Local Binary Pattern (LBP) Extraction for Coffee Classification Using Modular Neural Networks (MNN)

Abstract

This study aims to improve the accuracy of coffee bean classification by utilizing Local Binary Pattern (LBP) extraction with Modular Neural Network (MNN). Coffee, one of Indonesia's leading commodities, plays a vital role in the country's economy. Coffee production, especially arabica and robusta types, experienced fluctuations from 2018 to 2020, motivating this research to explore the latest technology for improving the accuracy of coffee type classification. This research focuses on the three main coffee types in Indonesia, namely arabica, robusta, and liberica. Although arabica and robusta are more common, liberica coffee is also gaining attention in the Indonesian coffee industry. This study illustrates the steps of LBP extraction, an important feature technique in image processing and pattern recognition, using a flowchart to facilitate the process. We tested the effectiveness of LBP extraction on MNN to enhance the quality of classification. It was able to correctly classify 2,000 images of coffee beans 87.25% of the time. Thus, this research makes an important contribution to the development of technology in the context of coffee plantations in Indonesia. LBP in MNN not only improves coffee bean classification accuracy but also opens up further development opportunities in the application of technology in plantations and agro-processing sectors.