Datasets

License Agreement


The Notre Dame Privacy-Safe Iris Presentation Attack Detection 2024 (ND-PSIPAD-2024) dataset consists of: (a) 4,167 images of authentic irises not covered by textured contact lenses, (b) 4,167 images of irises wearing various types of textured contact lenses, (c) 4,167 synthetically-generated images representing irises not covered by textured contact lenses, and (d) 4,167 synthetically-generated images mimicking irises wearing textured contact lenses. The associated metadata includes subject ID, race, gender, eye (left/right), sensor used in acquisition (for authentic images), and contact lens brand (for authentic images wearing contact lenses).

GitHub repository associated with this dataset: https://github.com/CVRL/PrivacySafeIrisPAD


License Agreement


The objective of collecting the RAITE 2023 Video dataset was to research on how person detection AI (Artificial Intelligence) models react to adversarial attacks and anomalies. Subjects, who gave consent to  participate in this collection, were asked to stand or walk in a simulated prison environment and were filmed by a few standard cameras. The subjects were asked to wear various types  of apparel, or keep banners with printed adversarial patterns to simulate visual  adversarial attacks. The dataset contains videos of subjects and metadata indicating which apparel was worn, or what other  adversarial objects were used by subjects during the experiment.


License Agreement


The WACV-2019-HDBIF Dataset is composed of 1,892 iris images used as the test set (S_test) in: A. Czajka, D. Moreira, K. Bowyer and P. Flynn, "Domain-Specific Human-Inspired Binarized Statistical Image Features for Iris Recognition," IEEE Winter Conference on Applications of Computer Vision (WACV), Waikoloa, HI, USA, 2019, pp. 959-967, doi: 10.1109/WACV.2019.00107. The dataset includes definitions of the genuine and impostor image pairs to fully replicate the evaluation as presented in the paper. 

License Agreement


The Synthetic Forensic Iris (UND-SFI-2024) dataset contains images of iris images that resemble those captured from deceased subjects by an equipment compliant with ISO/IEC 19794-6. The data is categorized into 18 disjoint ranges of PMI (Post-Mortem Interval). In each range, there are 10,000 images representing 1,000 non-existent identities. There are 10 images per “identity” that may be considered as same-eye images.



License Agreement


The Masked Physiological Monitoring (MPM) dataset contains 159 video recordings from 54 human subjects wearing protective face coverings. Each recording consists of a 1920x1080 resolution losslessly compressed RGB video recorded at 90 frames per second with simultaneous PPG collected from two fingertip oximeters. Each recording lasts a minimum of 3 minutes where subjects converse, move their head, and sit still, resulting in over 8 hours of data.



License Agreement


The Multi-Site Physiological Monitoring (MSPM) dataset consists of 103 sessions, each lasting just over 14 minutes on average, in which human subjects engage in a variety of activities designed to elicit interesting physiological phenomena such as a breath hold to increase blood pressure, or to provide a challenging context for performing remote photoplethysmography (rPPG) such as an adversarial attack. Sessions were recorded in RGB from three different angles and near-infrared zoomed in on the eyes, along with cardiac pulse at ten sites across the body, blood oxygenation, and blood pressure using a cuff-based monitor.



License Agreement


The UND AAAI 2023 Dataset contains (a) images of live (authentic) faces, (b) images of faces synthetically generated by deep learning-based generative adversarial networks, and (c) regions annotated by humans solving the synthetic face detection task, indicating features supporting their decisions.


This dataset contains modified samples from the Flickr-Faces-HQ (FFHQ), made available under Creative Commons BY-NC-SA 4.0 license by NVIDIA Corporation (https://github.com/NVlabs/ffhq-dataset/blob/master/LICENSE.txt).

