How Does Gender Balance In Training Data Affect Face Recognition Accuracy?

被引:32
作者
Albiero, Vitor [1 ]
Zhang, Kai [1 ]
Bowyer, Kevin W. [1 ]
机构
[1] Univ Notre Dame, Notre Dame, IN 46556 USA
来源
IEEE/IAPR INTERNATIONAL JOINT CONFERENCE ON BIOMETRICS (IJCB 2020) | 2020年
关键词
PERFORMANCE;
D O I
10.1109/ijcb48548.2020.9304924
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Deep learning methods have greatly increased the accuracy of face recognition, but an old problem still persists: accuracy is usually higher for men than women. It is often speculated that lower accuracy for women is caused by under-representation in the training data. This work investigates female under-representation in the training data is truly the cause of lower accuracy for females on test data. Using a state-of-the-art deep CNN, three different loss functions, and two training datasets, we train each on seven subsets with different male/female ratios, totaling forty two trainings, that are tested on three different datasets. Results show that (1) gender balance in the training data does not translate into gender balance in the test accuracy, (2) the "gender gap" in test accuracy is not minimized by a gender-balanced training set, but by a training set with more male images than female images, and (3) training to minimize the accuracy gap does not result in highest female, male or average accuracy
引用
收藏
页数:10
相关论文
共 39 条
[1]  
Albiero V., 2020, WINTER C APPL COMPUT
[2]  
Albiero V., 2020, WINT C APPL COMP VIS
[3]   Longitudinal Study of Automatic Face Recognition [J].
Best-Rowden, Lacey ;
Jain, Anil K. .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2018, 40 (01) :148-162
[4]  
Best-Rowden L, 2015, INT CONF BIOMETR, P214, DOI 10.1109/ICB.2015.7139087
[5]   Factors that influence algorithm performance in the Face Recognition Grand Challenge [J].
Beveridge, J. Ross ;
Givens, Geof H. ;
Phillips, P. Jonathon ;
Draper, Bruce A. .
COMPUTER VISION AND IMAGE UNDERSTANDING, 2009, 113 (06) :750-762
[6]  
Boyu Lu, 2019, IEEE Transactions on Biometrics, Behavior, and Identity Science, V1, P42, DOI 10.1109/TBIOM.2018.2890577
[7]   VGGFace2: A dataset for recognising faces across pose and age [J].
Cao, Qiong ;
Shen, Li ;
Xie, Weidi ;
Parkhi, Omkar M. ;
Zisserman, Andrew .
PROCEEDINGS 2018 13TH IEEE INTERNATIONAL CONFERENCE ON AUTOMATIC FACE & GESTURE RECOGNITION (FG 2018), 2018, :67-74
[8]  
Chen BC, 2014, LECT NOTES COMPUT SC, V8694, P768, DOI 10.1007/978-3-319-10599-4_49
[9]   Exploiting Effective Facial Patches for Robust Gender Recognition [J].
Cheng, Jingchun ;
Li, Yali ;
Wang, Jilong ;
Yu, Le ;
Wang, Shengjin .
TSINGHUA SCIENCE AND TECHNOLOGY, 2019, 24 (03) :333-345
[10]  
Cook Cynthia M., 2019, IEEE Transactions on Biometrics, Behavior, and Identity Science, V1, P32, DOI 10.1109/TBIOM.2019.2897801