A Water Behavior Dataset for an Image-Based Drowning Solution

Drowning is responsible for an estimated of 320,000 deaths annually worldwide, roughly 25% of those deaths are in swimming pools. This is probably due to the fact that a drowning person, to the untrained eye, will appear to be normally playing or floating in the water. While drowning, a person is unable to call for help, as the nervous system focuses on gathering oxygen for the lungs. To assist the lifeguards with their rescue mission, we propose a water behavior dataset curated to support the design of image-based methods for drowning detection. The dataset includes three major water activity behaviors (swim, drown, idle) that have been captured by overhead and underwater cameras. Moreover, we develop and test two methods to detect and recognize the drowning behavior using the proposed dataset. Both methods use deep learning and aim to support a fast and smart pool rescue system by watching for the early signs of drowning rather than looking for a drowned person. The results show a high performance of the presented methods validating our dataset, which is the first public water behavior dataset and the main contribution of the work.