Deep mutual learning for visual object tracking

作者：

Highlights：

• A novel offline training methodology is proposed and integrated into detection-based correlation-filter-based trackers.

• A detailed analysis of the effects of the proposed training methodology for the tracking performance is conducted.

• Extensive experiments on popular tracking benchmarks show the effectiveness of the proposed training methodology.

摘要

•A novel offline training methodology is proposed and integrated into detection-based correlation-filter-based trackers.•A detailed analysis of the effects of the proposed training methodology for the tracking performance is conducted.•Extensive experiments on popular tracking benchmarks show the effectiveness of the proposed training methodology.

论文关键词：Visual object tracking,Deep learning,Mutual learning

论文评审过程：Received 18 June 2020, Revised 21 November 2020, Accepted 14 December 2020, Available online 24 December 2020, Version of Record 5 January 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107796