PENGEMBANGAN APLIKASI MOBILE UNTUK DETEKSI CACAT BIJI KOPI ROBUSTA BERDASARKAN STANDAR NASIONAL INDONESIA

Abstract

Coffee is one of the most important raw materials for beverages worldwide. Indonesia is one of the world's leading coffee producers and exporters, and coffee consumption is currently increasing. To maintain the taste of coffee, it is necessary to preserve the quality of coffee beans. The government has established the Indonesian National Standard (SNI) 01-2907-2008 regarding coffee beans. The process of determining the quality of beans is carried out by processors who manually follow the steps outlined in the SNI. Therefore, to accelerate the quality determination process and provide knowledge to the general public, the author has developed a mobile-based coffee bean quality detection application to determine the grade of coffee beans using an Android platform. This application incorporates a machine learning model based on object detection. For the machine learning implementation, TensorFlow and Convolutional Neural Network (CNN) architecture are used. From the training data results, an average accuracy of 84% was achieved, indicating that the application generally detects correctly. Additionally, testing was conducted on several smartphones and at varying detection distances. The testing results showed that the optimal distance for detection is 15 cm to obtain accurate and detectable data. Furthermore, the smartphone with the best performance was the Smartphone A30, featuring a 16 MP resolution camera.