一种基于融合深度卷积神经网络与度量学习的人脸识别方法A face recognition method based on fusion of deep CNN and metric learning
吕璐,蔡晓东,曾燕,梁晓曦
摘要(Abstract)：
现有的卷积神经网络方法大多以增大类间距离为学习目标,而忽略类内距离的减小,这对于人脸识别来说,将导致一些非限制条件下(如姿态、光照等)的人脸无法被准确识别,为了解决此问题,提出一种基于融合度量学习算法和深度卷积神经网络的人脸识别方法。首先,提出一种基于多Inception结构的人脸特征提取网络,使用较少参数来提取特征;其次,提出一种联合损失的度量学习方法,将分类损失和中心损失进行加权联合;最后,将深度卷积神经网络和度量学习方法进行融合,在网络训练时,达到增大类间距离同时减小类内距离的学习目标。实验结果表明,该方法能提取出更具区分性的人脸特征,与分类损失方法及融合了其他度量学习方式的方法相比,提升了非限制条件下的人脸识别准确率。
关键词(KeyWords)： 多Inception结构;深度卷积神经网络;度量学习方法;深度人脸识别;特征提取;损失函数融合
基金项目(Foundation): 2016年广西科技计划项目(广西重点研发计划)(桂科AB16380264);; 2014年国家科技支撑计划课题(2014BAK11B02)~~
作者(Author): 吕璐,蔡晓东,曾燕,梁晓曦
DOI: 10.16652/j.issn.1004-373x.2018.09.013
参考文献(References)：
- [1]CHEN Dong,CAO Xudong,WEN Fang,et al.Blessing of dimensionality:high-dimensional feature and its efficient compression for face verification[C]//Proceedings of 2013 IEEE Conference on Computer Vision and Pattern Recognition.Portland:IEEE,2013:3025-3032.
- [2]SIMONYAN K,PARKHI O,VEDALDI A,et al.Fisher vector faces in the wild[C]//Proceedings of 2013 British Machine Vision Conference.British:[s.n.],2013:1-11.
- [3]TAIGMAN Y,YANG M,RANZATO M,et al.Deep Face:closing the gap to human-level performance in face verification[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition.Columbus:IEEE,2014:1701-1708.
- [4]SUN Yi,WANG Xiaogang,TANG Xiaoou.Deep learning face representation from predicting 10 000 classes[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition.Columbus:IEEE,2014:1891-1898.
- [5]SUN Yi,WANG Xiaogang,TANG Xiaoou.Deep learning face representation by joint identification-verification[J].Advances in neural information processing systems,2014,27:1988-1996.
- [6]YI Dong,LEI Zhen,LIAO Shengcai,et al.Learning face representation from scratch[J].Computer science,2014(11):1-9.
- [7]SCHROFF F,KALENICHENKO D,PHILBIN J.Face Net:a unified embedding for face recognition and clustering[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition.Boston:IEEE,2015:815-823.
- [8]SZEGEDY C,LIU W,JIA Y,et al.Going deeper with convolutions[C]//2014 IEEE Computer Vision and Pattern Recognition.[S.l.]:IEEE,2014:1-9.
- [9]SZEGEDY C,VANHOUCKE V,IOFFE S,et al.Rethinking the inception architecture for computer vision[C]//2016 IEEE Conference on Computer Science and Pattern Recognition.Las Vegas:IEEE,2015:2818-2826.
- [10]WEN Yandong,ZHANG Kaipeng,LI Zhifeng,et al.A discriminative feature learning approach for deep face recognition[J].Computers&operations research,2016,47(9):11-26.
- [11]JIA Yangqing,SHELHAMER E,DONAHUE J,et al.Caffe:convolutional architecture for fast feature embedding[J].Computer science,2014(3):675-678.
- [12]HUANG G B,MATTAR M,BERG T,et al.Labeled faces in the wild:a database for studying face recognition in unconstrained environments[J/OL].[2008-10-03].http://www.tamaraberg.com/papers/Huang_eccv2008-lfw.pdf.
- [13]ZHANG Kaipeng,ZHANG Zhanpeng,LI Zhifeng,et al.Joint face detection and alignment using multi-task cascaded convolutional networks[J].IEEE signal processing letters,2016,23(10):1499-1503.