Development of Visitor Interest Detection and Tracking System in The Museums

Understanding what interests’ visitors in museum collections is crucial for improving exhibitions and enhancing the visitor experience. Traditional methods such as surveys and direct observation have limitations, including bias and difficulties in collecting real-time data. This paper presents a web-based system that utilizes advanced computer vision techniques to track and measure visitor interest in real time. The system, developed using the Django framework with JavaScript and HTML for the front end, and SQLite3 for the database, was tested in a live museum setting. It employed a YOLO-based object detection model and the BotSort algorithm, achieving 88.5% accuracy and a 93.9% F1 score, even in low-light conditions. During a 1-week trial at the National Awakening Museum in Jakarta, the system detected 1,221 interested visitors and 6,666 passing visitors. Usability testing showed that 90% of the 10 respondents were very pleased with the system. The system is scalable, cost-effective, and well-suited for museums of all sizes.