WiDG: An Air Hand Gesture Recognition System Based on CSI and Deep Learning

Hand gesture recognition has become a hot research topic because it plays a crucial role in human-computer interaction applications. Channel State Information (CSI) is attracting more attention since it depicts more accurate communication links and can be leveraged to recognize target action in its coverage area. In this paper, we propose a device-free hand gesture recognition system based on CSI and deep learning models, called WiDG. This system can recognize handwritten digits from 0 to 9 in the air according to CSI changes caused by different hand movements. We build deep learning models to identify hand gestures. We conduct experiments in both non-through-the-wall and through-the-wall scenarios to evaluate system performance. The experimental results show that Convolutional Neural Networks (CNN) achieves 97.2% and 95.7% recognition accuracy in the non-through-the-wall scene and through-the-wall scene, respectively. In addition, we discuss the system parameters affecting recognition accuracy and compare system performance with WiNum. The results show that deep learning models can realize hand gesture recognition with a satisfactory performance using CSI.