Ensemble Approaches to Facial Action Unit Classification

Windeatt, T and Dias, K (2008) Ensemble Approaches to Facial Action Unit Classification In: 13th Iberoamerican Congress on Progress in Pattern Recognition, Image Analysis and Applications, 2008-09-09 - 2008-09-12, Havana, CUBA.

Preview	Text windeatt_cuba08.pdf Available under License : See the attached licence file. Download (58kB)
Preview	Text (licence) SRI_deposit_agreement.pdf Download (33kB)

Abstract

Facial action unit (au) classification is an approach to face expression recognition that decouples the recognition of expression from individual actions. In this paper, upper face aus are classified using an ensemble of MLP (Multi-layer perceptron) base classifiers with feature ranking based on PCA components. This approach is compared experimentally with other popular feature-ranking methods applied to Gabor features. Experimental results on Cohn-Kanade database demonstrate that the MLP ensemble is relatively insensitive to the feature-ranking method but optimized PCA features achieve lowest error rate. When posed as a multi-class problem using Error- Correcting-Output-Coding (ECOC), error rates are comparable to two-class problems (one-versus-rest) when the number of features and base classifier are optimized.

Item Type:	Conference or Workshop Item (Conference Paper)
Divisions :	Faculty of Engineering and Physical Sciences > Electronic Engineering > Centre for Vision Speech and Signal Processing
Authors :	Name Email ORCID Windeatt, T Dias, K
Date :	2008
Contributors :	Contribution Name Email ORCID http://www.loc.gov/loc.terms/relators/EDT RuizShulcloper, J http://www.loc.gov/loc.terms/relators/EDT Kropatsch, WG http://www.loc.gov/loc.terms/relators/PBL SPRINGER-VERLAG BERLIN,
Uncontrolled Keywords :	Ensembles, ECOC, FACS, Feature-ranking, FEATURE-SELECTION, FEATURE SUBSETS
Additional Information :	The original publication is available at http://www.springerlink.com
Depositing User :	Symplectic Elements
Date Deposited :	17 Feb 2012 11:59
Last Modified :	31 Oct 2017 14:20
URI:	http://epubs.surrey.ac.uk/id/eprint/135501

Actions (login required)

View Item

Downloads