Micro hand gesture recognition system using hybrid dilated convolution
Yaoyao Dong, Wei Qu, Tianhao Gao, Haohao Jiang, Pengda Wang · 2022
Gesture recognition is the latest human-computer interaction (HCI) technology, which allows users to naturally control electronic devices through the movement of fingers and palms without operating redundant devices. Radar gesture recognition technology offers significant advantages in terms of privacy and security, device reliability and design flexibility. In this paper, a model GestureNet suitable for radar gesture recognition is designed by using the smooth pseudo Wigner Ville processing of millimeter wave radar gesture echo and the knowledge of hybrid zero convolution neural network in deep learning. The results show that the recognition accuracy of the validation set of GestureNet reached 97.35% and the recognition accuracy of the test set reached 91.75%, indicating that the model has good generalisation ability, thus providing a strong guarantee for radar gesture recognition.