Gait Recognition for Co-Existing Multiple People Using Millimeter Wave Sensing

Zhen Meng,Song Fu,Jie Yan,Hongyuan Liang,Anfu Zhou,Shilin Zhu,Huadong Ma,Jianhua Liu,Ning Yang

DOI: https://doi.org/10.1609/aaai.v34i01.5430

2020-01-01

Proceedings of the AAAI Conference on Artificial Intelligence

Abstract:Gait recognition, i.e., recognizing persons from their walking postures, has found versatile applications in security check, health monitoring, and novel human-computer interaction. The millimeter-wave (mmWave) based gait recognition represents the most recent advance. Compared with traditional camera-based solutions, mmWave based gait recognition bears unique advantages of being still effective under non-line-of-sight scenarios, such as in black, weak light, or blockage conditions. Moreover, they are able to accomplish person identification while preserving privacy. Currently, there are only few works in mmWave gait recognition, since no public data set is available. In this paper, we build a first-of-its-kind mmWave gait data set, in which we collect gait of 95 volunteers 'seen' from two mmWave radars in two different scenarios, which together lasts about 30 hours. Using the data set, we propose a novel deep-learning driven mmWave gait recognition method called mmGaitNet, and compare it with five state-of-the-art algorithms. We find that mmGaitNet is able to achieve 90% accuracy for single-person scenarios, 88% accuracy for five co-existing persons, while the existing methods achieve less than 66% accuracy for both scenarios.