Statistically Learned Deformable Eye Models

Abstract

In this paper we study the feasibility of using standard deformable model fitting techniques to accurately track the deformation and motion of the human eye. To this end, we propose two highly detailed shape annotation schemes (open and close eyes), with $+30$ feature landmark points, high resolution eye images. We build extremely detailed Active Appearance Models (AAM), Constrained Local Models (CLM) and Supervised Descent Method (SDM) models of the human eye and report preliminary experiments comparing the relative performance of the previous techniques on the problem of eye alignment.

Cite

Text

Alabort-i-Medina et al. "Statistically Learned Deformable Eye Models." European Conference on Computer Vision Workshops, 2014. doi:10.1007/978-3-319-16178-5_19

Markdown

[Alabort-i-Medina et al. "Statistically Learned Deformable Eye Models." European Conference on Computer Vision Workshops, 2014.](https://mlanthology.org/eccvw/2014/alabortimedina2014eccvw-statistically/) doi:10.1007/978-3-319-16178-5_19

BibTeX

@inproceedings{alabortimedina2014eccvw-statistically,
  title     = {{Statistically Learned Deformable Eye Models}},
  author    = {Alabort-i-Medina, Joan and Qu, Bingqing and Zafeiriou, Stefanos},
  booktitle = {European Conference on Computer Vision Workshops},
  year      = {2014},
  pages     = {285-295},
  doi       = {10.1007/978-3-319-16178-5_19},
  url       = {https://mlanthology.org/eccvw/2014/alabortimedina2014eccvw-statistically/}
}