VASC Seminar Announcement
       =========================

Date: Monday, 12/03/01
Time: 3:30-4:40
Place: NSH 1507

Title: Quo Vadis Face Recognition?

Speaker: Ralph Gross

Abstract:

After more than 30 years of research in face recognition a number of people
have claimed results which suggest that the basic problem is solved.
Today, there are a number of companies in the field producing face recognition
systems for large-scale real-world deployment. However, there are persistent
discrepancies between these claims and the user experience of the systems
in question.
Little is known about the extent to which factors such as face pose, illumination,
expression, and occlusion affect the face recognition accuracy. Since the events
of September 11, and the emergence of commercial face recognition systems in
common places like airports and beaches, the general public has taken great
interest in this question.

In the first part of my talk I will present the findings of an extensive empirical
evaluation of a number of leading algorithms, including Visionics' FaceIt.
The image data we used consists of over 21000 images from 3 recently
constructed face databases: CMU PIE, Cohn-Kanade, and AR database.
We found that most of the algorithms were able to handle variations in
illumination and expression. However, recognition accuracy was highly sensitive
to variation in pose and to occlusion, especially around the eye region.

In the second part of my talk I will propose a new face recognition algorithm which
addresses the problem of face recognition across pose. Our algorithm can use any
number of gallery images per subject captured at arbitrary poses, and any number of
probe images, again captured at arbitrary poses. The algorithm operates by estimating
the eigen light-field of the subject's head from the input gallery or probe images.
Matching between the probe and gallery is again performed using the eigen light-fields.

The first part is based on our workshop paper which is available on the HID webpage
http://www.hid.ri.cmu.edu/Hid/publications.html

Quo Vadis Face Recognition?
Ralph Gross, Jianbo Shi and Jeff Cohn
Third Workshop on Empirical Evaluation Methods in Computer Vision, 2001

The second part is based on our F&G'02 submission
Eigen Light-Fields and Face Recognition Across Pose
Ralph Gross, Iain Matthews and Simon Baker

which for now is only available upon request.