Comparison of two methods for unsupervised person identification in TV shows

Abstract : —We address the task of identifying people appearing in TV shows. The target persons are all people whose identity is said or written, like the journalists and the well known people, as politicians, athletes, celebrities, etc. In our approach, overlaid names displayed on the images are used to identify the persons without any use of biometric models for the speakers and the faces. Two identification methods are evaluated as part of the REPERE French evaluation campaign. The first one relies on co-occurrence times between overlay person names and speaker/face clusters, and rule-based decisions which assign a name to each monomodal cluster. The second method uses a Conditionnal Random Field (CRF) which combine different types of co-occurrence statistics and pair-wised constraints to jointly identify speakers and faces.
Type de document :
Communication dans un congrès
12th International Workshop on Content-Based Multimedia Indexing, 2014, Klagenfurt, Austria. pp.1 - 6, 2014, Content-Based Multimedia Indexing (CBMI), 2014 12th International Workshop on. 〈10.1109/CBMI.2014.6849828〉
Liste complète des métadonnées

Littérature citée [29 références]  Voir  Masquer  Télécharger

https://hal.archives-ouvertes.fr/hal-01433260
Contributeur : Sylvain Meignier <>
Soumis le : samedi 1 avril 2017 - 00:25:00
Dernière modification le : jeudi 7 février 2019 - 17:55:48
Document(s) archivé(s) le : dimanche 2 juillet 2017 - 12:20:59

Fichier

Gay_CBMI_2014.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

Collections

Citation

Paul Gay, Grégor Dupuy, Carole Lailler, Jean-Marc Odobez, Sylvain Meignier, et al.. Comparison of two methods for unsupervised person identification in TV shows. 12th International Workshop on Content-Based Multimedia Indexing, 2014, Klagenfurt, Austria. pp.1 - 6, 2014, Content-Based Multimedia Indexing (CBMI), 2014 12th International Workshop on. 〈10.1109/CBMI.2014.6849828〉. 〈hal-01433260〉

Partager

Métriques

Consultations de la notice

192

Téléchargements de fichiers

100