ITL Kicks-Off the Multiple Biometric Grand Challenge
The Information Access Division (IAD) in ITL recently kicked off a new project called the Multiple Biometric Grand Challenge (MBGC). Over 100 people from seven different countries attended the first MBGC Workshop held on April 18, 2008 in Arlington, Virginia. The MBGC Program Manager is Jonathon Phillips (IAD).
The MBGC plans to investigate, test and improve the performance of face and iris recognition technologies on both still imagery and video sequences through a series of challenge problems and evaluations. The evaluation methodology in this program has been used successfully on several previous IAD programs, which laid the foundation for rapidly improving a technology’s state-of-the-art performance.
The Face Recognition Grand Challenge (FRGC) (2004-2006) and the Face Recognition Vendor Test (FRVT) 2006 were good examples of this methodology’s success. The FRVT 2006 report can be downloaded from http://face.nist.gov/frvt/frvt2006/FRVT2006andICE2006LargeScaleReport.pdf,
Continuing this successful methodology, the MBGC will provide the test data, a baseline algorithm, and software infrastructure to participants on which to test and improve their face and iris recognition algorithms. MBGC participants will report their results back to NIST and to the MBGC community at periodic workshops.
The MBGC Evaluation Team has designed three challenge problems: 1) the Portal Video Challenge, 2) the Still Face Challenge, and 3) the Video Face Challenge. The Portal Video Challenge will encourage face and iris recognition algorithm development on Near Infrared and High Definition video sequences taken as people walk through a portal. This application will enable rapid acquisition of biometrics at ports of entry while, at the same time, reducing invasiveness during acquisition. The Portal Video Challenge will also provide an opportunity for fusion of face and iris recognition technologies. The Still Face Challenge will focus on recognition from unconstrained indoor and outdoor imagery. The purpose of this challenge is to improve recognition performance on lower resolution still imagery and on imagery similar to mug shots, passports and credential cards. The Video Face Challenge seeks to improve face recognition capability in outdoor scenarios from video sequences taken at a distance. This challenge problem includes matching both still images against video and video against video.
The MBGC, after it’s completed, will be followed by an independent multiple biometric evaluation to document the new state-of-the-art and verify participants’ self-reported results. So far, over 100 organizations from 21 different countries have expressed an interest in the MBGC program.
The MBGC is sponsored by multiple U.S. Government agencies including the Department of Homeland Security, Science & Technology directorate; Director of National Intelligence's Intelligence Advanced Research Projects Activity (IARPA); the Federal Bureau of Investigation's Criminal Justice Information Services (CJIS) and Operational Technology Division (OTD); the multi-agency Technical Support Working Group (TSWG); and the Department of Defense's Biometric Task Force (BTF). More information is available on the MBGC website at http://face.nist.gov/mbgc.
Contact: Jonathon Phillips, ext. 5348
Cathy Schott, ext. 5594