Classification of Plant Leaf Diseases Using Convolutional Neural Networks

Abstract

Leaf diseases significantly impact agricultural productivity and economic stability. This study explores the use of Convolutional Neural Networks (CNN) for classifying plant leaf diseases, addressing limitations of traditional visual inspection methods. Utilizing a Kaggle dataset with three classes (Healthy, Powdery, Rust), data preprocessing techniques such as resizing, augmentation, and normalization enhanced model performance. The CNN model achieved 95% accuracy in classification, demonstrating its capability to detect intricate patterns on leaf surfaces. Despite challenges like dataset imbalance and limited disease categories, the research highlights the potential of integrating CNN with web or mobile platforms to aid farmers in disease identification. These findings align with previous studies and underscore the importance of deep learning in agricultural innovation. Future research should focus on expanding datasets, exploring advanced architectures, and validating models under real-world conditions to maximize utility and accuracy in diverse environments