期刊
IEEE TRANSACTIONS ON IMAGE PROCESSING
卷 27, 期 1, 页码 293-303出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TIP.2017.2756450
关键词
Face recognition; deep learning; small training data
资金
- European Unions Horizon Research and Innovation Program [640891]
- Science and Technology Plan Project of Hunan Province [2016TP1020]
- Natural Science Foundation of China [61502152]
- French research agency [ANR-16-CE23-0006, ANR-11-LABX-0025-01]
- Agence Nationale de la Recherche (ANR) [ANR-16-CE23-0006] Funding Source: Agence Nationale de la Recherche (ANR)
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 large training data sets are not publicly available and difficult to collect. In this paper, we propose a method to generate very large training data sets of synthetic images by compositing real face images in a given data set. 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 obtain state-of-the-art results on the CASIA NIR-VIS2.0 heterogeneous face recognition data set.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据