Leveraging Deep Learning for Drowning and Swimming Prevention

Drowning is a severe public health problem that has claimed many lives in Turkey. Our study focuses on a novel strategy to address this problem. We developed a semi-automated technique that uses drone technology and machine learning to prevent drowning in real time. Advanced convolutional neural network (CNN) models, including Xception, ResNet-50, and YOLOv8, were used in our approach. These various models were trained using a special dataset that included simulated drowning situations in the Turkish Aegean Sea and a collection of online images. Though fully automated operation cannot be ensured, this method greatly improves water safety by guiding the drone to the drowning event and alerting the relevant staff. The models performed admirably with relative accuracy rates of 82.1%, 83.40%, and 85.8%. This cutting-edge approach demonstrates how machine learning could fundamentally alter the way major health concerns are handled. It also demonstrates how, when integrated with traditional safety procedures and human supervision, technology can improve and assist human efforts to safeguard the public's health.