Identification of Paddy Leaves Disease using Convolutional Neural Networks

Abstract

The economic development of any country mainly depends on agriculture. Rice is a staple crop and a significant food source for a large portion of the world's population and rice usage is higher in Asia. But, the rice crop suffers from various diseases that greatly affect yield capacity and have an impact on food security. For the effective disease management and crop detection, the disease need to be detected early and diagnosis to be done at the earliest. By automating the disease detection process, the system can enable early intervention and timely management practices, leading to improved crop yield and reduced economic losses. The proposed system effectively identifies paddy leaf disease, in which a median filter is used to reduce the noise, a Fuzzy C-Means (FCM) algorithm were used for segmentation, and a Convolutional Neural Network (CNN) were uses for classification. Contrast, homogeneity, energy, entropy, and correlation features are extracted from the segmented image for training and testing the CNN classifier. The proposed system effectively detects Brown Spots (BS), Leaf Blasts (LB), and Bacterial Blight (BLB) disease. The proposed system was implemented by using MATLAB (R2021a) software and tested with the help of three data sets with respect to performance parameters: sensitivity, specificity, accuracy, and precision. The maximum achieved accuracy of the proposed system is 98.68%.