人脸数据库汇总—Part 3
原链接失效的已整理删除，可参考另外几部分的数据库：
人脸数据库汇总—Part 1 - holybin的专栏 - 博客频道 - CSDN.NET
人脸数据库汇总—Part 2 - holybin的专栏 - 博客频道 - CSDN.NET
人脸数据库汇总—Part 4 - holybin的专栏 - 博客频道 - CSDN.NET
（1）来自：http://pics.psych.stir.ac.uk/Other_face_databases.htm
Glasgow unfamiliar face database
Three imagesof each of 303 identities, plus20 video stills for training
Meissner Caucasian andAfrican American set
Around 100 ofeach race, each neutral andsmiling. Access by email permission.
Brazillian FEIdatabase
100 of eachsex, 14 images for each atvarious angles. 640x480 pixels. Alsoincludes cropped frontalimages, smile and neutral. WIth permission, I've mirrored the main imagecollections here, since downloadfrom Brazil can be quite slow: part1 part2 part3 part4 cropped1 cropped2
Radboudface database
67 identities, each with 8 expressions:Anger, disgust, fear, happiness, sadness,surprise, contempt, and neutral, 3gaze directions and five cameraangles. Access by request.
IndianFace database
11 images ofeach of 39 men, 22 womenfrom Indian institute of technology Kanpur,640x480 colour (not greyscale as itsays on their site).
SocialCognition and Social Neuroscience lab at Princeton
A variety ofcollections of faces, includingsome computer generated, rated on dimensionssuch as trustworthiness anddominance.
KarolinskaDirectedEmotional Faces - KDEF
70 individuals, each displaying 7different emotional expressions, each expressionbeing photographed (twice) from5 different angles.
（2）来自：http://www.facedetection.com/facedetection/datasets.htm
BioID Face Detection Database
1521 images with human faces, recordedundernatural conditions, i.e. varying illumination and complex background. Theeyepositions have been set manually (and are included in the set) forcalculatingthe accuracy of a face detector. A formula is presented tonormalize thedecision of a match or mismatch. This is, to my knowledge, thefirst attempt tofinally create a real test scenario with precise rules on howto calculate theaccuracy of a face detector - open for all to compare theirresults in ascientific way!
FGnet - FaceandGesture Recognition Working Group
European FGnet encourages developmentofface and gesture recognition techniques. Among other contributions worthhavinga look at, they provide resources especially useful forfacedetection/recognition. Have a look at "Benchmark Data" to accessthelist of useful datasets!
CarnegieMellon Test Images
（1）PIE Database The CMUPose, Illumination, and Expression (PIE) database, provided by SimonBaker
（2）frontal_images test imagesfor the facedetection task, provided by Henry A. Rowley
（3）profile_images testimages forthe face detection task, provided by Henry Schneiderman
（4）facial_expressions CMU-Pittfacial expression database, provided by Jeff Cohn.
BaoFaceDatabase
Lots of face images, mostly peoplefromAsia. Single face pictures are in the "one faces" subdirectory.
Beautycheck
Average female and male face (UniversityRegensburg,Germany) for Causes and Consequences of Human FacialAttractiveness.