ResNet50-Based Deep Learning Model for Early Detection and Classification of Mango Leaf Diseases

Abstract

Mango is not only considered to be the “king of fruits” for its delicious, incomparable taste but is also economically important fruit throughout the world. But mango is susceptible to a number of serious diseases which seriously affect fruit quality and yield. The major diseases of mango leaves include: Anthracnose, Bacterial Canker, Die Back, Gall Midge, Powdery Mildew, and Sooty Mould. By detecting them in an early stage, it can save crops from widespread damage and ensure better fruit production. Conventional detection methods of mango leaf diseases are cumbersome, time-consuming, and prone to human error. In this perspective, the research will seek to apply a class of Conv Net algorithms, a deep learning branch of artificial intelligence characterized by high accuracies realized in image classification tasks, in attempts to meet this challenge. In particular, this study aims at designing an efficient automated system capable of classifying mango leaf images into seven classes with high accuracy value, comprehensively including healthy and diseased leaves. It considers the target for achieving a 95% classification accuracy with the purpose of ensuring a dependable early detection of the disease. This dataset consists of images of mango leaves infected with the above-mentioned diseases, where CNN models are trained to recognize patterns indicative of each condition. The model will help reduce the manual effort involved in the identification of diseases, thereby facilitating timely intervention, which is very important to maintain better mango production. Results obtained show very promising accuracy levels; hence, a favorable solution is offered for farmers and other agricultural stakeholders in managing mango leaf diseases efficiently.

Topics & Concepts