View-specific subspace learning and re-ranking for semi-supervised person re-identification

作者：

Highlights：

• A pragmatic and effective framework for semi-supervised person re-ID is introduced.

• A view-specific subspace learning method is proposed to tackle view-specific biases.

• An effective re-ranking strategy with expanded cross neighborhood is proposed.

• A novel reciprocal content similarity and contextual similarity is put forward.

摘要

•A pragmatic and effective framework for semi-supervised person re-ID is introduced.•A view-specific subspace learning method is proposed to tackle view-specific biases.•An effective re-ranking strategy with expanded cross neighborhood is proposed.•A novel reciprocal content similarity and contextual similarity is put forward.

论文关键词：Person re-identification,Metric learning,Semi-supervised learning,Re-ranking

论文评审过程：Received 15 October 2019, Revised 4 June 2020, Accepted 28 July 2020, Available online 30 July 2020, Version of Record 4 August 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107568