Summary

Smart Info-Media Systems in Asia

2019

Session Number:SS2

Session:

Number:SS2-5

Personal Identification by CNN Using Footstep Waveforms

Yoshiki Goto,  Akitoshi Itai,  

pp.44-48

Publication Date:2019/9/4

Online ISSN:2188-5079

DOI:10.34385/proc.57.SS2-5

PDF download (1.3MB)

Summary:
It is known that the footstep includes personal characteristics. We often recognize a person from walking footsteps in limited situation. If the high accuracy personal identification using footstep is possible, a novel surveillance systems like a crime prevention system, or a biometric system are expected. However, in conventional method, individual characteristics of footsteps have not been mentioned. The classification accuracy in conventional research has not been reached to actual identification. The convolution neural network(CNN) is often used to various recognition systems. In this paper, we apply the CNN to footstep identification, whose training dataset is the clipped footstep waveform. We discuss the optimal number of steps when constructing a system from several dataset patterns.