According to that license, one is allowed to redistribute and adapt FFHQ samples for non-commercial purposes, as long as one (a) gives appropriate credit by citing the FFHQ creator’s paper, (b) indicate any changes that one made, and (c) distribute any derivative works under the same license. In response to these requirements, we: (a) cited the paper indicated at https://github.com/NVlabs/ffhq-dataset in the paper publishing the UND AAAI 2023 Dataset, (b) inform that the modifications made to the original FFHQ samples include cropping the image around the detected face and rescaling such cropped samples to the 224x224 pixel resolution, and (c) the derivative work is distributed as the AAAI 2023 paper.

https://github.com/CVRL/AI-Guidance/  




License Agreement


The UND WACV 2023 CYBORG Dataset contains (a) images of live (authentic) faces, (b) images of faces synthetically generated by deep learning-based generative adversarial networks, and (c) regions annotated by humans solving the synthetic face detection task, indicating features supporting their decisions.


This dataset contains modified samples from the Flickr-Faces-HQ (FFHQ), made available under Creative Commons BY-NC-SA 4.0 license by NVIDIA Corporation (https://github.com/NVlabs/ffhq-dataset/blob/master/LICENSE.txt). According to that license, one is allowed to redistribute and adapt FFHQ samples for non-commercial purposes, as long as one (a) gives appropriate credit by citing the FFHQ creator’s paper, (b) indicate any changes that one made, and (c) distribute any derivative works under the same license. In response to these requirements, we: (a) cited the paper indicated at https://github.com/NVlabs/ffhq-dataset in the paper publishing the UND WACV 2023 CYBORG Dataset, (b) inform that the modifications made to the original FFHQ samples include cropping the image around the detected face and rescaling such cropped samples to the 224x224 pixel resolution, and (c) the derivative work is distributed as the AAAI 2023 paper. UND WACV 2023 CYBORG Dataset contains modified samples from the (https://creativecommons.org/licenses/by-nc/4.0/legalcode).

According to the request of the licensor (https://github.com/tkarras/progressive_growing_of_gans), one is allowed to use any of the material in their own work, as long as appropriate credit is given to the creators by mentioning the title and author list of their paper: Tero Karras, Timo Aila, Samuli Laine, Jaakko Lehtinen, “Progressive Growing of GANs for Improved Quality, Stability, and Variation,” ICLR 2018.



LivDet-Iris-2023-Part1-Notre Dame License Agreement, LivDet-Iris-2023-Part1-Clarkson License Agreement


LivDet-Iris-2023 dataset contains images of live (authentic) irises and images of irises synthetically-generated by deep learning-based generative adversarial networks. The primary goal of creating and sharing this dataset is to allow researchers to participate in LivDet-Iris 2023 competition by delivering to the organizers the presentation attack detection scores associated with these images. After the LivDet-Iris 2023 competition is concluded, this dataset may be a useful benchmark allowing to compare future solutions with those submitted to the competition.


License Agreement


The Multi-Demographic Retouched Faces (MDRF) database has a total of 3600 images, where 1200 belong to original class and 2400 belong to retouched class; 1200 belong to each of the Caucasian, East Asian, and South Asian classes; and 1800 each to the male and female classes. Human responses to the task of detecting retouching in a subset of these images have also been included. The shared folder includes all the Caucasian images and instructions and resources for the other two origin groups.


The goal here is to analyze the effects of retouching on faces from different demographics and test performance of retouching detection algorithms.

License Agreement


This includes a database of Nigerian face images, that was collected by the Biometrics Vision and Computing (BVC) group at the University of Nigeria in 2018. The BVC group through the University of Nigeria retains ownership and copyright of the BVC-face-poses dataset. The BVC-face data is distributed via the University of Notre Dame.


License Agreement


This data set is a non-sequestered partition of the VII-Q dataset, and it contains 4466 near-infrared (NIR) images of 203 distinct eyes (103 left and 100 right eyes) of 144 distinct human subjects.

All data is de-identified. Assembly of this data set was supported by the US National Institute of Standards and Technology.


License Agreement


The BVC-UNN-face data set was collected by the Biometrics Vision and Computing (BVC) group at the University of Nigeria. It includes a database of face images of Nigerians.

