Deep features for person re-identification on metric learning
作者：
Highlights：
• We reviewed several person Re ID approaches. In doing so, we proposed taxonomies for these two components.
• For our taxonomies, combining advanced approaches to data enhancement, we conduct comprehensive experiments on metric learning methods.
• Our results show quantitatively that, the powerful expression of deep features reduces the influence of metric learning on overall performance.
摘要
•We reviewed several person Re ID approaches. In doing so, we proposed taxonomies for these two components.•For our taxonomies, combining advanced approaches to data enhancement, we conduct comprehensive experiments on metric learning methods.•Our results show quantitatively that, the powerful expression of deep features reduces the influence of metric learning on overall performance.
论文关键词：Person re-identification,Deep features,Metric learning,Empirical comparison
论文评审过程：Received 15 July 2019, Revised 24 March 2020, Accepted 4 May 2020, Available online 8 May 2020, Version of Record 1 November 2020.
论文官网地址：https://doi.org/10.1016/j.patcog.2020.107424
