MPII Human Shape |
MPII Human Shape is a family of expressive 3D human body shape models and tools for human shape space building, manipulation and evaluation. Human shape spaces are based on the widely used statistical body representation and learned from the CAESAR dataset, the largest commercially available scan database to date. As preprocessing several thousand scans for learning the models is a challenge in itself, we contribute by developing robust best practice solutions for scan alignment that quantitatively lead to the best learned models. We make the models as well as the tools publicly available. Extensive evaluation shows improved accuracy and generality of our new models, as well as superior performance for human body reconstruction from sparse input data.
@article{pishchulin17pr,
author = {Leonid Pishchulin and Stefanie Wuhrer and Thomas Helten and Christian Theobalt and Bernt Schiele}
title = {Building Statistical Shape Spaces for 3D Human Modeling},
journal = {Pattern Recognition},
year = {2017}
}
We are making shape spaces, fitted scans and code freely available for non-commercial and educational purposes only.
Statistical human shape spaces learned from the meshes fitted to the pre-processed CAESAR body scans:
WSX and NH denote additional posture normalization of the pre-processed scans using the methods of Wuhrer et al. and Neophytou & Hilton, respectively.
CAESAR-fitted meshes used for learning statistical human shape spaces:
CAESAR + WSX fitted meshes (364 MB)
CAESAR + NH fitted meshes (363 MB)
Source code (GitHub) for data pre-processing, 3D human shape modeling, fitting and evaluation.
Below we visualize the first 10 PCA components of the learned human shape spaces. Components correspond to extreme shape variations in each dimension.
Building Statistical Shape Spaces for 3D Human Modeling
Leonid Pishchulin, Stefanie Wuhrer, Thomas Helten, Christian Theobalt and Bernt Schiele
Pattern Recognition, 2017