The BVC group through the University of Nigeria retains ownership and copyright of the BVC-UNN-face dataset. This data is distributed via the University of Notre Dame upon receipt of a properly executed copy of the license agreement.

For details on publishable images, please click here.




License Agreement


This dataset contains images of 575 subjects who identified as Caucasian female, with a total of 6,669 images, and of 687 subjects who identified as Caucasian male, with a total of 8,419 images.

This dataset may be useful for studying accuracy differences across female / male demographics.


Data Type: RGB Face Video, Pulse waveforms and Heart rate, Approximate Download Size: 7 TB



This dataset overlaps with DDPM specifically for remote pulse detection. It consists of losslessly compressed RGB videos and ground truth pulse waveforms and heart rate (HR) for 86 subjects. The data was collected in an interview scenario with subjects freely moving, talking, and exhibiting facial expressions. Each video lasts around 10 minutes, recorded at 90 frames-per-second giving several million visible-light frames at 1920x1080 resolution. Pulse data was interpolated to the video sampling rate, such that each frame has a waveform and HR pair. Subject metadata describing age, gender, race, and ethnicity is included. Predefined train, validation, and test splits are also included to compare with results presented in the original paper.


Data Type: Video with corresponding force plate data, Approximate Download Size: 20 GB 


License Agreement


This dataset consists of videos of 89 female athletes performing 582 evaluative jumps for the purpose of predicting ACL injury risk.  The dataset includes three videos from different angles for each jump as well as force plate data.  For more information please see the detailed description Here.



Data Type: Visible Face Images, Approximate Download Size: 4.2 GB


License Agreement


The TIM test includes a total of 675 images organized into 225 image triads. The triads comprise two image of the same identity and one image of a different identity. The task is to select the odd-one-out (i.e., the image of the different identity). The images were sampled from the Good, the Bad, and the Ugly Challenge  and show frontal view of faces with variation in illumination, expression, and subject appearance (hair, accessories). The age of the subjects can be estimated assuming that the photos were taken in between 2004 and 2005.

Data Type: RGB Face Video, NIR Face Video, LWIR Face Video, Pulse waveforms, Heart rate, Approximate Download Size: 12 TB


License Agreement


The Deception Detection and Physiological Monitoring (DDPM) dataset captures an interview scenario in which the interviewee attempts to deceive the interviewer on selected responses. The interviewee is recorded in RGB, near-infrared, and long-wave infrared, along with cardiac pulse, blood oxygenation, and audio. After collection, data were annotated for interviewer/interviewee, curated, ground-truthed, and organized into train/test parts for a set of canonical deception detection experiments. The dataset contains almost 13 hours of recordings of 70 subjects, and over 8 million visible-light, near-infrared, and thermal video frames, along with appropriate meta, audio, and pulse oximeter data. 



Data Type: IR Iris Still, Approximate Download Size: 2.7 GB


License Agreement


The NDIris3D dataset contains a total of 6,850 images: 3,488 images acquired by LG4000, and 3,362 images acquired by AD100 from the same 89 subjects with and without textured contact lenses, and for varying illumination setups in both LG4000 and AD100 sensors. The dataset may be used in research on iris presentation attack detection (especially related to application of photometric stereo in PAD), or to assess the impact of contact lenses on matching performance. Note: the NDIris3D dataset is part of the larger LivDet-Iris 2020 test set. If you plan to test you algorithms with Notre Dame’s part of the LivDet-Iris 2020 benchmark, use NDIris3D appropriately to make fair comparisons with the LivDet competition winner (e.g., do not use NDIris3D in training).

Data Type: IR Iris Still, Approximate Download Size: 2.4 GB


License Agreement


This database offers iris images (with and without contact lenses) of the same eyes captured shortly one after another with illumination coming from two different locations. 5,796 iris images in total were acquired by the LG IrisAccess 4000 sensor from 119 subjects. This set is divided into four subsets used in the experiments: (a) 1,800 images of irises wearing regular (with dot-like pattern) textured contact lenses, as shown in Fig. 6a in the WACV 2019 paper; (b) 864 images of irises wearing irregular (without dot-like pattern) textured contact lenses, as shown in Fig. 6b in the WACV 2019 paper; (c) 1,728 images of irises wearing clear contact lenses (without any visible pattern), and (d) 1,404 images of authentic irises without any contact.

