Visual tracking based on Distribution Fields and online weighted multiple instance learning

作者：

Highlights：

• We adopt Distribution Field (DF) layer as feature instead of traditional Haar-like one to robustly model the target.

• We derive an online weighted-geometric-mean MIL classifier to select the most discriminative layers.

• Our tracker is more robust while needing fewer features than the traditional Haar-like one and the original DFs one.

• The experiments show higher performance of our tracker than five state-of-the-art ones.

摘要

•We adopt Distribution Field (DF) layer as feature instead of traditional Haar-like one to robustly model the target.•We derive an online weighted-geometric-mean MIL classifier to select the most discriminative layers.•Our tracker is more robust while needing fewer features than the traditional Haar-like one and the original DFs one.•The experiments show higher performance of our tracker than five state-of-the-art ones.

论文关键词：Distribution Fields,Weighted-geometric-mean multiple instance learning,Discriminative classifier,Object tracking

论文评审过程：Received 14 January 2013, Revised 24 July 2013, Accepted 23 September 2013, Available online 1 October 2013.

论文官网地址：https://doi.org/10.1016/j.imavis.2013.09.003