The Headspace dataset

The Headspace dataset is a set of 3D images of the human head, consisting of 1518 subjects wearing tight fitting latex caps to reduce the effect of hairstyles. Due to poor fitting of some caps and/or excessive hair bulge effects and other reasons (eg. declared head trauma), we excluded 306 subjects when building our 3D morphable models, thus using 1212 3D images evenly split between males and females. The indices of the subjects that we used are provided with the dataset.

The data is in OBJ format with BMP textures and is available online for download. The dataset is subject to the same non-commercial restrictions as the LYHM models above, and the same user agreement form can be used. The download file is a compressed tarball (headspace.tar.gz), with file size 38GB.

Directories are per subject and named with subject indices from 00001 to 01518. Within each directory are the following files, where root is 12 digit 3D image identifier.

root.obj : ASCII geometry file with 3D points, mesh connectivity and texture coordinates
root.bmp : the bit mapped texture file (available for over 80% of subjects)
root.mtl : material file for OBJ rendering
subject.txt : contains age/gender etc information of the subject
landmarks.txt : a set of 3D landmark coordinates extracted using the Zhu-Ramanan mixture-of-trees algorithm.

Matlab code is included for viewing the dataset, or other utilities can be used, such as MeshLab.

For information on how to access the data, follow the link to the Liverpool-York Head Model website. Note that restrictions apply to use of the data.