Air Writing Gesture Recognition Using FMCW Radar and Deep Learning

초록

Hand gesture recognition is getting huge attention amongst the research community in the domain of human-computer interaction (HCI). Amongst different HCI methods, air-writing recognition is gaining significant importance. In this paper, we present a deep learning framework to recognize alphabets written in the air using a Frequency Modulated Continuous Wave (FMCW) radar. Human subjects were asked to perform gestures in front of radar and the signal captured through radar is localized in two dimensions. Afterwards, a deep learning model is trained on the patterns generated by localizing hand movements. Data from eight human volunteers is captured and promising accuracy of above 90% is achieved.

키워드