<![CDATA[Accurate and real-time visitor data are needed to support smart tourism management. However, conventional counting methods still have limitations in dynamic outdoor tourism environments. This study develops and evaluates a real-time video-based visitor counting system by integrating YOLOv11 for person detection and ByteTrack for multi-object tracking. This approach extends visitor counting evaluation to uncontrolled open-air tourist destinations, where lighting variation, background complexity, visitor movement, and crowd density may affect detection and tracking performance. The system was evaluated using nine Full HD videos from five tourist destinations in West Sumatra, recorded under daylight and afternoon conditions with low to medium visitor densities. The YOLOv11–ByteTrack system achieved an average counting accuracy of 84.02%, MAE of 7.22 visitors per video, MAPE of 15.98%, and an average processing speed of 36.23 FPS. The average accuracy exceeded those of YOLOv3 and YOLOv8, which achieved 75.71% and 77.15%, respectively. These findings suggest that YOLOv11–ByteTrack has practical potential as a real-time visitor counting approach in smart tourism management, particularly for monitoring visitor flows, assessing site capacity, controlling visitor density, and supporting data-driven infrastructure planning.]]>
Copyrights © 2026