Data Type: 3D Face, Visible Face Images, Approximate Download Size: 72 GB


Information about the FRGC program may be found here. Note: the FRGC 1.0a data collection has been superseded by the FRGC v2.0 collection and is no longer available. As of 4/2/2014 the Bee Software has been separated from the FRGC 2.0 dataset. If you wish to download it as well as the FRGC 2.0 dataset, please download and execute the Bee software license agreement. Publications using this database must cite the paper listed in the license agreement.



Data Type: Face 3D, Approximate Download Size: 29 GB


The ND-2006 dataset contains a total of 13,450 images containing 6 different types of expressions (Neutral, Happiness, Sadness, Surprise, Disgust and Other). A total of 888 distinct persons, with as many as 63 images per subject are available in this data set. The data set corresponds exactly to the data set described in: Faltemier, T.C.; Bowyer, K.W.; Flynn P.J.; Using a Multi-Instance Enrollment Representation to Improve 3D Face Recognition, Proc. First IEEE International Conference on Biometrics: Theory, Applications, and Systems, September 2007, pp.1 - 6. To obtain this data set, retrieve the license agreement and follow instructions above.  Publications using this database must cite the paper listed in the license agreement.


Data Type: IR Iris Still, Approximate Download Size: 20.5 GB


A technical report describing this data can be found here. The data set contains 64,980 iris images obtained from 356 subjects (712 unique irises) between January 2004 and May 2005. To obtain this data set, retrieve the license agreement and follow instructions above.
Publications using this database must cite the paper listed in the license agreement.

Data type: Face Still image and Face Video, Approximate Download Size: 71 GB


Information about the PaSC effort may be found at http://www.nist.gov/itl/iad/ig/face.cfm and http://www.cs.colostate.edu/pasc. To obtain this data set, retrieve the license agreement and follow instructions above. 
Publications using this database must cite the paper listed in the license agreement.

Data Type: Face 3D, Approximate Download Size: 1.5 GB


This data set contains 3D face scans for 107 pairs of twins. There are 107 x 2 = 214 individuals, each with a 3D face scan with a smiling expression and a scan with a neutral expression, and so 214 x 2 = 428 scans. The scans were acquired with a Minolta Vivid 910. To obtain this data set, retrieve the license agreement and follow instructions above.  
Publications using this database must cite the paper listed in the license agreement.

Data type: IR & Visible Face Still, Approximate Download Size: 404 GB


The image collection comprises visible-light and near-IR face images of 574 subjects acquired from fall 2011 to spring 2012. There are a total of 2,341 visible-light face images of the 574 persons. There are a total of 22,264 near-IR face images, coming from two 230 subject-sessions. A total of 402 subjects had both visible-light and near-IR face images collected in one or more sessions in fall 2011 and also one or more sessions in spring 2012. To obtain this data set, retrieve the license agreement and follow instructions above. 
Publications using this database must cite the paper listed in the license agreement.


Data type: IR Iris still, Approximate Download Size: 122 GB


This data set was initially released for the Cross Sensor Iris Recognition Challenge associated with the BTAS 2013 Conference. This dataset consists of 27 sessions of data with 676 unique subjects. An average session contains 160 unique subjects which have multiple images from both the LG2200 and LG4000 iris sensors.  There are 29,986 images from the LG4000 and 116,564 images from the LG2200. Every subject occurs in at least two sessions across the entire data set. This data set spans three years, 2008 to 2010. The initial images are taken from both sensors and are 640 by 480. There are additional images included in this data set, known as the modified LG2200 images.  The original images have been stretched vertically by 5% to compensate for the non-unit aspect ratio of the digitizer used in the LG2200 computer-hosted runtime acquisition system (this elongation was suggested by Imad Malhas of IrisGuard inc.  in 2009). Hence these additional images are of size 640 by 504.  To obtain this data set, retrieve the license agreement and follow instructions above. 
Publications using this database must cite the paper listed in the license agreement.


