Open AccessPerancangan sistem identifikasi penyakit pada daun kelapa sawit (Elaeis guineensis Jacq.) dengan algoritma deep learning convolutional neural networks
Author(s) -
Gusti Ashari Wira Satia,
Erick Firmansyah,
Arif Umami
Publication year - 2022
Publication title -
jurnal ilmiah pertanian/jurnal ilmiah pertanian
Language(s) - English
Resource type - Journals
eISSN - 2502-5988
pISSN - 1829-8346
DOI - 10.31849/jip.v19i1.9556
Subject(s) - convolutional neural network , palm oil , computer science , artificial intelligence , elaeis guineensis , deep learning , agricultural engineering , machine learning , machine vision , artificial neural network , plant disease , agriculture , classifier (uml) , engineering , agroforestry , microbiology and biotechnology , environmental science , ecology , biology
The effectiveness and efficiency of operations are essential in increasing the production and profitability of oil palm plantations. It can be performed through the application of precision farming principles. One of the main obstacles for oil palm to produce optimally according to their potential is disease attacks on leaves. However, the weakness of the manual observation method is the limited ability of the observer in assessing a disease that attacks leaves. Therefore, it is necessary to have a companion system for smallholders to detect and control diseases with minimal environmental impact properly. Most of the visual-based identification efforts in precision agriculture use the concepts of computer vision and machine learning. This study's problem was the need for machine learning and computer vision-based software to identify diseases to realize sustainable oil palm plantation practices. Disease detection includes a description of the name of the disease in oil palm plantations. In this study, designing a disease recognition based on computer vision and machine learning had used the convolutional neural network (CNN). The application used the Android operating system in real-time. The test results on the model showed that the model had been able to predict with an accuracy rate of 85.5%