The goal of our experiements is to determine the performance characteristics of gait recognition systems across varying speed of walking. For all experiments described below the training and gallery sets are identical. The list of gallery and probe image sets for each experiment can be accessed by clicking on the respective links.

A note on the results..... The results are presented in three formats i.e. the similarity files, the similarity matrix (that can be loaded into MATLAB) and the image representation of the similarity matrix.

EXPERIMENT 1:

How well does the gait recognition algorithm perform within each gait? 

Experiment 1.1:
 

	slowWalk		fastWalk	ball
vr03_7	Gallery	Probe

Results:

HMM Approach

Similarity matrix (image)

Similarity matrix (text)

Similarity files

CMS Curve

Experiment 1.2:
 

	slowWalk	fastWalk		ball
vr03_7		Gallery	Probe

Results:

HMM Approach

Similarity matrix (image)

Similarity matrix (text)

Similarity files

CMS curve

Experiment 1.3:
 

	slowWalk	fastWalk	ball
vr03_7			Gallery	Probe

Results:

HMM Approach

Similarity matrix (image)

Similarity matrix (text)

Similarity files

EXPERIMENT 2:

How well does the gait recognition algorithm perform within each view? 

Experiment 2.1:
 

	slowWalk		fastWalk
vr03_7	Gallery	Probe

Results:

HMM Approach

Similarity matrix (image)

Similarity matrix (matrix)

Similarity files

Experiment 2.2:
 

	slowWalk		fastWalk
vr07_7	Gallery	Probe

Results:

HMM Approach

Similarity matrix (image)

Similarity matrix (text)

Similarity files

EXPERIMENT 3:

How well does the gait recognition algorithm generalize across different types of gaits for the same view?

Experiment 3.1:
 

	slowWalk	fastWalk
vr03_7	Gallery	Probe

Results:

HMM Approach

Similarity matrix (image)

Similarity matrix (matrix)

Similarity files

CMS curve

Experiment 3.2:
 

	slowWalk	fastWalk
vr07_7	Gallery	Probe

Results:

HMM Approach

Similarity matrix (image)

Similarity matrix (text)

Similarity files