ENHANCED SELF-CHECKOUT SYSTEM USING YOLOV10
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Malaysia Sarawak
Abstract
Description
In today’s digital era, the demand for efficient, accurate, and contactless retail experiences is rapidly increasing. This project presents the design and development of an enhanced self-checkout system that incorporates YOLOv10 for real-time object detection and a hand-tracking virtual keyboard for touch-free payment interaction. The system addresses common issues found in traditional self-checkout machines, such as slow processing speed and limited adaptability to dynamic retail environments. A structured survey involving students from the Faculty of Computer Science and Information Technology at UNIMAS was conducted to identify key user requirements, which guided the system design. The prototype demonstrates improved detection accuracy and system responsiveness, with testing confirming its ability to function effectively under various conditions. Specifically, YOLOv10 achieved 100.0% accuracy in both single-object and stress test scenarios, significantly outperforming the CNN baseline, which scored only 33.3% and 0.0% respectively. User feedback also indicated high satisfaction, with 93.3% confirming receipt accuracy, 86.7% rating the system's response speed between 4 and 5, and 100% finding the layout easy to navigate. These results suggest that the integration of advanced computer vision techniques and intuitive user interaction can significantly enhance the performance, usability, and reliability of self-checkout systems in modern retail settings.