Author:
Li Beibei,Zhu Jiansheng,Li Wen
Abstract
AbstractWith business process optimization, technological advancement, equipment capability enhancement, and other means, the Railway Passenger Service Department in China is consistently working to improve the efficiency and convenience of passenger entry and exit procedures at railway stations. Concerning passengers’ checkout, not only conventional identification approaches based on manual control, identification card, and magnetic thermal paper ticket are supported, but also a recent contactless identification process based on face recognition is applied in some stations. To further improve the contactless identification ability for checkout, an advanced contactless checkout process based on gait-augmented face recognition is innovatively put forward, in which a weakly-supervised body segmentation network named Dwsegnet and an improved GaitSet model are proposed. Through comparison with various models, the effectiveness of both Dwsegnet and the improved GaitSet is validated. Specifically, the contactless identification rate of gait-augmented face recognition is improved by 2.31% when compared to single-modal face recognition, which demonstrates the superiority of the proposed process.
Funder
China Academy of Railway Sciences
Publisher
Springer Science and Business Media LLC
Reference42 articles.
1. Qian, H., Yi, J. & Fu, Y. Review of human action recognition based on deep learning. J. Front. Comput. Sci. Technol. 15(3), 438–455 (2021).
2. Song, Q., Peng, L., Qin, Y.-K. & Pei, X.-J. Gait and surface electromyography characteristics in hurdle athletes in a total gait cycle. Chin. J. Tissue Eng. Res. 21(24), 3851–3857 (2017).
3. Cheng, Q. & Sheng, G. Research on human gait test and recognition algorithm in fare gate system. Comput. Meas. Control 22(5), 1487–14891522 (2014).
4. Li, S. Z. Encyclopedia of Biometrics: I–Z Vol. 2 (Springer, 2009).
5. Shirke, S., Pawar, S. & Shah, K. Literature review: Model free human gait recognition. In 2014 Fourth International Conference on Communication Systems and Network Technologies 891–895 (IEEE, 2014).