Detecting Human–Human Interactions Using Smartphones

People use online communication tools such as social networking sites on a daily basis, but they are often disconnected from face-to-face interactions. Therefore, we consider the use of sensor networks to recognize human activities and reflect them in the online world. In this study, we developed a prototype system that detects handshakes from the acceleration and angle of smartphones and exchanges contact information between them in real time. It uses 5G local area networks to collect sensor data and multivariate time-series classification models to recognize human activities. Experimental results show that human activities can be roughly recognized using only smartphone sensors, and the position of the sensor affects the accuracy. This system will serve as a basis for more sophisticated systems in future.