Deep compact discriminative representation for unconstrained face recognition
作者：
Highlights：
• Two losses are for compact and discriminative deep face features.
• The intra-class variations and inter-class variations are both restrained.
• ACD loss alleviates the imbalanced computation of CD loss.
• The methods avoid the inescapable step of selecting training samples.
• Six face recognition tasks are conducted to evaluate the performance.
摘要
•Two losses are for compact and discriminative deep face features.•The intra-class variations and inter-class variations are both restrained.•ACD loss alleviates the imbalanced computation of CD loss.•The methods avoid the inescapable step of selecting training samples.•Six face recognition tasks are conducted to evaluate the performance.
论文关键词：Convolutional neural network,Compact discriminative loss,Advanced compact discriminative loss,Deep compact discriminative representation,Face recognition
论文评审过程：Received 22 March 2018, Revised 21 March 2019, Accepted 22 March 2019, Available online 27 March 2019, Version of Record 10 April 2019.
论文官网地址：https://doi.org/10.1016/j.image.2019.03.015
