A NOVEL APPROACH TO COFFEE BEAN SORTING: YOLO ALGORITHM IMPLEMENTATIO

Abstract

The coffee industry's growing demand for high-quality coffee beans has spurred the need for efficient and accurate coffee bean sorting methods. Traditional manual selection by humans lacks consistency due to factors like human error, lack of training, and long working hours. On the other hand, existing sorting machines can only select beans based on size, neglecting essential characteristics like shape that contribute to coffee flavor. To address these limitations, we propose an automatic coffee bean selector using the YOLO (You Only Look Once) algorithm integrated with a Raspberry Pi microcontroller.

The YOLO algorithm, known for its real-time object detection capabilities at 45 FPS, has demonstrated promising results in video analysis and image understanding tasks. We adapt this algorithm to detect coffee beans based on their shape and size, optimizing the coffee bean selection process. The Raspberry Pi, a versatile microcomputer, provides a suitable platform for running the YOLO algorithm, utilizing Python as the programming language, which aligns with the algorithm's requirements.

In this study, we develop a coffee bean selection system, named "Green Beans," to automate the process and reduce the dependency on manual labor. By integrating the YOLO algorithm with the Raspberry Pi, we aim to achieve high accuracy and efficiency in coffee bean sorting. This automation becomes crucial as the industry witnesses increasing demands, putting a strain on human resources using the manual method.

We conduct comprehensive testing of the YOLO-based coffee bean selector to evaluate its performance in detecting coffee beans. The results of the testing reveal the algorithm's effectiveness in identifying coffee beans accurately and efficiently. By leveraging the power of computer learning and modern technology, we bridge the gap between manual and machine-based coffee bean selection, paving the way for improved coffee production quality.