GAN-based Synthetic Radar Micro-Doppler Augmentations for Improved Human Activity Recognition

Deep neural networks (DNNs), and, in particular, convolutional neural networks (CNNs), have recently received much attention in a wide range of research areas, including radar-based human activity recognition. However, obtaining a large training dataset still remains a challenging task due to the costs and resources required for data collections. In this paper, we present a method for extending adversarial learning to the generation of synthetic radar time-frequency (TF) domain signatures which provides the ability to adapt to different operational environments. The classification results achieved with a deep CNN trained on generated images prove the efficiency of proposed algorithm over the state of the art methods.