FRVT 2006: Quo Vadis face quality

J. Ross Beveridge; Geof H. Givens; P. Jonathon Phillips; Bruce A. Draper; David S. Bolme; Yui Man Lui

doi:10.1016/j.imavis.2009.09.005

FRVT 2006: Quo Vadis face quality

J. Ross Beveridge, Geof H. Givens, P. Jonathon Phillips, Bruce A. Draper, David S. Bolme, Yui Man Lui

Research output: Contribution to journal › Article › peer-review

24 Scopus citations

Abstract

A study is presented showing how three state-of-the-art algorithms from the Face Recognition Vendor Test 2006 (FRVT 2006) are effected by factors related to face images and people. The recognition scenario compares highly controlled images to images taken of people as they stand before a camera in settings such as hallways and outdoors in front of buildings. A Generalized Linear Mixed Model (GLMM) is used to estimate the probability an algorithm successfully verifies a person conditioned upon the factors included in the study. The factors associated with people are: Gender, Race, Age and whether they wear Glasses. The factors associated with images are: the size of the face, edge density and region density. The setting, indoors versus outdoors, is also a factor. Edge density can change the estimated probability of verification dramatically, for example from about 0.15 to 0.85. However, this effect is not consistent across algorithm or setting. This finding shows that simple measurable factors are capable of characterizing face quality; however, these factors typically interact with both algorithm and setting.

Original language	English
Pages (from-to)	732-743
Number of pages	12
Journal	Image and Vision Computing
Volume	28
Issue number	5
DOIs	https://doi.org/10.1016/j.imavis.2009.09.005
State	Published - May 2010
Externally published	Yes

Funding

The work was funded in part by the Technical Support Working Group (TSWG) under Task T-1840C. P.J.P. was supported by the Department of Homeland Security, Director of National Intelligence, Federal Bureau of Investigation and National Institute of Justice. The identification of any commercial product or trade name does not imply endorsement or recommendation by Colorado State University or the National Institute of Standards and Technology.

Keywords

Biometric quality
Face recognition
Generalized linear mixed models
Image covariates

Access to Document

10.1016/j.imavis.2009.09.005

Cite this

@article{1064308a97154ad381a171a3b3170149,

title = "FRVT 2006: Quo Vadis face quality",

abstract = "A study is presented showing how three state-of-the-art algorithms from the Face Recognition Vendor Test 2006 (FRVT 2006) are effected by factors related to face images and people. The recognition scenario compares highly controlled images to images taken of people as they stand before a camera in settings such as hallways and outdoors in front of buildings. A Generalized Linear Mixed Model (GLMM) is used to estimate the probability an algorithm successfully verifies a person conditioned upon the factors included in the study. The factors associated with people are: Gender, Race, Age and whether they wear Glasses. The factors associated with images are: the size of the face, edge density and region density. The setting, indoors versus outdoors, is also a factor. Edge density can change the estimated probability of verification dramatically, for example from about 0.15 to 0.85. However, this effect is not consistent across algorithm or setting. This finding shows that simple measurable factors are capable of characterizing face quality; however, these factors typically interact with both algorithm and setting.",

keywords = "Biometric quality, Face recognition, Generalized linear mixed models, Image covariates",

author = "Beveridge, {J. Ross} and Givens, {Geof H.} and Phillips, {P. Jonathon} and Draper, {Bruce A.} and Bolme, {David S.} and Lui, {Yui Man}",

year = "2010",

month = may,

doi = "10.1016/j.imavis.2009.09.005",

language = "English",

volume = "28",

pages = "732--743",

journal = "Image and Vision Computing",

issn = "0262-8856",

publisher = "Elsevier",

number = "5",

}

TY - JOUR

T1 - FRVT 2006

T2 - Quo Vadis face quality

AU - Beveridge, J. Ross

AU - Givens, Geof H.

AU - Phillips, P. Jonathon

AU - Draper, Bruce A.

AU - Bolme, David S.

AU - Lui, Yui Man

PY - 2010/5

Y1 - 2010/5

N2 - A study is presented showing how three state-of-the-art algorithms from the Face Recognition Vendor Test 2006 (FRVT 2006) are effected by factors related to face images and people. The recognition scenario compares highly controlled images to images taken of people as they stand before a camera in settings such as hallways and outdoors in front of buildings. A Generalized Linear Mixed Model (GLMM) is used to estimate the probability an algorithm successfully verifies a person conditioned upon the factors included in the study. The factors associated with people are: Gender, Race, Age and whether they wear Glasses. The factors associated with images are: the size of the face, edge density and region density. The setting, indoors versus outdoors, is also a factor. Edge density can change the estimated probability of verification dramatically, for example from about 0.15 to 0.85. However, this effect is not consistent across algorithm or setting. This finding shows that simple measurable factors are capable of characterizing face quality; however, these factors typically interact with both algorithm and setting.

AB - A study is presented showing how three state-of-the-art algorithms from the Face Recognition Vendor Test 2006 (FRVT 2006) are effected by factors related to face images and people. The recognition scenario compares highly controlled images to images taken of people as they stand before a camera in settings such as hallways and outdoors in front of buildings. A Generalized Linear Mixed Model (GLMM) is used to estimate the probability an algorithm successfully verifies a person conditioned upon the factors included in the study. The factors associated with people are: Gender, Race, Age and whether they wear Glasses. The factors associated with images are: the size of the face, edge density and region density. The setting, indoors versus outdoors, is also a factor. Edge density can change the estimated probability of verification dramatically, for example from about 0.15 to 0.85. However, this effect is not consistent across algorithm or setting. This finding shows that simple measurable factors are capable of characterizing face quality; however, these factors typically interact with both algorithm and setting.

KW - Biometric quality

KW - Face recognition

KW - Generalized linear mixed models

KW - Image covariates

UR - http://www.scopus.com/inward/record.url?scp=76449103193&partnerID=8YFLogxK

U2 - 10.1016/j.imavis.2009.09.005

DO - 10.1016/j.imavis.2009.09.005

M3 - Article

AN - SCOPUS:76449103193

SN - 0262-8856

VL - 28

SP - 732

EP - 743

JO - Image and Vision Computing

JF - Image and Vision Computing

IS - 5

ER -

FRVT 2006: Quo Vadis face quality

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this