The first challenge is training data. 10-100 videos of people beginning to drown would be a good start. It also might be highly pool dependent. The different sh

The first challenge is training data. 10-100 videos of people beginning to drown would be a good start. It also might be highly pool dependent. The different shapes, styles and attractions at the pool will all make a difference. Wave pool vs slope vs dropoff etc