Development of Artificial Intelligence System to Support Stadium Access Innovation for Exercise

Abstract

          In an era characterized by rapid urbanization and increasingly busy lifestyles, access to sports facilities for physical activity continues to face numerous challenges, particularly due to time constraints and overcrowding. The development of an AI-powered system capable of providing real-time data on facility usage presents a promising solution to alleviate these barriers and facilitate safer, more informed decisions about where and when to exercise especially during periods of disease outbreaks. This study investigated the application of object detection technology using the YOLO (You Only Look Once) model for identifying, capturing, and counting individuals engaged in physical activity within sports facilities. The primary objective is to develop and evaluate the performance of an artificial intelligence system designed to enhance access to public exercise spaces.

To achieve this, motion video capture devices were installed to collect data from 30 voluntary participants exercising in various zones of three distinct sports facilities: a football field, a fitness center, and a swimming pool. Digital image data were then analyzed using the YOLOv8 model to assess the system’s detection and counting capabilities. System accuracy was measured as a percentage to compare performance across facility types.

          The findings demonstrate that the AI system developed through the application of the YOLOv8 object detection model is highly effective in identifying, locating, and counting individuals in sports venues. The system generates output in the form of annotated images showing detected individuals and their respective locations. The overall accuracy rate across all tested facilities reached 94.67%, highlighting the model’s strong potential for real-time monitoring and intelligent support in the management and accessibility of exercise facilities.