Data Type: 3D + 2D Ear Images, Approximate Download Size: 35 GB


1800 3D (and corresponding 2D) profile (ear) images from 415 human subjects captured between 2003 and 2005. Corresponds to data used in Yan and Bowyer, "Biometric recognition using three-dimensional ear shape," PAMI 29(8), August 2007.

Data Type: Visible Face Images, Visible Face Video, NIR Iris and Ocular Region, Approximate Download Size: 98.5 GB


Information about the FOCS program may be found here. Please be sure to execute all three required documents to receive download information. Publications using this database must cite the paper listed in the license agreement.

Data type: Face Still image, Approximate Download Size: 4.3 GB


The ND-IIITD Retouched Faces database is a dataset of original face images and retouched versions of those face images. The database contains 2600 original images and 2275 altered images.  It is meant for use in the problem of developing methods to classify a face image as original or retouched.  To obtain this data set, retrieve the license agreement and follow instructions above. 
Publications using this database must cite the paper listed in the license agreement.

Data type: Face Still image, Approximate Download Size: 2.2 GB


To obtain this data set, retrieve the license agreement and follow instructions above. Publications using this database must cite the paper listed in the license agreement.

Data type: IR Iris still, Approximate Download Size: 3.9 GB


This data set contains iris images of subjects without contact lenses, with soft contact lenses, and with cosmetic contact lenses, acquired using an LG4000 and an IrisGuard AD100 iris sensor. The data set contains 4,200 TIFF files from the LG4000 sensor, 900 TIFF images from the AD100 sensor, and four metadata files describing the images.  For a more thorough description, please see the README document. This data set corresponds to that in:  Doyle, J.S; Bowyer, K.W.; Flynn, P.J., "Variation in accuracy of textured contact lens detection based on sensor and lens pattern," Biomentrics: Theory, Applications and Systems (BTAS), 2013 IEEE Sixth International Conference on, vo.,no.,pp1,7, Sept. 29 2013-Oct.2 2013. To obtain this data set, retrieve the license agreement and follow instructions above. 
Publications using this database must cite the paper listed in the license agreement.

Data type: IR Iris Still, Approximate Download Size:1 GB


This is a set of iris images acquired using an LG 4000 sensor. It is divided into left and right iris images, and it includes the gender of each subject.  In the basic part of the dataset there is one image per left and right iris of each of 750 males and 750 females, for 3000 total images. An additional part of the dataset, for a different set of subjects, includes three images per iris. 
Publications using this database must cite the paper listed in the license agreement.

Data type: IR Iris Still image, Approximate Download Size: 1.5 GB


This is a full set of iris images, with and without textured contact lenses, used in the LivDet-Iris 2017 competition (http://iris2017.livdet.org). It is built with samples taken from the Notre Dame Contact Lens Detection 2015 (NDCLD15). The training subset consists of 600 images of authentic irises (with no contacts, either soft or cosmetic) and 600 images of textured contact lenses manufactured by Ciba, UCL and ClearLab. The testing subset is split into "known spoofs" and "unknown spoofs”. The "known spoofs" dataset includes 900 images of textured contact lenses produced by Ciba, UCL and ClearLab (as in the training set) and 900 images of authentic irises. The unknown spoofs dataset includes 900 images of textured contact lenses produced by Cooper and J&J (i.e., not represented in the training set) and 900 images of authentic irises. All images were acquired using either an IrisAccess LG4000 or an IrisGuard AD100 sensor.  To obtain this data set, retrieve the license agreement and follow instructions above. Publications using this database must cite the paper listed in the license agreement.


