Gait Recognition

Shuai Zheng¹ Kaiqi Huang¹ Junge Zhang¹, Dacheng Tao²,Bo Xie^1,2,3, Ran He¹, Tieniu Tan¹

¹NLPR, Institute of Automation, Chinese Academy of Sciences ²University of Technology, Sydney ³Georgia Tech

Introduction

Various popular feature learning methods such as sparse coding, Robust PCA, CCA, and Group sparse coding have been explored in multi-view multi-modality gait recognition and intelligent visual surveillance system applications.

Gait recognition is a popular pattern recognition problem for which attracts a lot of researchers from different communities such as computer vision, machine learning, biomedical, forensic studying and robotics.This problem also has great potential in industries such as visual surveillance. Hence, The Intelligent Recognition & Digital Security Group, which was formed in 1998 by Prof. Tieniu Tan at NLPR (National Laboratory of Pattern Recognition), developed this CASIA dataset since Dec. 10, 2001. This CASIA Gait Recognition Dataset current contains 4 subset: Dataset A(standard dataset), Dataset B(multiview gait dataset) and Dataset C (infrared gait dataset), and Dataset D(gait and its corresponding footprint dataset). The contributions of the project are two folds. 1) We create a new database contains gait and foot pressure image(footprint), which would help further studies on gait. It becomes Gait CASIA Part D sub dataset. 2) We demonstrate how to address most important problem in gait recognition by employing some popular machine learning techniques such RobustPCA, Group Sparse Coding and etc.

More information can be found at this link.

CASIA Gait Part D Dataset

Multi-view Gait Recognition with RobustPCA and Group Sparse Coding

Duration

Jan, 2008-Jan, 2010

Publications

S. Zheng, K. Huang, T. Tan and D.Tao, ” A Cascade Fusion Scheme for Gait and Cumulative Foot Pressure Image Recognition“, Pattern Recognition, 2012.[bib]

S. Zheng, J. Zhang, K. Huang, R. He, and T. Tan, “Robust View Transformation Model for Gait Recognition,” in International Conference on Image Processing(ICIP), Brussels, Belgium, 2011.[bib] – How to employ Robust PCA with Partial Least Square to address multi-view gait recognition problem?

S. Zheng, K. Huang, and T. Tan., “Evaluation framework on translation-invariant representation for cumulative foot pressure image,” in International Conference on Image Processing(ICIP), Brussels, Belgium, 2011. [bib] – A new type of gait data called cumulative foot pressure image!

S. Zheng, B. Xie,D. Tao, and K. Huang, “Multi-View Pedestrian Recognition using Shared Dictionary Learning with Group Sparsity,” in International Conference on Neural Information Processing (ICONIP), Shanghai, China, 2011.(oral)[slides][dataset][code][bib] – How to use L-1_inf norm to address the multi-camera tracking problem?
Patrick Connor. Combining underfoot pressure features for shod and unshod gait biometrics. (Submitted)

Datasets

CASIA Gait Part B (GEI in Matlab Mat format):

CASIA Gait Part D-part1

CASIA Gait Part_D-part2

CASIA Gait Part D Static Cumulative Foot Pressure Image and its walking side sequence: [Gait&Footprint]

CASIA Gait Part D: Foot Pressure Dynamic Data with in mat format, Thanks to Patrick Connor.

Code

Warning: THESE RESOURCES COME WITH NO WARRANTIES, USE AT YOUR OWN RISK.

Multiview Gait Recognition Baseline for CASIA Part B (Matlab/Octave code, it requires to download GEI features above. Please cite our paper if you use it.) [link]

RobustPCA code[link] (How to use: Replace SVD-based View transformation model (VTM) in the baseline code. This is a version of robust pca code where it has been modified a few lines. The original code is from Prof. Yi Ma’s website.)

Multiview Gait Recognition with RobustPCA[link]

share dictionary learning with group sparsity.[link]

Toy example about L1-inf sparse coding [link].

Patents

Tieniu Tan, Kaiqi Huang, Shuai Zheng, “An Robust foot impression based human analysis”, Applied on 2009.12.23, FIle NO. 200910243728.6.[link]

Links

[1] What’s the latest research by many others on gait recognition beyond our work? [link]

[2] Gait Recognition project page in CBSR.[link]

Extension Applications for foot pressure recognition:

(NB: Ping me if you think your papers should or should not be presented here.)

Patrick Charles Connor. Comparing and Combining Underfoot Pressure Features for Shod and Unshod Gait Biometrics. IEEE International Conference on Technologies for Homeland Security (HST), 2015.

7 thoughts on “Gait Recognition”

iftikhar ansari
at

Sir can please send me the matlab code to extract gait energy from gait cycles (From silhouettes).
Thank You

Iftikhar Ansari
Research Scholar at Indian Institute Of Technology, Delhi
adminPost author
at

Hi, Iftikhar, You can get Gait energy by doing these. 1) Compute the cumulative of the silhouttes in one gait cycle, 2) Divide this by the number of frames in one gait cycle. The results would be similar to what GEI we provided in this website. Let me know if you need more help in this. I will try to find that bit code for you and release it here. Cheers. Kyle
Lillie
at

I read a lot of interesting posts here. Probably you spend a lot of time writing, i know how to save you a lot of work, there is an online tool that creates high quality, google friendly posts in seconds, just search in google – laranitas free content source
nimisha
at

Hello Sir,
Can you mail me a code.?
Because I want to take trials on live video..
This is just for learning about different pattern recognition methods..
Researcher amr
at

Please Admin,

I urgently need the matlab code for generating the GEI from the silhouette.

Thanks in advance
Om Prakash
at

Dear sir,
I gone through the the gait recognition system you proposed. I found it very interesting and useful for my work. I want to begin in computer vision area by gait analysis. Could you please provide me matlab code which you have written?
Thanks and regards,
Om Prakash
LEI
at

Sir,
In your paper “A Cascade Fusion Scheme for Gait and Cumulative Foot Pressure Image Recognition”,how to determine the dimensions of pbHOGs? Do pbHOGs extracted from every frame have the same dimensions?Is the dimensions of concatenated pbHOG representation same with that of pbHOGs or (N*dimensions of pbHOGs) ?(N is the number of frames in one gait cycle)
THX ~
LEI

Comments are closed.