Human identification with FMCW radar

What is it about?

Gait-based human identification aims to identify individuals by their walking style. In this paper, we investigate the use of micro-Doppler (m-D) signatures retrieved from a FMCW radar sensor to identify individuals based on their natural gait characteristics.

Featured Image

Why is it important?

We have established a currently largest uncontrolled gait dataset using a frequency-modulated continuouswave (FMCW) radar, including the gait data of 20 subjects who can walk around naturally and freely indoors. Then, an effective radar data processing method was proposed to eliminate noise and artifacts interference caused by multipath, thus high-quality gait m-D signatures can be extracted from raw radar records. Finally, we fine-tuned a pre-trained ResNet-50 with transfer learning for human identification, which outperforms the state-of-the-art methods.