Explicit occlusion detection based deformable fitting for facial landmark localization

Xiang Yu, Fei Yang, Junzhou Huang, Dimitris N. Metaxas

Research output: Chapter in Book/Report/Conference proceedingConference contribution

11 Scopus citations

Abstract

This paper addresses the problem of facial landmark localization on partially occluded faces. We proposes an explicit occlusion detection based deformable fitting model for occluded landmark localization. Most recent shape registration methods apply landmark local search and attempt to simultaneously minimize both the model error and localization error. However, if part of the shape is occluded, those methods may lead to misalignment. In this paper, we introduce regression based occlusion detection to restrict the occluded landmarks' error propagation from passing to the overall optimization. Assuming the parameter model being Gaussian, we propose a weighted deformable fitting algorithm that iteratively approaches the optima. Experimental results in our synthesized facial occlusion database demonstrate the advantage of our method.

Original languageEnglish (US)
Title of host publication2013 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition, FG 2013
DOIs
StatePublished - 2013
Event2013 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition, FG 2013 - Shanghai, China
Duration: Apr 22 2013Apr 26 2013

Publication series

Name2013 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition, FG 2013

Other

Other2013 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition, FG 2013
Country/TerritoryChina
CityShanghai
Period4/22/134/26/13

All Science Journal Classification (ASJC) codes

  • Computer Vision and Pattern Recognition

Fingerprint

Dive into the research topics of 'Explicit occlusion detection based deformable fitting for facial landmark localization'. Together they form a unique fingerprint.

Cite this