Identification of Diseases on Corn Leaves Using CNN Denoising (DeCNN)

Abstract

Indonesia is a country with large agricultural commodities, especially corn. The need for corn increases every year both for food and feed. In agricultural production, if it cannot be identified correctly and quickly it will have a negative impact on crop yields, especially on the amount of harvest and crop quality. Currently every country is focusing on agricultural automation to increase crop yields and crop quality and accuracy to meet market demand. To respond to this, this research was conducted to reduce failed yields caused by diseases that attack corn. The aim of this study is to improve the accuracy of the Convolutional Neural Network algorithm in identifying diseases on corn leaves by combining Gaussian filters. This research was conducted with 4 classes on 4000 images, this research will produce data based on the class of healthy leaves, gray diseased leaves, blight diseased leaves. and leaf rust disease. the accuracy results obtained by combining the gaussian filter in this study are 99%, and it is hoped that further development in this research will be in the form of a mobile application that can help farmers.