Comparative Analysis of Transfer Learning Models for The Detection and Classification of Grape Leaf Diseases

Grape cultivation is a vital part of India’s agricultural sector, contributing significantly to the national economy and the livelihoods of many farmers. However, grape production faces ongoing threats from various diseases, mainly fungal, which thrive in the humid, moderate climate from January to May, causing substantial yield losses. Traditional disease detection methods, like manual inspection, are often inadequate due to their time-consuming nature, high costs, and susceptibility to human error. To address these challenges, this study compares several state-of-the-art transfer learning (TL) models, including VGG16, VGG19, InceptionV3, MobileNet, MobileNetV2, ResNet50, and Xception, to automate the detection and classification of grape leaf diseases. Among these models, MobileNet demonstrated superior performance, compared to other TL models. These findings highlight MobileNet’s effectiveness for accurate disease detection in grape cultivation, offering a promising solution to enhance the efficiency and reliability of the detection process. By enabling early intervention, this approach can reduce the impact of diseases on grape yields, thus supporting the economic stability of farmers and promoting sustainable grape farming practices in India.