Fine Grained Group Gesture Detection Using Wearable Devices

1 MOTIVATION People tend to form groups in real world activities. One way to detect groups is to first recognize each user’s activity and then analyze their cooperative or collaborative relationship [1]. However, in some cases wemay just want to find out whether people belong to the same group rather than identifying the specific activity they are performing, so groups may be detected based on either proximity[2] or gesture similarities[3]. Nowadays wearable devices are extremely popular as personal health and fitness tracking devices. These wearables can be effectively used for the identification of user status in groups that perform the same activity at the same location, which can be helpful in many applications such as emergency response, disaster recovery, and sport activities. For instance, we can help guide people towards emergency exits during a fire evacuation or identifying the group of supporters of a particular team in a sports game. In this work, we utilize the sensor data collected from smartwatches and apply signal processing algorithms to accurately identify the group synchronization status.