Identifikasi Citra Kualitas Minyak Kelapa Sawit Berbasis Android Menggunakan Algoritma Convolutional Neural Network

Abstract

The Central Statistics Agency reports that the average development of palm cooking oil consumption at the household level in Indonesia during the 2015-2020 period has increased by 2.32% per year. The use of cooking oil repeatedly is commonplace among the people of Indonesia and quite a lot. Even though the use of cooking oil can endanger health because the frying process at high temperatures can damage the chemical structure of the oil. Therefore, in this study, image processing was carried out to identify the quality of palm oil using the Convolutional Neural Network (CNN) algorithm. This research was conducted through several stages, namely dataset collection, dataset preprocessing, CNN algorithm implementation, testing, and development of information systems. The dataset consists of image data of palm cooking oil that has not been used, palm cooking oil used for frying twice, and palm cooking oil used for frying more than twice. The total amount of data is 3000 image data. Distribution of training data and test data using the Pareto division of 80:20. Based on the test, the best accuracy is 97.08%. This research produces an android-based information system that can identify the quality of cooking oil based on the classification.