2025 Volume 81 Issue 17 Article ID: 25-17055
Approximately half of all drowning incidents are caused by rip currents. To prevent such accidents, AI-based rip current detection systems have been in operation on five beaches, providing real-time hazard notifications to lifeguards and beachgoers. This systems have significantly reduced rip current incidents at the monitored beaches. However, the detected information is limited to the fixed camera view, making it difficult for users to spatially recognize the location of rip currents. Therefore, we developed an AR ap-plication that visualizes rip currents areas linked to the existing detection system. In this study, a suitability of the AI model created on the adjacent beach for detecting rip currents on the study coast was confirmed. Then, the accuracy of converting image data to GPS coordinates, and the accuracy of AR-based rip current visualization were evaluated. The results showed a GPS conversion error of 4.1 meters, and an AR visualization error of 9.9 meters. The maxi-mum error relative to the center of the detected rip current area was approximately 14 meters, which is considered acceptable for rip currents awareness.
