Preliminary Face Recognition Grand Challenge Results
P. Jonathon Phillips
, Patrick J. Flynn
, Todd Scruggs
Kevin W. Bowyer
, William Worek
National Institute of Standards and Technology, 100 Bureau Dr., Gaithersburg, MD 20899
Computer Science & Engineering Depart., U. of Notre Dame, Notre Dame, IN 46556
SAIC, 4001 N. Fairfax Dr., Arlington, VA 22203
The goal of the Face Recognition Grand Challenge
(FRGC) is to improve the performance of face recogni-
tion algorithms by an order of magnitude over the best
results in Face Recognition Vendor Test (FRVT) 2002.
The FRGC is designed to achieve this performance goal
by presenting to researchers a six-experiment challenge
problem along with a data corpus of 50,000 images.
The data consists of 3D scans and high resolution still
imagery taken under controlled and uncontrolled con-
This paper presents preliminary results of the
FRGC for all six experiments.
The preliminary results
wards achieving the stated goals.
In the past few years, a number of new face recog-
nition techniques have been proposed.
The new tech-
niques include recognition from three-dimensional (3D)
recognition from high resolution still images,
face recognition, multi-algorithm, and preprocessing al-
gorithms to correct for illumination and pose variations.
These techniques hold the potential to improve perfor-
mance of automatic face recognition signiﬁcantly over
the results in the Face Recognition Vendor Test (FRVT)
The Face Recognition Grand Challenge (FRGC)
is designed to achieve an order of magnitude increase
in performance over the best results in FRVT 2002 by
encouraging the development of algorithms for all of
the above proposed methods.
To facilitate the devel-
opment of new algorithms, a data corpus consisting of
50,000 recordings divided into training and validation
partitions was provided to researchers.
The starting point for measuring the increase in
(HCInt) of the FRVT 2002.
The images in the HCInt
corpus were taken indoors under controlled lighting.
The performance point selected as the reference is a
veriﬁcation rate of 80% (error rate of 20%) at a false ac-
cept rate (FAR) of 0.1%. This is the performance level
of the top three FRVT 2002 participants.
An order of
magnitude increase in performance is therefore deﬁned
as a veriﬁcation rate of 98% (2% error rate) at the same
ﬁxed FAR of 0.1%.
Participants in FRGC submitted a set of raw sim-
ilarity scores to the FRGC organizers on 14 January
This paper provides a summary of performance
from these submitted scores. A more detailed descrip-
tion of the FRGC challenge problem, data, and experi-
ments is given in Phillips et al .