Adaptive appearance model tracking for still-to-video face recognition
作者：
Highlights：
• Propose a face recognition system using Adaptive Appearance Model Tracking (AAMT).
• AAMT learns face models online and matches these models against reference stills.
• AAMT incorporates greater diversity of facial representation in the models.
• AAMT selects facial captures more reliably to update an individual׳s face model.
• Provides a higher level of performance in still-to-video face recognition.
摘要
Highlights•Propose a face recognition system using Adaptive Appearance Model Tracking (AAMT).•AAMT learns face models online and matches these models against reference stills.•AAMT incorporates greater diversity of facial representation in the models.•AAMT selects facial captures more reliably to update an individual׳s face model.•Provides a higher level of performance in still-to-video face recognition.
论文关键词：Biometrics,Video surveillance,Face recognition,Watch-list screening,Single sample per person,Face tracking,Online and incremental learning,Adaptive appearance modeling
论文评审过程：Received 20 December 2014, Revised 23 June 2015, Accepted 5 August 2015, Available online 15 August 2015, Version of Record 28 September 2015.
论文官网地址：https://doi.org/10.1016/j.patcog.2015.08.002
