Deep Learning Techniques for Automation and Industrial Applications

3Fruit Leaf Classification Using Transfer Learning Techniques

Taha Siddiqui^*, Surbhit Chopra and Mamta Arora

Department of Computer Science and Technology (CST), Manav Rachna University, Faridabad, India

Abstract

Approximately 60% of India’s provincial families rely on horticulture for their living expenses, and India has a reputation of having the second-largest agricultural area in the world. Forty-two percent of the most important food crops are produced from crops that could be catastrophically or chronically damaged. Due to advancements in deep learning methods, we can minimize this rate to a great extent. In this research chapter, we have applied various deep learning and transfer learning methods to accurately predict the disease of damaged plants so that we can cure them at the initial stage. The models are trained on the image dataset, which contains various categories of plants like mango and pomegranate. The results show that ResNet outperformed Inception, VGG19, and CNN by giving an accuracy of 88% and 87.5% for pomegranate and mango, respectively.

Keywords: Deep learning, transfer learning, leaf disease detection, image preprocessing

3.1 Introduction

Plant infections harm agricultural production and if they are not identified on time, meal insecurity may rise, especially the primary crops, such as mango and pomegranate, which are crucial as agricultural output becomes part of people’s food supply. Effective plant disease prevention and management start with early ...