Open AccessDetection of Rice Plant Diseases using Convolutional Neural Network
Author(s) -
Intan Yuniar Purbasari,
Basuki Rahmat,
C S Putra Pn
Publication year - 2021
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/1125/1/012021
Subject(s) - convolutional neural network , staple food , computer science , pooling , deep learning , population , rice plant , artificial intelligence , agriculture , layer (electronics) , agricultural engineering , pattern recognition (psychology) , agronomy , geography , engineering , medicine , biology , chemistry , environmental health , archaeology , organic chemistry
Agricultural sector is the main sector that plays an important role in the national economy, from absorbing labours to playing as a contributor for foreign exchange. Indonesia is an agrarian country whose livelihood of the majority population is farming. Rice is one of the cultivation plants that becomes the staple food of most population. Thus, rice availability and its quality are factors that must be put in high consideration, either for national consumption or for export quality. This study aims to detect diseases in rice plants (by observing the leaves) that may cause a decrease in rice production or may result in bad quality rice using an artificial intelligent approach. The method used in this research is Convolutional Neural Network (CNN). This CNN is the result of the development of multilayer perception (MLP) which is used to manage two-dimensional data. The input from CNN is in the form of 2-dimensional data which is then propagated on a network with parameters at different weights and linear operations. CNN is one method of deep learning. The CNN method has many types of layers, namely the convolution layer, the subsampling / pooling layer, and the fully connected layer. This study uses different CNN architectures to find the best accuracy value. This study used four types of leaf diseases in rice plants with each type of disease consisting of 2,239 training image data and 168 image data. This research has succeeded in detecting diseases in leaf images automatically with the best training accuracy obtained at 91%.