Rider Water Wave-enabled deep learning for disease detection in rice plant

Abstract

Rice diseases have degraded the production of the rice plant, which produces economic loss. To control and minimize the effects of attacks, the diseases are required to be recognized at a premature stage. Premature detection of infections can improve the yield from quantitative as well as qualitative losses, diminish the usage of pesticides, and improve the economic growth of the country. Hence, this paper devises a new method, namely Rider Water Wave-based neural network (RWW-NN) for finding the disease in the rice plant, where the training of NN is completed using the RWW, which is formed by assimilating Rider Optimization algorithm (ROA) and Water wave optimization (WWO). Initially, the pre-processing is done by using histogram equalization from the input image. Then, the segmentation is completed using Segmentation Network (SegNet), and then the CNN features are employed for feature mining in order to acquire the optimal features for disease recognition. These features are fed to NN for disease detection wherein the RWW is introduced for training the optimal weights. The RWW-based NN acquires greatest accuracy of 0.908, F-measure of 0.907, sensitivity of 0.862, and specificity of 0.947 based on K-value using Rice disease dataset.