BTAS 2013 will feature four tutorials as listed below. Registration details for tutorials can be found here.

Quick links:
TABULA RASA: September 29, 9am – 12pm (Francis Scott Key Salon A)
Face Recognition: September 29, 9am – 12pm (Francis Scott Key Salon B)
Iris Recognition: September 29, 2pm – 5pm  (Francis Scott Key Salon A)
Forensic Recognition: September 29, 2pm – 5pm  (Francis Scott Key Salon B)


 

Spoofing and Anti-Spoofing in Biometrics: lessons learned from the TABULA RASA project
Sébastien Marcel
Idiap Research Institute

Abstract
This BTAS tutorial will present the main research outcome of the TABULA RASA project. TABULA RASA (Trusted Biometrics under Spoofing Attacks) is a European funded project (7th Framework Program) that is addressing some of the issues of direct (spoofing) attacks to trusted biometric systems. This is an issue that needs to be addressed urgently because it has recently been shown that conventional biometric techniques, such as fingerprints and face, are vulnerable to direct (spoofing) attacks.

Direct attacks are performed by falsifying the biometric trait and then presenting this falsified information to the biometric system, one such example is to fool a fingerprint system by copying the fingerprint of another person and creating an artificial or gummy finger which can then be presented to the biometric system to falsely gain access. This issue effects not only companies in the high security field but also emerging small and medium sized enterprises (SMEs) that wish to sell biometric technologies in emerging fields.

Biography
Sébastien Marcel received the Ph.D. degree in signal processing from Université de Rennes I in France (2000) at CNET, the research center of France Telecom (now Orange Labs). He is currently interested in pattern recognition and machine learning with a focus on biometrics. He is a senior researcher at the Idiap Research Institute (CH), where he heads a research team and conducts research on face recognition, speaker recognition and spoofing attacks detection. In 2010, he was appointed Visiting Associate Professor at the University of Cagliari (IT) where he taught a series of lectures in face recognition. He is also lecturer at the Ecole Polytechnique Fédérale de Lausanne (EPFL) where he is teaching on “Fundamentals in Statistical Pattern Recognition”. He serves on the Program Committee of several scientific journals and international conferences in pattern recognition and computer vision. Finally he is the principal investigator of international research projects including MOBIO (EU FP7 Mobile Biometry), TABULA RASA (EU FP7 Trusted Biometrics under Spoofing Attacks) and BEAT (EU FP7 Biometrics Evaluation and Testing).

URL
http://www.idiap.ch/~marcel/professional/BTAS_2013.html

return to top


 

Face Recognition by Humans and Machines
Alice J. O’Toole
The University of Texas at Dallas

Abstract
Human recognition skills are considered to be the gold standard for computer-based face recognition systems. At our best, we are remarkably robust at face recognition in suboptimal viewing conditions and across changes in viewpoint, illumination, and appearance (e.g., due to aging). Humans are also able to exploit the identity information carried in facial motions. Machine-based face recognition has improved steadily over the last two decades, but is still largely limited to recognition with frontal face images. Characteristics of the human system can provide useful hints for improving the capabilities of face recognition algorithms. Moreover, direct comparisons between the performance of humans and machines can provide an important benchmark for evaluating the state of the art in automatic face recognition. In this tutorial, I will cover the three topics:

1.) Predicting human face recognition performance
Human face perception and recognition has been studied by psychologists for decades. This work converges on an understanding of the factors that affect human accuracy under a variety of conditions. These factors can be categorized into photometric variables (e.g., illumination, pose, resolution), inherent characteristics of faces (e.g., demographic categories), and subject factors (e.g., age). Interactions among these variables, including the interaction that produces the well-known other-race effect, are surprisingly relevant for predicting the performance of algorithms in real world applications. I will review the characteristics of human face processing, with an emphasis on the relevance of these characteristics for understanding machine performance.

2.) Neural representations of faces and people in motion
The human visual system carries out preprocessing operations that transform image-based face representations into representations of the identity, intent, and emotional state of a person. Particular areas in the brain are specialized for processing faces or bodies, and are tuned to biological motions, including facial expression, gesture, and gait. I will provide a brief overview of the neural architecture of the brain’s person-processing network.

3.) Comparisons between human and machine-based face recognition
Over the last two decades, the U.S. Government has sponsored international competitions for face recognition algorithms. Beginning in 2005, direct comparisons between the performance of humans and machines have shown that the best algorithms surpass humans on identity matching of frontal images with variable illumination. I will present a timeline of these comparisons, beginning in 2005 and continuing to the present. These comparisons illustrate the relative strengths of humans and machines. I will also present a fusion strategy for combining human and machine judgments to achieve greater accuracy than can be achieved with either system operating alone. The most recent comparisons with highly challenging images show that humans perform more accurately than machines on difficult face recognition tasks. We will show that the human advantage in these highly challenging cases comes from the ability to exploit identity specific information in the body.

