We have developed the AI Intelligent Pool Automated Drowning Prevention and Alert System, which offers several key benefits, including adaptability to different venue environments, assistance in maintaining lifeguards＇ attention, enhanced judgment efficiency, and prediction of swimming patterns. By identifying pre-drowning, mid-drowning, and post-drowning stages with a recognition accuracy of over 90%, lifeguards and safety personnel can proactively respond to potentially risky situations.

Our AI-based system expedites the assessment of swimming patterns by considering multiple parameters. The system＇s adaptability to various venues is achieved through multi-point ROI and perspective transformation techniques. Additionally, low-latency motion patterns, zone-specific audio-visual alarms, and a dual drowning prevention safety mechanism are employed to assist lifeguards in promptly identifying anomalies.

Our system has proven to be effective in large-scale venues. This can be adapted for medium- and small-scale venues by changing the number of cameras and computing platform capabilities as necessary. In future, our goal is to collaborate with AI camera developers and embed the system with edge computing-enabled AI cameras to enable data processing and analysis to be performed locally on the camera itself, reducing latency and enhancing real-time monitoring capabilities.