Face Recoginition Demonstration - FaceL
We have developed a simple to use and fun application that lets anyone play with face recognition in realtime over a live video stream.

Anyone running Mac OS X may download a ready-to-run application. For Windows users there is a beta release of a ready-to-run application. Finally, the source has been successfully compiled and tested using Kubuntu linux.

There are two websites with additional information:

• CSU Local FaceL Website
• Main SourceForge Website

Face Recognition Covariate Analysis
A key question in the study of face recognition algorithms is what factors influence performance, including such things as the age of the people and the relative focus of the images. Our group has pioneered the use of statistical techniques such as Generalized Linear Models to measure the influence of multiple factors on human face recognition algorithm performance. Much of this work has been supported by the Technical Support Working Group.

Our most recent major study has looked at performance of three face recognition algorithms that were part of the 2006 Face Recognition Vendor Test (FRVT). Prior to studying performance in the vendor test, we carried out a study of performance for three algorithms in the Face Recognition Grand Challenge (FRGC). Here are links to additional information on each set of studies:

• FRVT 2006 related papers and supplementary material.
• FRGC related papers and supplementary material.

A recent paper presented at the The IEEE Third International Conference on Biometrics: Theory, Applications and System summarizes findings of how age, gender, race, expression, time between images and image resolution have been shown to influence face recognition performance in past studies. More about this analysis may be found at:

• A Meta-Analysis of Face Recognition Covariates

Earlier papers on covariate analysis are in the general list of publications on the Computer Vision Group Homepage.

Evaluation of Face Recognition Algorithms
The Evaluation of Face Recognition Algorithms project began as a DARPA sponsored effort in 2000. This effort has its own website summarizing the effort and its accomplishments, particularlythe CSU Face Identification Evaluation System, which may be downloaded through the page under the algorithms link. This project is largely complete, although many researchers continue to download our benchmark software.