Keep it Accurate and Diverse: Enhancing Action Recognition Performance by Ensemble Learning

Research output: Research - peer-reviewArticle in proceeding

Abstract

The performance of different action recognition tech- niques has recently been studied by several computer vision researchers. However, the potential improvement in classi- fication through classifier fusion by ensemble-based meth- ods has remained unattended. In this work, we evaluate the performance of an ensemble of action learning techniques, each performing the recognition task from a different per- spective. The underlying idea is that instead of aiming a very sophisticated and powerful representation/learning technique, we can learn action categories using a set of relatively simple and diverse classifiers, each trained with different feature set. In addition, combining the outputs of several learners can reduce the risk of an unfortunate selec- tion of a learner on an unseen action recognition scenario. This leads to having a more robust and general-applicable framework. In order to improve the recognition perfor- mance, a powerful combination strategy is utilized based on the Dempster-Shafer theory, which can effectively make use of diversity of base learners trained on different sources of information. The recognition results of the individual clas- sifiers are compared with those obtained from fusing the classifiers’ output, showing enhanced performance of the proposed methodology.
Close

Details

The performance of different action recognition tech- niques has recently been studied by several computer vision researchers. However, the potential improvement in classi- fication through classifier fusion by ensemble-based meth- ods has remained unattended. In this work, we evaluate the performance of an ensemble of action learning techniques, each performing the recognition task from a different per- spective. The underlying idea is that instead of aiming a very sophisticated and powerful representation/learning technique, we can learn action categories using a set of relatively simple and diverse classifiers, each trained with different feature set. In addition, combining the outputs of several learners can reduce the risk of an unfortunate selec- tion of a learner on an unseen action recognition scenario. This leads to having a more robust and general-applicable framework. In order to improve the recognition perfor- mance, a powerful combination strategy is utilized based on the Dempster-Shafer theory, which can effectively make use of diversity of base learners trained on different sources of information. The recognition results of the individual clas- sifiers are compared with those obtained from fusing the classifiers’ output, showing enhanced performance of the proposed methodology.
Original languageEnglish
Title of host publicationIEEE Conference on Computer Vision and Pattern Recognition Workshop (CVPRW), 2015
PublisherIEEE Computer Society Press
Publication date7 Jun 2015
Pages22-29
ISBN (Print)978-1-4673-6759-2
DOI
StatePublished - 7 Jun 2015
Publication categoryResearch
Peer-reviewedYes
Event2015 IEEE Conference on Computer Vision and Pattern Recognition Workshop - Boston, United States
Duration: 6 Jun 201512 Jun 2015

Conference

Conference2015 IEEE Conference on Computer Vision and Pattern Recognition Workshop
LandUnited States
ByBoston
Periode06/06/201512/06/2015

Download statistics

No data available
ID: 210011557