Impact of EfficientNetB3 for stratification of Tomato Leaves Disease

Abstract

India's rapidly expanding agriculture sector relies on early leaf detection. For the agricultural sector to succeed in its mission of feeding a large population, early detection of leaf problems in plants and the development of predictive systems are essential. To identify diseases in tomato plants, image analysis and machine vision techniques have recently been applied to images of tomato leaves. Using custom-made techniques or deep learning architectures, these methods are utilized to retrieve leaf image features. In order to identify tomato plant diseases, this paper employs a transfer learning based EfficientNetB3 model. The dataset includes 11 different types of leaves and is collected from an internet database. The EfficientNetB3 model is trained over the course of 15 iterations with a batch size of 32 using two optimizers, Adamax and Adam. Analysis using the Adam optimizer reveals an accuracy value of 0.94, consistent with the classification report's findings.