B.S. (Bachelor of Science)
Department of Computer Science
The ongoing global pandemic caused by novel Covid-19 virus has infected 248 million and killed 5 million people so far worldwide. The scarcity of vaccines in different parts of the world has forced people to follow safety and precautionary measures. Even where the vaccines are widely available, a significant number of breakthrough cases have been reported recently. Touching the face with an infected hand is one of the leading causes of spreading the virus. Since face-touching is a compulsive behavior, it is hard to prevent them without a constant reminder. In this work, we present FaceShield application deployed on a commercial off-the-shelf smartwatch to alert the user to prevent them from touching their face. We discuss the challenges and their solutions related to limited resources, high variance in face-touching activities, battery longevity and accuracy while deploying a Random Forest model on a commercial smartwatch to detect the activity in real-time. We have shown that FaceShield is a practical application that can be installed and used on a resource-limited device like smartwatch without significantly draining the battery.
Lake, Rami I., "FaceShield: Using Machine Learning to Prevent Face Touching on a Commercially Available Smartwatch" (2022). Honors Capstones. 1428.