各位学者，MSU-PRIP实验室共享了一系列数据库，包括人脸模拟画像，人脸活体检测，人对年龄的估计等；希望对相关领域的学者们有帮助:
University Distinguished Prof. Anil K. Jain
Pattern Recognition and Image Processing (PRIP) Lab
Department of Computer Science and Engineering (CSE)
Michigan State University (MSU)
http://biometrics.cse.msu.edu/pubs/databases.html
Databases - " PRIP Hand-Drawn Composite (PRIP-HDC) dataset", Nov. 2014. [Description] [data]
- " PRIP Viewed Software-Generated Composite dataset (PRIP-VSGC) dataset", Nov. 2014. [Description] [data]
- " UFR: Identifying a Person of Interest from a Media Collection", July 2014. [Description] [data] [protocols]
- " MSU Mobile Face Spoofing Database", June 2014. [Description]
- " FG-NET Human Age Estimation", June 2013. [Description] [download]
- " Tattoo Sketch and Image Dataset", March 2013. [Description] [download]
- " Long Distance Heterogeneous Face Database (LDHF-DB)", November 2012. [Description]
1. " PRIP Hand-Drawn Composite (PRIP-HDC) dataset", Nov. 2014.
The Pattern Recognition and Image Processing (PRIP) Hand-Drawn Composite (PRIP-HDC) database used in the paper [1] contains 265 hand-drawn composites togethr with the mated mugshots. These hand-draw composites are drawn based on the verbal description by the eyewitness or victim; so they are completely different from the othe public-domain sketch databases such as CUHK sketch database [2], which are drawn by looking at the photo. Of the 265 total hand-drawn composites, 73 were drawn by Lois Gibson [2], 43 were drawn by Karen Taylor [3], 56 were provided bythe Pinellas County Sheriff's Office (PCSO), 46 were drawn by forensic artists employed by the Michigan State Police, and 47 were downloaded from the Internet. Due to the intellectual property (IP) issue, currently we are not able to release all the 265 composites, except for the 47 hand-drawn composites publicly availabe from the Internet. This released dataset contain 47 pairs of hand-drawn composites-mugshot pairs.
2. " PRIP Viewed Software-Generated Composite dataset (PRIP-VSGC) dataset" Nov. 2014.
The PRIP Viewed Software-Generated Composite (PRIP-VSGC) database used in the paper [1] contains 123 subjects (look at the ID filed (e.g., 001, 002) in the image file name to find the correspondance) from the AR database. For each of the 123 photographs from the ARdatabase used in the PRIP-VSGC database, three composites were created. Two composites were created using FACES [2] and the third was created using Identi-Kit [3].Due to the intellectual property (IP) issue, currently we are only able to release the 123 composites generated using Identi-Kit, which were created in the work [4].
3. " UFR: Identifying a Person of Interest from a Media Collection", July 2014.
We will make the following data publicly available SOON 1) Detailed testing protocols, e.g., file names of individual subsets (gallery, probe);(2) Pose-corrected versions of all the face images in the LFW database;(3) pose-corrected video frames from the YTF database;(4) 21 forensic sketches drawn based on low-quality videos from the YTF database;(5) human estimates (by MTurk workers) of gender and race of all the subjects in LFW and YTF databases
4. " MSU Mobile Face Spoofing Database", June 2014.
Automatic face recognition is now widely used in applications ranging from de-duplication to user authentication and mobile payment for mobile phones. This popularity of face recognition has raised concerns about face spoof attacks, where a photo or video of an authorized person’s face could be used to gain access to facilities or services. Fortunately, research on face spoof detection has attracted significant attention since 2010. A number of face spoof detection techniques have been proposed along with publicly available face spoof databases for performance benchmark. Noticeable works include [1] [2] [3] from the Idiap Institute, [4] [5] [6] from the CASIA Institute and [7] from NUAA. In spite of the above endeavor, generalization ability of current face spoof detection algorithms has not been adequately addressed. Therefore, more realistic face spoof databases are needed in public domain for cross-database benchmark. And a face spoof database captured with smart phones is especially important to facilitate spoof detection research on mobile phone applications. To this end, we have collected our MSU Mobile Face Spoofing Database (MSU MFSD). The public available MSU MFSD Database for face spoof attack consists of 280 video clips of photo and video attack attempts to 35 clients. This Database was produced at the Michigan State University Pattern Recognition and Image Processing (PRIP) Lab, in East Lansing, US. Figure 1 shows example images of genuine and spoof faces of one of the subjects in the MSU database. Compared to other public domain face spoof databases, the MSU database has the following desirable properties:
Mobile phone is used to capture both genuine faces and spoof attacks, simulating the application of mobile phone unlock; The printed photos used for attacks are generated with a state of the art color printer on larger sized paper. Hence, the printed photos in the MSU database have much better quality than the printed photos in the Idiap and CASIA databases.
5. " FG-NET Human Age Estimation", June 2013.
The human age estimates for fg-net are recorded in a .csv file, with one line for each image in the dataset.Each line has the following format:ground_truth_age,file_name,human_age_estimateswhere human_age_estimates is a comma separated list of 10 human age estimates for the file.lease cite the following paper if you used our data:H. Han, C. Otto, X. Liu and A. K. Jain, "Demographic Estimation from Face Images: Human vs. Machine Performance", IEEE Transaction on Pattern Analysis and Machine Intelligence, 2014, (To appear).
6. " Tattoo Sketch and Image Dataset" March 2013
Among various soft biometric traits, tattoos, in particular, have received substantial attention over the past several years due to their prevalence among the criminal section of the population and their saliency in visual attention. However, in many cases, the tattoo image of a suspect may not be available (e.g., scenarios without surveillance cameras). In these circumstances, just like a face sketch [1], a sketch of a tattoo can be drawn based on the description provided by an eyewitness or the victim.This dataset is built for the research for matching tattoo sketches to tattoo images. A tattoo image was first shown to a subject for one minute. Ten minutes later, the subject was asked to draw a tattoo sketch (a line drawing image) on a white paper according to his/her memory. The current data set consists of 100 tattoo sketches drawn by two different subjects, each sketch has a corresponding tattoo image, which is provided by the Michigan State Police (MSP).Please cite as:Hu Han and Anil K. Jain, "Tattoo Based Identification: Sketch to Image Matching", ICB, Madrid, Spain, June 4-7, 2013. [1] Hu Han, Brendan Klare, Kathryn Bonnen, Anil K. Jain: Matching Composite Sketches to Face Photos: A Component-Based Approach. IEEE Transactions on Information Forensics and Security 8(1): 191-204 (2013).
7. " Long Distance Heterogeneous Face Database (LDHF-DB)", November 2012.
LDHF database contains both visible (VIS) and near-infrared (NIR) face images at distances of 60m, 100m, and 150m outdoors and at a 1m distance indoors. Face images of 100 subjects (70 males and 30 females) were captured; for each subject one image was captured at each distance in daytime and nighttime. All the images of individual subjects are frontal faces without glasses, and collected in a single sitting.
Short distance (1m) visible light images were collected under a fluorescent light by using a DSLR camera with Canon F1.8 lens, and NIR images were collected using the modified DSLR camera and NIR illuminator of 24 IR LEDs without visible light. Long distance (over 60m) VIS images were collected during the daytime using a telephoto lens coupled with a DSLR camera, and NIR images were collected using the DSLR camera with NIR light provided by RayMax300 illuminator.
|
发表于 2015-5-4 14:54:29
发表于 2015-5-26 13:49:47