Recognizing social gestures with a wrist-worn smartband

The ability to recognize social gestures opens the door for the development of enhanced pervasive computing applications that are responsive to users´ social interactions. In this paper, we explore the feasibility of using a smartband for social gesture recognition. We apply logistic regression, a supervised machine learning technique, to accelerometer data collected in a study of 32 users performing 12 social gestures. Our experimental results show promise for recognizing social gestures with a smartband; our simple approach achieves an average accuracy of 86% for classification of social gestures.