Automatic Learning of Gait Signatures for People Identification

This work targets people identification in video based on the way they walk (i.e. gait). While classical methods typically derive gait signatures from sequences of binary silhouettes, in this work we explore the use of convolutional neural networks (CNN) for learning high-level descriptors from low-...

Full description

Saved in:
Bibliographic Details
Published inAdvances in Computational Intelligence Vol. 10306; pp. 257 - 270
Main Authors Castro, Francisco Manuel, Marín-Jiménez, Manuel J., Guil, Nicolás, Pérez de la Blanca, Nicolás
Format Book Chapter
LanguageEnglish
Published Switzerland Springer International Publishing AG 2017
Springer International Publishing
SeriesLecture Notes in Computer Science
Subjects
Convolutional Neural Network
Human Action Recognition
Near Neighbour
Optical Flow
Support Vector Machine
Online AccessGet full text
ISBN9783319591469
3319591460
ISSN0302-9743
1611-3349
DOI10.1007/978-3-319-59147-6_23

Cover

More Information
Summary:This work targets people identification in video based on the way they walk (i.e. gait). While classical methods typically derive gait signatures from sequences of binary silhouettes, in this work we explore the use of convolutional neural networks (CNN) for learning high-level descriptors from low-level motion features (i.e. optical flow components). We carry out a thorough experimental evaluation of the proposed CNN architecture on the challenging TUM-GAID dataset. The experimental results indicate that using spatio-temporal cuboids of optical flow as input data for CNN allows to obtain state-of-the-art results on the gait task, both for identification and gender recognition, with an image resolution eight times lower than the previously reported results (i.e. $$80\times 60$$  pixels).
Bibliography:Original Abstract: This work targets people identification in video based on the way they walk (i.e. gait). While classical methods typically derive gait signatures from sequences of binary silhouettes, in this work we explore the use of convolutional neural networks (CNN) for learning high-level descriptors from low-level motion features (i.e. optical flow components). We carry out a thorough experimental evaluation of the proposed CNN architecture on the challenging TUM-GAID dataset. The experimental results indicate that using spatio-temporal cuboids of optical flow as input data for CNN allows to obtain state-of-the-art results on the gait task, both for identification and gender recognition, with an image resolution eight times lower than the previously reported results (i.e. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$80\times 60$$\end{document} pixels).
ISBN:9783319591469
3319591460
ISSN:0302-9743
1611-3349
DOI:10.1007/978-3-319-59147-6_23