Frankenstein: Learning Deep Face Representations using Small Data

Abstract : Deep convolutional neural networks have recently proven extremely effective for difficult face recognition problems in uncontrolled settings. To train such networks very large training sets are needed with millions of labeled images. For some applications, such as near-infrared (NIR) face recognition, such larger training datasets are, however, not publicly available and very difficult to collect. We propose a method to generate very large training datasets of synthetic images by compositing real face images in a given dataset. We show that this method enables to learn models from as few as 10,000 training images, which perform on par with models trained from 500,000 images. Using our approach we also improve the state-of-the-art results on the CASIA NIR-VIS heterogeneous face recognition dataset.
Type de document :
Pré-publication, Document de travail
2016
Liste complète des métadonnées


https://hal.inria.fr/hal-01306168
Contributeur : Guosheng Hu <>
Soumis le : vendredi 22 avril 2016 - 14:51:44
Dernière modification le : mercredi 13 juillet 2016 - 19:36:10

Fichier

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

Identifiants

  • HAL Id : hal-01306168, version 1
  • ARXIV : 1603.06470

Collections

Citation

Guosheng Hu, Xiaojiang Peng, Yongxin Yang, Timothy Hospedales, Jakob Verbeek. Frankenstein: Learning Deep Face Representations using Small Data. 2016. <hal-01306168>

Partager

Métriques

Consultations de
la notice

527

Téléchargements du document

680