Robust discriminative tracking via structured prior regularization

作者：

Highlights：

• Video tracking is cast into a prior regularized semi-supervised learning task.

• A generative model is utilized support the discriminative tracker.

• Target structure is maintained by a set of structured distributions.

• The reliability of features used in track update is quantified.

摘要

•Video tracking is cast into a prior regularized semi-supervised learning task.•A generative model is utilized support the discriminative tracker.•Target structure is maintained by a set of structured distributions.•The reliability of features used in track update is quantified.

论文关键词：Visual tracking,Semi-supervised learning,Multi-objective optimization,Random Forest

论文评审过程：Received 29 January 2016, Revised 10 July 2017, Accepted 10 November 2017, Available online 21 November 2017, Version of Record 5 December 2017.

论文官网地址：https://doi.org/10.1016/j.imavis.2017.11.003