Face Recognition Databases From Other Research Groups

We list some face databases widely used for face recognition related studies, and summarize the specifications of these databases as below.

1. LFW (Labeled Faces in the Wild) Database

o Source: The LFW is built by University of Massachusetts, Amherst,

o Purpose: LFW is a database of face photographs designed for studying the problem of unconstrained face recognition. Variation in clothing, pose, background, and other variables is large in LFW.

o Properties:

Properties	Descriptions
# of subjects	5749
# of images/videos	13,233
Static/Videos	Static
Single/Multiple faces	Single
Gray/Color	color
Resolution	250*250
Face pose	Various poses
Facial expression	Various expressions
Illumination	Various illuminations
3D data	N/A
Ground truth	Identifications of subjects

o Reference: refer to the paper: Gary B. Huang, Manu Ramesh, Tamara Berg, and Erik Learned-Miller, "Labeled Faces in the Wild: A Database for Studying Face Recognition in Unconstrained Environments",
University of Massachusetts, Amherst, Technical Report 07-49, October, 2007.

2. Face Recognition Grand Challenge (FRGC) Database

o Source: the FRGC database is jointly sponsored by several government agencies interested in improving the capabilities of face recognition technology.

o Purpose: the primary goal of the FRGC database is to evaluate face recognition technology. It may also be used for eye detection.

o Properties:

Properties	Descriptions
# of subjects	222 (large still training set) 466 (validation set)
# of images/videos	12,776 (large still training set) 943 8 (3D training set) 4007 8 (validation set)
Static/Videos	Static
Single/Multiple faces	Single
Gray/Color	Color
Resolution	17042272 or 12001600
Face pose	Frontal view
Facial expression	Neutral and smiling
Illumination	Controlled and uncontrolled illumination
3D data	Yes (range and texture)
Ground truth	Positions of eyes, nose, and mouth Identifications of subjects

o Reference: Please refer to the paper: P. J. Phillips, P. J. Flynn, T. Scruggs, K. W. Bowyer, J. Chang, K. Hoffman, J. Marques, J. Min, and W. Worek, Overview of the face recognition grand challenge, Proc. of CVPR05, no. 1, pp. 947:954, June 2005 and the original document under the directory BEE_DIST\doc.

Georgia Tech Face Database

o Source: this database is constructed by Georgia Institute of Technology.

o Purpose: this database is primarily used for face recognition. It may also be used for face detection.

o Properties:

Properties	Descriptions
# of subjects	50
# of images/videos	750
Static/Videos	Static
Single/Multiple faces	Single
Gray/Color	color
Resolution	640*480
Face pose	Nearly frontal-view or quarter-profile images
Facial expression	Various
Illumination	Various
Accessories	Glasses
3D data	N/A
Ground truth	Identifications of subjects Coordinates of left-upper corner and right-bottom corner of face rectangle

UMIST_Face_Database

o Source: this database was constructed by the University of Manchester Institute of Science and Technology that merged with the Victoria University of Manchester to form the University of Manchester.

o Purpose: this database is primarily used for face recognition.

o Properties:

Properties	Descriptions
# of subjects	20
# of images/videos	564
Static/Videos	Static
Single/Multiple faces	Single
Gray/Color	Eight-bit gray
Resolution	92*112
Face pose	From profile to frontal views
Facial expression	neutral
Illumination	N/A
Accessories	Glasses
3D data	N/A
Ground truth	Cropped face region Identifications of subjects

o Reference: Please refer to the paper:Daniel B Graham and Nigel M Allinson, Characterizing Virtual Eigensignatures for General Purpose Face Recognition, Face Recognition: From Theory to Applications, NATO ASI Series F, Computer and Systems Sciences, vol. 163,
H. Wechsler, P. J. Phillips, V. Bruce, F. Fogelman-Soulie and T. S. Huang (eds), pp. 446-456, 1998.

ORL Database of Faces

o Source: this database is constructed by AT&T Laboratories Cambridge.

o Purpose: this database is primarily used for face recognition.

o Properties:

Properties	Descriptions
# of subjects	40
# of images/videos	400
Static/Videos	Static
Single/Multiple faces	Single
Gray/Color	Eight-bit gray
Resolution	92*112
Face pose	Moderate pose variation (up and down, quarter-profile to frontal-view)
Facial expression	3 facial expressions: neutral, smiling, closed eye
Illumination	N/A
Accessories	Glasses
3D data	N/A
Ground truth	Cropped face region Identifications of subjects

o Reference: Please refer to the paper: F. S. Samaria and A. C. Harter ‘Parameterisation of a stochastic model for human face identification’, Proc. of 2^nd IEEE workshop on Applications of Computer Vision, pp. 138-142, 1994.