Database: Faces & Sketchs 人脸识别数据集

Database: Faces & Sketchs [UMIST] [Yale] [Purdue] [XM2VTS] [UCD] [BioID] [MIT] [FERET] [AT&T] [PIE] [IMM] [CMU]  [Champion] [Face Expression] [Face Sketch]

UMIST Face Database: The UMIST Face Database consists of 564 images of 20 people. Each covers a range of poses from profile to frontal views. Subjects cover a range of race/sex/appearance. Each subject exists in their own directory labelled 1a, 1b, ... 1t and images are numbered sequentially as they were taken. The files are all in PGM format, approximately 220 x 220 pixels in 256 shades of grey. Pre-cropped versions of the images may be made available by contacting Daniel Graham. top Yale Face Database: The Yale Face Database (size 6.4MB) contains 165 grayscale images in GIF format of 15 individuals. There are 11 images per subject, one per different facial expression or configuration: center-light, w/glasses, happy, left-light, w/no glasses, normal, right-light, sad, sleepy, surprised, and wink. top PurdueU Database: Available at: http://rvl1.ecn.purdue.edu/~aleix/aleix_face The Purdue U database contains 3203 face images of 126 people, where 2 sessions per person (14 days away ):Each person, each session contains 13 pictures under 3 changes respectively on expression, illumination, glass occlusion and scarf occlusion. The images are at resolution 768x576 pixels in RAW data with 24-bits gray level. top The XM2VTS Database: Multi-model personal verification is one of the most promising approaches to user-friendly highly security personal verification systems. The XM2VTS Database contains 1,180 Color frontal images ( 720X576 PPM image file, one picture per person, 4 sessions ), 2,360 side-profile images ( right profile, left profile, per person, per session ), 32 KHz 16-bit sound, video sequences, 3D model. top The UCD colour face image database for face detection: The database has two parts. The first part contains a 100 colour images of faces with variations in background (indoors, outdoors, complex, simple) facial structural components (beards, moustaches, and glasses), poses (frontal, near-frontal, profile), orientation (upright and upside down), facial expressions, imaging conditions (poor quality, good quality) occluded areas, age, gender, race and size. The images have been captured from the following sources: Digital Cameras Pictures scanned in using a scanner Images from the World Wide Web Images from existing face recognition and detection databases. It should be noted that all the pictures in the database are original pictures from the source without performing any processing such as colour correction, sharpening, noise removal, etc. This ensures that the images in the database reflect actual images that might be encountered by face detection algorithms. No restrictions are imposed on the face size or image resolution. Details of the database are shown in Table 1 in terms of 3D pose, orientation, structural components and occlusion. While no standard terms exist to define face pose and orientation, researchers often use their own terminology such as frontal, near-frontal, half-profile, profile, semi-frontal, tilted etc. This paper defines the terminology used to describe faces in the UCD Colour Face Image Database as follows: 3D Pose of the Face: Frontal: A frontal view is a mugshot–type view where the sagittal plane of the body divides the face in half. Profile: A side-view of the face where the plane of the image divides the face in half. Intermediate: A face pose that is neither frontal nor profile. The definitions above are subjective as accurate pose estimation is an entirely different area of research. Our definitions only serve the purpose of comparing the performance of face detection algorithms for different poses. Orientation of the Face: Upright: A face is considered upright if the major axis of the best-fit ellipse makes an angle of less than ±150 with the vertical axis. Rotated: Faces outside the above range are defined as rotated. Occluded Face:  A face is occluded if the entire face region is not clearly visible. Other faces or objects in the image can occlude faces. Structural Components:  Structural components such as glasses, sunglasses, beards and moustaches present a real challenge in face detection and are also noted for each face in the image. Table 1 UCD Colour Face Image Database top BioID Face DB: Available at: http://www.humanscan.de/support/downloads/facedb.php The dataset consists of 1521 gray level images with a resolution of 384x286 pixels. Each one shows the frontal view of a face of one out of 23 different test persons. top MIT CBCL Face Data Set: Available at: http://www.ai.mit.edu/projects/cbcl/software-datasets/FaceData2.html target= A training set consists of 6,977 cropped images (2,429 faces and 4,548 nonfaces), and the test set consists of 24,045 images (472 faces and 23,573 nonfaces ). top FERET Database: Available at: http://www.nist.gov/srd/ This database is a large collection of male and female faces. Each image contains a single person with certain expression. top AT&T (Olivettti) Database (ORL): Available at: http://www.uk.research.att.com/ This database contains 40 subjects, 10 images per subject. (10 images per person) and 150 profile face images (5 images per person) top PIE Database: Available at: http://www.ri.cmu.edu/projects/project_418.html The database contains faces of 68 people, 13 different poses, 43 different illumination conditions, and with 4 different expressions. top IMM Face Database: Available at: http://www.imm.dtu.dk/~aam/aamexplorer/ The IMM Face Database comprises 240 still images of 40 different human faces, all without glasses. The gender distribution is 7 females and 33 males. Images were acquired in January 2001 using a 640x480 JPEG format with a Sony DV video camera, DCR-TRV900E PAL. top CMU Face Database: The image dataset is used by the CMU Face Detection Project and is provided for evaluating algorithms for detecting frontal and profile views of human faces. This test set was collected at CMU by Henry Schneiderman and Takeo Kanade. Those images are not restricted in terms of subject matter or background scenery. They were collected from various news Web sites. We provide ground truth for face location in the following two files. (Note: Some faces are listed in both files if they in between frontal view and profile view)      frontal view ground truth     http://vasc.ri.cmu.edu/idb/images/face/profile_images/testing_frontal_ground_truth.dat     profile view ground truth      http://vasc.ri.cmu.edu/idb/images/face/profile_images/testing_profile_ground_truth.dat The CMU profile face set consists of 208 images with 441 faces of which 347 are profile views. top The Champions database: This database contains 236 images of Champions database, and each image involves one face. top Japanese Female Facial Expression (JAFFE) Database: The JAFFE database consists of 213 images of 7 different emotional facial expressions. top Face Sketch Database: The face sketch Database is provided by the Chinese University of Hong Kong, which contains 21 face sketches and their corresponding photo images. Besides, there are another 15 photo images without corresponding sketches. All images in the Database are frontal and gray. top collected by F.R&A Group, July, 2006

[VIPSL] [Database & Software]

from: http://see.xidian.edu.cn/vipsl/database_Face.html