Development of a Robust CNN Model for Mango Leaf Disease Detection and Classification: A Precision Agriculture Approach

Abstract

In recent years, convolutional neural network (CNN) models and deep learning techniques have gained significant attention for plant disease detection. Despite advances, achieving high accuracy across diverse classes remains challenging. Existing CNN models have demonstrated moderate accuracy in classifying a limited number of mango leaf diseases. So, a crucial necessity exists to broaden the scope of precision. Our investigation introduces a CNN model that achieves an impressive 99% accuracy across eight classes of mango leaf diseases. Using advanced data processing, image augmentation, and feature extraction methodologies rooted in artificial intelligence and deep learning, we systematically explored over 20 CNN architectures and various hyperparameters to develop a robust model. Given the global significance of mango cultivation, our model was rigorously trained and tested for reliability. Detailed results and materials are available on GitHub. Additionally, we integrated our CNN model into an Android app, “Mango-SCN”, designed for easy use in managing mango leaf diseases, accessible even to nonexperts.