Near drowning pattern detection using neural network and pressure information measured at swimmer's head level
Mohamed Kharrat, Yuki Wakuda, Noboru Koshizuka, Ken Sakamura · 2012
It is difficult for a person who cannot swim to call for help while he face a drowning incident. This make from drowning incidents very dangerous as it can occur silently. In this research we consider the use of wearable sensors to identify victims at early drowning stage. For this we attached a pressure sensor logging unit at the head level of a professional lifeguard, then we asked him to imitate near drowning pattern. We process the obtained dataset with neural networks at 20 second time window. The trained neural network succeed to classify the near drowning and normal swimming pattern.