Facial expression recognition has been addressed mainly working on 2D images or videos. In this paper, the problem of person-independent facial expression recognition is addressed on 3D shapes. To this end, an original approach is proposed that relies on selecting the minimal-redundancy maximal-relevance features derived from a pool of SIFT feature descriptors computed in correspondence with facial landmarks of depth images. Training a Support Vector Machine for every basic facial expression to be recognized, and combining them to form a multiclass classifier, an average recognition rate of 77.5% on the BU-3DFE database has been obtained. Comparison with competitors approaches using a common experimental setting on the BU-3DFE database, shows that our solution is able to obtain state of the art results.
Person Independent 3D Facial Expression Recognition by a Selected Ensemble of SIFT Descriptors / S. Berretti; B. Ben Amor; M. Daoudi; A. Del Bimbo. - STAMPA. - (2010), pp. 47-54. (Intervento presentato al convegno 3rd Eurographics/ACM SIGGRAPH Symposyum on 3D Object Retrieval tenutosi a Norrkooping, Svezia nel May 2) [10.2312/3DOR/3DOR10/047-054].
Person Independent 3D Facial Expression Recognition by a Selected Ensemble of SIFT Descriptors
BERRETTI, STEFANO;DEL BIMBO, ALBERTO
2010
Abstract
Facial expression recognition has been addressed mainly working on 2D images or videos. In this paper, the problem of person-independent facial expression recognition is addressed on 3D shapes. To this end, an original approach is proposed that relies on selecting the minimal-redundancy maximal-relevance features derived from a pool of SIFT feature descriptors computed in correspondence with facial landmarks of depth images. Training a Support Vector Machine for every basic facial expression to be recognized, and combining them to form a multiclass classifier, an average recognition rate of 77.5% on the BU-3DFE database has been obtained. Comparison with competitors approaches using a common experimental setting on the BU-3DFE database, shows that our solution is able to obtain state of the art results.File | Dimensione | Formato | |
---|---|---|---|
3dor10.pdf
Accesso chiuso
Descrizione: Articolo principale
Tipologia:
Versione finale referata (Postprint, Accepted manuscript)
Licenza:
Tutti i diritti riservati
Dimensione
2.52 MB
Formato
Adobe PDF
|
2.52 MB | Adobe PDF | Richiedi una copia |
I documenti in FLORE sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.