Recognition of Hand-Towards-Face Movements Prototype to Fight COVID-19 Using an IMU Sensor and Deep Learning Model Embedded in a Single-Board Computer

Narvaez A.M.A. · Smart Innovation, Systems and Technologies (2021) · DOI: 10.1007/978-3-030-75680-2_97

Resumen

People touch their faces an average of 16 to 23 times per h; this unconscious habit is very difficult to eradicate and allows the SARS-CoV-2 virus to enter the body after a contaminated surface was touched with hands, causing COVID-19 disease that could become a serious illness for older people with underlying medical problems. To help combat this behavior, a Recognition of hand-towards-face movements prototype has been proposed by using an IMU Sensor and Deep Learning Model embedded in Single-Board Computer. It works using linear and angular accelerations from an MPU 6050 IMU sensor and a Raspberry Pi Model 3B+ that collects data and executes a TensorFlow Lite model that determines whether a person is going to touch their face or not to avoid unintentional infections.

Instituciones: Universidad Nacional de Ingeniería