Design of optimised model for nitrogen deficiency in rice leaves

Abstract

A major crop for agricultural productivity is rice. This study aims to create a convolutional neural network model that is precisely predicting nitrogen deficiency in rice plants. Convolutional neural networks (CNNs) must be tested with a variety of configurations for various numbers of convolutional layers, filter size in each layer, number of convolution filters in each layer, and pool size sampling the images in order to get optimal performance. In this paper, rice leaf dataset was used to predict nitrogen deficiency in rice crop. Secondary data is used to perform convolutional neural network. From which 30% of the total data were used for testing and 70% of the images were used for training the model. After comparing the Adam optimiser accuracy and RMSprop optimiser accuracy, it is clearly seen that Adam optimiser gives higher accuracy. The model achieved 99% of classification accuracy using genetic algorithm (GA).