Inception V3, Support Vector Machine (SVM), Random Forest, and Decision Tree. The approach currently under contemplation conducts a comparative analysis of machine learning and deep learning models in order to identify diseases that manifest in apple leaves. Apple Rust, Apple Scab, and Apple Spot are the particular products that are the focus of the attack. An examination is conducted on the "Apple Leaves Disease Dataset" with respect to apple leaves. When compared to all other fitted models, VGG19 has demonstrated the highest level of test accuracy ever established, at an impressive 95 percent.