Penerapan Deep Learning Pada Jenis Penyakit Tanaman Kelapa Sawit Menggunakan Algoritma Convolutional Neural Network

Abstract

The problem of Plant Destruction Organisms (OPT), especially related to disease, has always been an issue in the management of oil palm plantations. Oil palm has diseases caused by pests and others that can affect the growth and fruiting process. For this reasearch, the aims to identify whether or not oil palm plants are healthy through the color of their leaves, so that it will facilitate the performance of farmers. Deep Learning (DL) is a field of science from machine learning by doing deeper learning for many layers. Convolutional Neural Network (CNN) is one of the DL algorithms designed to process data in two-dimensional form such as images. Therefore, in this study, the CNN method will be applied to classify the health of oil palm plants based on the color of the leaves. The data used are 3000 data with test scenarios for training data and testing data are 90%:10%, 80%:20%, 70%:30% and 65%:35%. Based on the 4 test scenarios, the best accuracy obtained is 99.90% for the scenario of 65% of training data and 35% of testing data. While the lowest level of accuracy is 99.50% for the scenario of 90% training data and 10% testing data.