Probabilistic Constrained Adaptive Local Displacement Experts
Abstract
Generic non-rigid face fitting, namely the task of finding the configuration of a shape model describing a face in an image under variations in identity, illumination, pose and expression, is addressed in this work through an ensemble of local patch-based displacement experts. To account for appearance variations, these displacement experts are parameterized bilinearly, allowing the experts to adapt to the face at hand. The problem is formulated probabilistically, where the objective is to maximize the likelihood of predicted displacements marginalized over the adaptation parameters. The efficacy of the proposed formulation is compared empirically against a two existing methods.
Cite
Text
Saragih et al. "Probabilistic Constrained Adaptive Local Displacement Experts." IEEE/CVF International Conference on Computer Vision Workshops, 2009. doi:10.1109/ICCVW.2009.5457686Markdown
[Saragih et al. "Probabilistic Constrained Adaptive Local Displacement Experts." IEEE/CVF International Conference on Computer Vision Workshops, 2009.](https://mlanthology.org/iccvw/2009/saragih2009iccvw-probabilistic/) doi:10.1109/ICCVW.2009.5457686BibTeX
@inproceedings{saragih2009iccvw-probabilistic,
title = {{Probabilistic Constrained Adaptive Local Displacement Experts}},
author = {Saragih, Jason M. and Lucey, Simon and Cohn, Jeffrey F.},
booktitle = {IEEE/CVF International Conference on Computer Vision Workshops},
year = {2009},
pages = {288-295},
doi = {10.1109/ICCVW.2009.5457686},
url = {https://mlanthology.org/iccvw/2009/saragih2009iccvw-probabilistic/}
}