Modified ride-NN optimizer for the IoT based plant disease detection

Abstract

Internet of Things (IoT) has emerged prolifically in the recent years as they aid in lot of applications. In reference to the agriculture sector, automated technologies for the plant disease recognition have varying benefits, and at the same time has potential challenges. In this work, an automated plant disease detection model has been developed for the IoT environment. The proposed scheme places the nodes over the simulation environment for capturing the plant leaf images. The system maintains a sink node, which collects the information from the automated plant disease detection module and helps in IoT based monitoring. The images from the nodes are pre-processed through the median filter for making it suitable for the plant disease detection. Then, the segmentation is done over the image, and from the image, segment level and the pixel level features are extracted. This work develops a novel classifier, named sine cosine algorithm based rider neural network (SCA based RideNN) for the disease detection such that the weights in the neural network are chosen optimally. The entire detection performance is validated using the metrics, like accuracy, sensitivity, specificity, and energy of nodes on different IoT environments. The simulation results reveal that the proposed approach has improvement detection performance with the accuracy of 0.9156.