A COMPREHENSIVE EVALUATION OF MOBILENET ARCHITECTURE FOR TOMATO DISEASES.

Abstract

The potential of deep learning models to automate and enhance various processes has garnered significant attention for their use in agricultural applications in recent years. One notable application is the use of convolutional neural networks (CNNs) for classifying plant diseases. An extensive assessment of the MobileNet architecture for the task of classifying tomato diseases is presented in this research. Because of its lightweight architecture, MobileNet is renowned for its effectiveness and adaptability for embedded and mobile devices. We use a publicly available dataset to investigate MobileNet's effectiveness in classifying various tomato illnesses. Comparing MobileNet to other deeper topologies, experimental results show how successful it is at achieving high accuracy with reduced computational complexity. We obtained 97% accuracy, classifying nine disease categories plus one healthy category using the leaves of the tomato plant as a feature.