Structured and weighted multi-task low rank tracker
作者：
Highlights：
• Propose structured and weighted multi-task low rank tracker with novel task definition.
• Weighted nuclear norm adaptively assigns different tracking importance on different rank components of multiple tasks, and avoids over-shrink.
• Take advantage of the local and global multi-task tracking modals simultaneously, and mine their structure information.
• Simultaneously learn and update the adaptively discriminative subspace and classifier.
• The developed tracker is a general model for most existing multi-task trackers.
摘要
•Propose structured and weighted multi-task low rank tracker with novel task definition.•Weighted nuclear norm adaptively assigns different tracking importance on different rank components of multiple tasks, and avoids over-shrink.•Take advantage of the local and global multi-task tracking modals simultaneously, and mine their structure information.•Simultaneously learn and update the adaptively discriminative subspace and classifier.•The developed tracker is a general model for most existing multi-task trackers.
论文关键词：Robust multi-subtask learning,Structured and weighted low rank,Group-sparsity regularization,Normalized collaboration metric
论文评审过程：Received 16 May 2017, Revised 13 October 2017, Accepted 2 April 2018, Available online 6 April 2018, Version of Record 16 May 2018.
论文官网地址：https://doi.org/10.1016/j.patcog.2018.04.002
