Drowning Detector

Abstract

The goal of this project is to design a device that can give a warning message to people when drowning instances in a pool might occur. The device consists of a main waterproof housing that holds key components including a raspberry pi, a camera, a battery, an Intel Neural Compute Stick 2 and potentially a solar panel. An additional camera will be mounted on the house in order to keep a log of who comes in the backyard and who is leaving. The raspberry pi connects directly to the camera attachment and the Intel Stick, the latter of which allows for the drowning detection AI to run properly. The AI was built using Tensorflow and other AI tools and was trained to detect if a person is drowning. Once the device detects that someone might be drowning, it sends a notification to an android app and triggers a loud alarm to get the attention of people to assist the person currently drowning. If the person has the app downloaded then the app will display a CPR video in case the person doesn’t know how to perform CPR. This will allow for a better chance of saving the person who is drowning.