Bean vision: a microcontroller-based intelligent hulling and sorting system for robusta coffee bean using yolo v8/
Laika Joy S. San Diego, Janna Victoria M. Cerera, Danilyn B. Magluyoan, Gio Sebastian O. Pacia, Kathleen Nicole S. Pascual, and John Kenneth F. Villaflores.--
- Manila: Technological University of the Philippines, 2025.
- xiv, 232pages: 29cm.
Bachelor's thesis
College Of Engineering.--
Includes bibliographic references and index.
The Philippine coffee industry faces significant challenges in processing Robusta coffee beans due to inefficiencies and quality inconsistencies caused by traditional manual methods. This study proposes a hybrid system that integrates mechanical and electronic technologies to automate the processing of Robusta coffee. The system includes a vibratory mesh sorting mechanism that separates the beans into medium and large sizes, improving consistency and uniformity. The mechanical hulling machine efficiently removes husks, while YOLO V8 computer vision technology is incorporated for quality sorting. The trained YOLO V8 model, based on a dataset of 11,400 samples, achieves an impressive 97% accuracy in detecting the correct bean quality classification. Although the speed of the YOLO V8 model may not yet surpass manual sorting, its high accuracy offers significant potential for enhancing quality control and reducing labor dependence in the future. The system was evaluated using the ASEAN Standard for Coffee Beans (ASEAN Stan 31: 2013) to ensure compliance with industry quality standards through the verification of the Philippine Professional Coffee Cuppers to the beans used for data collection and trained in the computer vision YOLO v8 algorithm model. The system increased the sustainability and global competitiveness of the Philippine coffee sector by modernizing traditional post-harvest processing methods.
Hybrid automation system Mechanical hulling machine YOLOv8 computer vision