Data Type: Face Still, Approximate Download Size: 57.5 GB


33,287 visible-light frontal face images captured from 487 human subjects from 2002 through 2004.  Each subject was photographed with a high-resolution digital camera (1600 x 1200 or 2272 x 1704) under different lighting and expression conditions.  Many subjects were photographed every week for 10 weeks in the Spring of 2002, 13 weeks in the fall of 2002 and 15 weeks in the spring of 2003.  The number of images per subject ranges from 4 to 227 with an average of 68.  Hence, this database provides a significant amount of "repeat data" to assess performance of face recognition systems with respect to time elapsed since enrollment.

Data Type: 3D + 2D Face Images, Approximate Download Size: 2.5 GB


953 3D ( and corresponding 2D) frontal face images from 277 human subjects captured in 2003.  These images were acquired with a Minolta Vivid 900 3D range scanner.

Data Type: IR Iris Still, Approximate Download Size: 7 GB


This is a data set of iris images that was used in a study of the effects of wearing contact lenses on the performance of iris recognition: "Degradataion of Iris Recognition Performance Due to Non-Cosmetic Prescription Contact Lenses", Sarah E. Baker, Amanda Hentz, Kevin W. Bowyer, and Patrick J. Flynn, Computer Vision and Image Understanding 114 (9), 1030-1044, September 2010.  To obtain this data set, retrieve the license agreement and follow instructions above. 
Publications using this database must cite the paper listed in the license agreement.

Data Type: Face Still, Approximate Download Size: 250 GB


This data set contains 24,050 color photographs of the faces of 435 attendees at the Twins Days Festivals in Twinsburg, Ohio in 2009 and 2010. All images were captured by Nikon D90 SLR cameras. Images were captured under natural light in "indoor" and "outdoor" configurations ("indoor" was a tent). Facial yaw varied from -90 to +90 degrees in steps of 45 degrees (zero degrees was frontal). To obtain this data set, retrieve the license agreement and follow instructions above. 
Publications using this database must cite the paper listed in the license agreement.

Data Type: IR + Visible Face Images, Approximate Download Size: 3 GB


2292 IR frontal face images and 2292 visible frontal face images from 82 human subjects captured from 2002-2004.

Data Type: IR Face and Iris Video, IR Iris Video, Visible Face Images and Video, Approximate Download Size: 110 GB (+UTD data: 36 GB)


Information about the MBGC program maybe found here. The MBGC v2 and MBGC v1 data sets are separately licensed; MBGC v1 licensees must complete a new license for MBGC v.2.  Please be sure to complete all three required forms to receive the full MBGC 2.0 data collection. To obtain this data set, retrieve the license agreement and follow instructions above. 
Publications using this database must cite the paper listed in the license agreement.

Data Type: Visible Face Video , Approximate Download Size: 36 GB 


Notre Dame distributes a subset of the Database of Moving Faces and People data set, assembled by A.J. O'Toole and H. Abdi at the University of Texas at Dallas. The subset contains between 1 and 9 videos of 297 unique human subjects, with a total of 1019 videos  and a data set size of 36GB. 

To obtain this data set, you must agree to, and your institution must execute, both the data license agreement and the permission form.

Data Type: Visible Light Ear, Approximate Download Size: 487 MB


464 visible light profile (ear) images from 114 human subjects captured in 2002.

Approximate Download Size: 2.5 GB


942 3D (and corresponding 2D) profile (ear) images from 302 human subjects captured in 2003 and 2004.

Data type: IR Iris Still, Approximate Download Size: 2.7 GB


As detailed in Robust Detection of Textured Contact Lenses in Iris Recognition Using BSIF, James S. Doyle and Kevin W. Bowyer, IEEE Access, Volume 3, 2015. Digital Object Identifier 10.1109/ACCESS.2015.2477470, this database of 7300 images was constructed to evaluate contact lens detection under various experimental scenarios. The main dataset is composed of 6000 images for model training and 1200 images for model evaluation. Images were acquired using either an IrisAccess LG4000 or an IrisGuard AD100 sensor; both sensors are equally represented. The dataset is composed of images from one of three equally-represented classes: No Lens, Soft Lens, and Textured Lens.

