The Implementation of CNN on Website-based Rice Plant Disease Detection

Abstract

Rice is the staple food of Indonesian society. As Indonesia's population continues to grow, this implies that the need for rice consumption will also increase in the future. Therefore, it is necessary to have a strategy to maintain and increase rice harvest production in Indonesia. Rice farmers need to get maximum support to maintain the quality of the yield and rice produced. Unfortunately, in order to harvest rice at the right time with good quality, farmers often face various obstacles that can cause crop failure. Harvest failure in rice can be caused by various factors e.g. the disease that infects rice plants. To reduce crop failure caused by rice plant diseases, this research proposes a website-based system with the aim of detecting rice plant diseases to optimize agricultural sector. This system was developed by applying the Deep Learning method. The method of image processing was implemented using a Convolutional Neural Network with the GoogLeNet architecture which is then integrated into a website-based application. The results showed an increase in accuracy in the increasing number of epochs for CNN training models. This application is expected to be able to assist rice farmers in analyzing diseases in rice plants that are planted, so that prevention and handling can be carried out in accordance with the aim of minimizing losses from crop failure.