An Intelligent System for Automated Identification and Categorization of Plant Diseases

Abstract

Early crop disease detection is an important area of research towards Precision Agriculture (PA). Innovations like Artificial Intelligence (AI) and Internet of Things (IoT) have paved the way for technology-driven agriculture. Towards this end, many researchers contributed in crop disease detection using learning-based methods. However, decease detection accuracy can be further enhanced by improving such models. In the current paper, we propose a deep learning-based crop disease detection framework. We enhanced one of the deep learning models, the Convolutional Neural Network (CNN) to improve its accuracy. We also defined an algorithm named Learning based Crop Disease Detection (LbCDD) which exploits our enhanced CNN for efficient disease detection and classification. It is a multi-class classification model designed to classify all possible diseases. We used PlantVillege dataset for our empirical study. Experimental results showed that LbCDD outperforms many existingmethods. Our framework can be used to be part of a Decision Support System (DSS) in PA applications.