Face Recognition Workshop
Face recognition workshop, Tuesday 31 October, 1.00pm - 5.00pm
PRESENTERS: Patrick Grother and Mei Ngan NIST
Patrick Grother is a staff scientist at the National Institute of Standards in Technology responsible for biometric standards development, algorithm testing and analysis. He leads the ongoing MINEX, IREX and MBE-STILL evaluations of fingerprint, iris and face recognition technologies. These support biometrics in national scale identity management. He currently serves as editor of four ISO standards in the areas of interchange of fingerprint minutiae and facial data, performance testing and modular fusion processes. He was the recipient, in 2003 and 2007, of Department of Commerce Gold Medals for, respectively, tests of border management technologies and for the specification of biometric data elements and procedures needed for the US Government's Personal Identity Verification (PIV) program. He advises US Government agencies on a number of biometric projects.
Mei Ngan graduated from the University of Maryland, College Park with a B.S. in Computer Engineering and received her M.S. in Computer Science from the Johns Hopkins University. She now works for the National Institute of Standards and Technology (NIST) with focus on research and evaluation of face recognition and tattoo recognition technologies.
This half day workshop will cover face recognition including the new generation of algorithms, demographics, NIST results, items from NIST test roadmap.
Target Audience: End-users, integrators, application developers and others interested in the technical performance aspects of face detection and recognition in photographs collected with variable image quality constraints and subject non-cooperation.
Face recognition performance and measurement
1. Across algorithms, datasets, demographics, levels of subject cooperation
2. With motion video vs. still.
3. Online vs. offline.
Hunting down face recognition errors
1. Sources: Capture, detection, enrollment, feature extraction, matching 1:1 and 1:N
2. The anatomy of the impostor distribution. Are distributions stable? Do fat tails exist?
3. The anatomy of the genuine distribution. High scores. Quality. Standards and conformance.
4. Nature-vs-capture: Subjects-specific vs. image-specific effects.
5. Demographics effects in biometrics: ISO/IEC TR 22116
6. Image quality: Standards. Measurement and standards conformance.
Face recognition under attack
1. Passive vs. active. Zero effort and levels beyond.
2. Tests for morphing, manipulation, presentation attack detection.
3. Role of ISO/IEC 30107.
State of the industry
1. Retrospective evolution of the FR performance 2000 – 2017.
2. CNNs in an open universe: A breakthrough?
3. Test programs and benchmarks worldwide: Megaface, IJB-x, LFW, FRVT
4. Comparing tests in academia, corporations, labs, pilots, and operations.
Member: £100 + VAT = £120
Non-member: £200 + VAT = £240