Logistic Similarity Metric Learning for Face Verification

Lilei Zheng 1 Khalid Idrissi 1 Christophe Garcia 1 Stefan Duffner 1 Atilla Baskurt 1
1 imagine - Extraction de Caractéristiques et Identification
LIRIS - Laboratoire d'InfoRmatique en Image et Systèmes d'information
Abstract : This paper presents a new method for similarity metric learning , called Logistic Similarity Metric Learning (LSML), where the cost is formulated as the logistic loss function, which gives a probability estimation of a pair of faces being similar. Especially, we propose to shift the similarity decision boundary gaining significant performance improvement. We test the proposed method on the face verification problem using four single face descriptors: LBP, OCLBP, SIFT and Gabor wavelets. Extensive experimental results on the LFW-a data set demonstrate that the proposed method achieves competitive state-of-the-art performance on the problem of face verification.
Keywords : face verification Metric learning linear transformation face recognition cosine similarity
Type de document :
Communication dans un congrès
40th IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2015, Apr 2015, Brisbane, Australia
Liste complète des métadonnées

Littérature citée [26 références]  Voir  Masquer  Télécharger
https://hal.archives-ouvertes.fr/hal-01158949
Contributeur : Lilei Zheng <>
Soumis le : mardi 2 juin 2015 - 11:16:26
Dernière modification le : jeudi 19 avril 2018 - 14:38:05
Document(s) archivé(s) le : mardi 15 septembre 2015 - 09:26:32

Fichier

Liris-7045.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01158949, version 1

Collections

EC-LYON | LIRIS | UNIV-LYON2 | INSA-LYON | UNIV-LYON1

Citation

Lilei Zheng, Khalid Idrissi, Christophe Garcia, Stefan Duffner, Atilla Baskurt. Logistic Similarity Metric Learning for Face Verification. 40th IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2015, Apr 2015, Brisbane, Australia. 〈hal-01158949〉

Exporter

BibTeX TEI DC DCterms EndNote

Partager

Métriques

Consultations de la notice

294

Téléchargements de fichiers

607

Contact

support.ccsd.cnrs.fr
hal.support@ccsd.cnrs.fr
Logo CCSD