Data type: Face Still image, Approximate Download Size: 200 GB


To obtain this data set, retrieve the license agreement and follow instructions above. Publications using this database must cite the paper listed in the license agreement.

Data Type: 3D + 2D Ear Images, Approximate Download Size: 2 GB


738 3D (and corresponding 2D) profile (ear) images from 235 human subjects captured between 2003 and 2005.

Data Type: IR Iris, Approximate Download Size: 4 GB


This data set contains sequences of iris images of different persons, acquired using an LG4000 iris sensor.  Images are from Spring 2008, Spring 2009, and Spring 2010. This allows two different one-year template aging studies, 2008-2009 and 2009-2010, and one two-year template aging study, 2008- 2010. The data set contains 11,776 TIFF files.  It is used in the paper: Analysis of Template Aging in Iris Biometrics, Samuel P. Fenker and Kevin W. Bowyer, IEEE Compter Society Biometrics Workshop, June 2012. To obtain this data set, retrieve the license agreement and follow instructions above. 
Publications using this database must cite the paper listed in the license agreement.

Data Type: IR Face and Iris Video, IR Iris Video, Visible Face Images and Video, Approximate Download Size: N/A


Information about the MBGC program may be found here. To obtain this data set, retrieve the license agreement and follow instructions above. 
Publications using this database must cite the paper listed in the license agreement.


Data type: IR Iris still, Approximate Download Size: 2.2 GB


This data set contains 6797 images collected from 23 subjects (46 different irises) between January 2004 and October 2008. It corresponds to the data set used in a chapter in the book: Template Aging in Iris Biometrics:  Evidence of Increased False Reject Rate in ICE 2006, Sarah Baker, Kevin W. Bowyer, Patrick K. Flynn and P. Jonathon Phillips, in Handbook of Iris Recognition, Mark Burge and Kevin W. Bowyer, editors, Springer, 2012.  to demonstrate the effects of elapsed time between probe and gallery image acquisition on iris recognition system performance.
To obtain this data set, retrieve the license agreement and follow instructions above. Publications using this database must cite the paper listed in the license agreement.

Data Type: LWIR Face Still, Approximate Download Size: 360 MB


2,492 LWIR frontal face images from 241 human subjects captured in 2002.  All images are 320 x 240 and were captured by a Merlin-Uncooled camera purchased from Indigo Systems in 2001.

Data Type: 3D + 2D Hand Images, Approximate Download Size: 14.5 GB


1191 3D (and corresponding 2D) images of the back (non-palm) portion or 223 different human hands captured between 2003 and 2005.

Data type: Group Video, Approximate Download Size: 2.1 GB


Comprising 190 subjects recorded in 28 crowd videos over a two year period, SN-Flip captures variations in illumination, facial expression, scale, focus, and pose. The videos were recorded with point-and-shoot camcorders from the Cisco Flip family of products, so the image quality is representative of typical web videos.  Ground truth information for subject identities and social groups is included to facilitate future research in vision-driven social network analysis.
To obtain this data set, retrieve the license agreement and follow instructions above. Publications using this database must cite the paper listed in the license agreement.


Data type: IR Iris still, Approximate Download Size: 83.5 GB


This data set was initially released for the Cross Sensor Iris Recognition Challenge associated with the BTAS 2012 conference. This dataset consists of 27 sessions of data with 676 unique subjects. An average session contains 160 unique subjects which have multiple images from both the LG2200 and LG4000 iris sensors.  There are 29,939 images from the LG4000 and 117,503 images from the LG2200. Every subject occurs in at least two sessions across the entire data set. This data set spans three years, 2008 to 2010. The initial images are taken from both sensors and are 640 by 480. There are additional images included in this data set, known as the modified LG2200 images.  The original images have been stretched vertically by 5% to compensate for the non-unit aspect ration of the digitizer used in the LG2200 computer-hosted runtime acquisition system (this elongation was suggested by Imad Malhas of IrisGuard Inc. in 2009). Hence these additional images are of size 640 by 504. 
To obtain this data set, retrieve the license agreement and follow instructions above. Publications using this database must cite the paper listed in the license agreement.

