The FaceGrabber Database and Software

The FaceGrabber Software


FaceGrabber is an open source software tool for capturing facial videos with Microsoft Kinect 2.0. (UPDATE: FaceGrabber now also supports Kinect 1.0). FaceGrabber is written in C++ and permits the concurrent recording of various formats, including the raw 2D and 2.5D video streams, 3D point clouds and the 3D registered face model provided by the Kinect. (UPDATE: FaceGrabber now also supports registered 2D outputs as well as infrared captures). The software is also able to convert different data formats and playback recorded results directly in 3D. More details can be found in the documentation or in our paper.


The FaceGrabber Database


We publish a public video face database which was captured using FaceGrabber. The database comprises 40 individuals (33 male), performing the six universal emotions (disgust, sadness, happiness, fear, anger, surprise) and two additional sequences. The first additional sequence is a six-second (neutral) head scan sequence, the second one is a ten-second do-what-you-like sequence.


Acquiring the FaceGrabber Software and/or Database


We would like to hear from your work with FaceGrabber, so feel free to let us know and we can link to your project page and/or paper!

The FaceGrabber software is released under an MIT license. It is available on GitHub via the following link: https://github.com/DanielMerget/FaceGrabber

For convenience, we provide a compiled executable version of FaceGrabber together with all required library .dll files (this version is outdated by now). The .pdb files are optional in case you want to debug the application.

The password for all downloads is FaceGrabber@MMK (case-sensitive):

FaceGrabber_binary_x64.zip (~45 MB) outdated! Check out our github page and build FaceGrabber from source if you need the new features (https://github.com/DanielMerget/FaceGrabber).

FaceGrabber_binary_x64_pdb_files.zip (~7 MB) outdated! See above.


The FaceGrabber database is released under an MIT license. Since the raw database is very big (>50GB), we provide a JPEG-compressed version (~9GB). If you specifically need the raw data, please contact us directly. All files are in binary format. If you need other formats such as ASCII, you can use the FaceGrabber software to convert it. We furthermore separate the different capture types in order to reduce the download size, e.g., if only a subset of the data is needed. We kindly ask you to download the preview first, this helps you find out which data you actually need and keeps unnecessary traffic to a minimum.

Finally, we provide miscellaneous data (age, gender) plus some Matlab-code to give an idea of how to read the database.

The password for all downloads is FaceGrabber@MMK (case-sensitive):

Preview.zip (~175 MB)

FaceGrabberDB_KinectColorRaw.zip (~3.1 GB)

FaceGrabberDB_KinectDepthRaw.zip (~2.7 GB)

FaceGrabberDB_Landmarks2D.zip (~260 MB)

FaceGrabberDB_Landmarks3D.zip (~350 MB)

FaceGrabberDB_FaceRaw.zip (~ 900 MB)

Misc.zip (~ 4KB)


Acknowledgement


If you use the FaceGrabber software or database for your research, please cite our paper:

D. Merget, T. Eckl, M. Schwörer, P. Tiefenbacher, and G. Rigoll, “ Capturing Facial Videos with Kinect 2.0: A Multithreaded Open Source Tool and Database”, in Proc. WACV, IEEE, 2016.


Contribute to FaceGrabber


We always appreciate contributions to FaceGrabber, whether it be a tiny patch for a bug in the software, a new feature, newly captured individuals for the database, or something else. For additions to the database, please make sure that an appropriate form of consent (including at least: non-commercial use and publication) is signed for every individual. We are looking forward to hearing from you.


Contact


For any inquiries or questions, please contact Daniel Merget.

hoch