Biography
Prof. Alice O’Toole received the B.A.in psychology from The Catholic University of America, Washington DC in 1983. She received the M.Sc. (1985) and Ph.D. (1988) in Experimental Psychology from Brown University, Providence, RI. She was a postdoctoral fellow at the University of Dijon, in Dijon, France, and at the Ecole Nationale de Telecommunications in Paris, France. She has been on the faculty of The University of Texas at Dallas since 1989, and has been a Professor in Cognition and Neuroscience since 1999. Over the last 25 years, she has published numerous articles on human perception and memory for faces from the behavioral, computational, and neural perspectives. Her current work includes functional magnetic resonance imaging studies of human face processing, performance comparisons between humans and machines on face processing tasks, and eye-movement studies of person perception.

Prof. O’Toole currently holds the Aage and Margareta Moller Endowed Chair in the School of Behavioral and Brain Sciences. She has been awarded research fellowships from the French Embassy to the United States and from the Alexander von Humboldt Foundation in Germany. Professor O’Toole serves as an Associate Editor of Psychological Science and The British Journal of Psychology. She is on the editorial board of the Vision Science Society and has served as an Area Chair for the IEEE International Workshop on Automatic Face and Gesture Recognition (2013). Her research over the years has been supported by funding from the National Institute of Mental Health, DARPA, the Technical Support Working Group of U.S. Dept. of Defense, the Office of Naval Research, and Texas Instruments.

URL
http://www.utdallas.edu/~otoole/BTAS_2013.html

return to top


Iris Recognition: From Origin to Current Research Frontiers
Kevin Bowyer
University of Notre Dame

Abstract
This tutorial is intended for attendees who may be familiar with some other area of biometrics, and who want to understand the basic principles of iris recognition and to be aware of current major research directions. Earlier versions of this tutorial have been given at the IEEE Automated Face and Gesture Recognition conference, the IEEE Workshop on Applications of Computer Vision, and at the IEEE Homeland Securities Technology conference. The tutorial includes a historical perspective on the origin and development of iris recognition, an explanation of the principles of the Daugman approach to iris recognition, and an introduction to major current research directions.

Biography
Kevin W. Bowyer is the Schubmehl-Prein Professor and the Chair of the Department of Computer Science and Engineering at the University of Notre Dame. He has made major contributions in multiple areas of biometrics research, including iris recognition, face recognition, multi-biometrics and other areas. His research group has been involved in numerous government-sponsored biometrics programs, including the Human ID Gait Challenge, the Face Recognition Grand challenge, the Iris Challenge Evaluation, the Face Recognition Vendor Test 2006, and the Multiple Biometric Grand Challenge. He has served as Editor-in-Chief of the IEEE Biometrics Compendium and the IEEE Transactions on Pattern Analysis and Machine Intelligence. He served as the General Chair of the 2007, 2008 and 2009 IEEE International Conference on Biometrics Theory Applications and Systems, and the 2011 International Joint Conference on Biometrics. He is a Fellow of the IAPR, a Fellow of the IEEE and a Golden Core Member of the IEEE Computer Society. His most recent book is the Handbook of Iris Recognition.

URL
http://www.nd.edu/~kwb/iris_recognition.htm

return to top


 

1-to-1 Forensic Facial Comparison
Richard W. Vorder Bruegge
Federal Bureau of Investigation

Abstract
Academic and industrial scientists engaged in biometric research would benefit from insight into the processes through which human analysts perform biometrics-related tasks in operational scenarios. For example, a better understanding of the processes currently used by examiners in forensic facial comparison would allow such researchers to identify avenues of research that would provide additional scientific validation and general improvements for those processes. The purpose of this tutorial is to provide the student with a better understanding of the current approach used by human analysts and identify some of the technical advances that have shown promise as an aid to the analyst.

Biography
Richard W. Vorder Bruegge is a Senior Level Photographic Technologist for the Federal Bureau of Investigation, which he joined in 1995. Since that time his work has involved analysis of film, video, and digital images that relate to crime and intelligence matters, as well as testifying in court. He has been Chair of the Facial Identification Scientific Working Group since 2009. The FBI has designated him as the Bureau’s subject matter expert for face and iris recognition. He is a fellow of the American Academy of Forensic Sciences and was named a Director of National Intelligence Science and Technology Fellow in January 2010. He is the lead instructor for the FBI in facial comparison analysis.

URL
https://www.fiswg.org/doc/pdf/FISWG_GuidelinesforFacialComparisonMethods_v1.0_2012_02_02.pdf

https://www.fiswg.org/document/viewDocuments;jsessionid=2303B0B4708EB00FF4F47E8380FC9AF5

 

return to top