Data Type: Group Video, Approximate Download Size: 260 MB


The database contains 14 crowd videos of 90 subjects, five of whom appear in multiple videos and 85 of whom appear in one video. These videos were acquired between November 2009 and May 2010.
To obtain this data set, retrieve the license agreement and follow instructions above. Publications using this database must cite the paper listed in the license agreement.

Dataset Description

Data Type: Video, Approximate Download Size: 3.3 GB


License Agreement


The VBOLO dataset was collected in several sessions, at various checkpoints within public transportation facilities such as tunnels, bridges, and hallways. These capture environments include different camera mount heights and depression angles, illuminations, backgrounds, resolutions, pedestrian poses, and distractors. This dataset provides a good scenario for the facial ReID problem. This dataset uses a small set of known individuals - ``actors'', who move in and out of the surveillance cameras' fields of view, together with the unknown persons denoted as ``distractors''. The ``actors'' change clothing randomly between each ``appearance'' in a camera's field of view.

Compared to a typical body-based ReID dataset, which has only a few images for each subject, the VBOLO dataset has a large number of annotations for each subject from consecutive video frames, which mimic a real scenario for surveillance tracking and detection. This is significantly challenging for matching, because: 1) Faces change size significantly e.g. , from 12x12 to 150x150) and exhibit significant pose variations as well. 2) The cameras supplying the probe and gallery images may have different resolutions and points of view.


Dataset


Advances in image restoration and enhancement techniques have led to discussion about how such algorithms can be applied as a pre-processing step to improve automatic visual recognition. In principle, techniques like deblurring and super-resolution should yield improvements by de-emphasizing noise and increasing signal in an input image. But the historically divergent goals of computational photography and visual recognition communities have created a significant need for more work in this direction. To facilitate new research, we introduce a new benchmark dataset called UG^2, which contains three difficult real-world scenarios: uncontrolled videos taken by UAVs and manned gliders, as well as controlled videos taken on the ground.

Data type: Face Still image, Approximate Download Size: 247 MB


Images used in the experiments in “Face Recognition Accuracy of Forensic Examiners, Superrecognizers, and Algorithms,” Proceedings of the National Academy of Sciences, 2018 (load paper from https://doi.org/10.1073/pnas.1721355115). Publications using this data set must cite the paper listed in the license agreement.

Data type: feature patterns, Approximate Download Size: 192 KB


The FACE Features Set comprises feature patterns for imagery that is amenable to human-assisted face clustering. The features were computed for faces observed in blurry point-and-shoot videos, images of women seen before and after the application of makeup, and photographs of twins. 
Please follow the instructions in the README file regarding citation.

Data Type: Synthetic Face Images, 3D Head Models, Approximate Download Size: 211 GB


License Agreement


The dataset contains two types of data:

1. A set of 3D head models (.abs files) and their corresponding 2D RGB registration image (.ppm files), obtained using a Konica-Minolta ‘Vivid 910’ 3D scanner, of real identities (subjects), either Male or Female in gender, and Caucasian or Asian in ethnicity. 

2. A set of RGB face images, masked faces without context and background 800x600 in size, of fully synthetic subjects (identities) that do not exist in reality. The synthetic identities are generated by consistent sampling of facial parts from face images of different real identities, sampled from, either Male or Female in gender, and Caucasian or Asian in ethnicity.

Since all the identities in this dataset are synthetic, i.e. they do not exist, they can be used freely without any privacy concerns. These synthetic face images were generated using Python and OpenGL, with minimal training, and can be used as – (1) supplemental training data to train CNNs, (2) additional distractor face images in the gallery for face verification experiments.