Deep Learning based Leaf Disease Detection using Convolutional Neural Network

Rice serves as a staple food for numerous tropical and subtropical nations worldwide. However, the maintenance and care of extensive paddy fields spanning hectares can be arduous for farmers. Identifying specific diseases becomes a challenging task, often resulting in farmers resorting to excessive chemical usage to safeguard their crops. Unfortunately, these chemicals can be detrimental to consumers and have negative long-term effects. Hence, there is a pressing need for a rapid classification method that distinguishes between healthy and diseased paddy plants. This research proposes a solution by swiftly categorizing paddy plants as diseased or healthy and identifying the affected areas in case of disease. The study employs an intelligent edge detector to determine the extent of the affected region. The image dataset for this module consists of 1766 pictures sourced from open repositories, including both healthy and diseased paddy leaves. The proposed Deep Learning model in this study utilizes a Convolutional Neural Network (CNN) for disease detection, achieving an impressive accuracy of nearly 96.7%• Compared to prior works, this classification model exhibits improved accuracy. By combining CNN with the Canny edge detector, the method harnesses the strengths of both approaches to effectively detect rice leaf diseases. This information on the percentage of diseased crops can aid in disease management and reducing the excessive use of pesticides, thus benefiting society at large.