Thermal to Visual Person Re-Identification Using Collaborative Metric Learning Based on Maximum Margin Matrix Factorization

作者：

Highlights：

• We address the problem of thermal to visual person re-identification.

• We propose collaborative metric learning method to address the problem.

• The proposed collaborative metric learning uses maximum margin matrix factorization.

• We obtain the more generalised metric due to the maximum margin.

• Proposed method outperforms the baseline algorithms in few-shot learning settings.

摘要

•We address the problem of thermal to visual person re-identification.•We propose collaborative metric learning method to address the problem.•The proposed collaborative metric learning uses maximum margin matrix factorization.•We obtain the more generalised metric due to the maximum margin.•Proposed method outperforms the baseline algorithms in few-shot learning settings.

论文关键词：Thermal to visual person re-identification,Cross-domain image retrieval,Collaborative metric learning,Matrix factorization

论文评审过程：Received 24 January 2022, Revised 6 August 2022, Accepted 21 September 2022, Available online 24 September 2022, Version of Record 1 October 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.109069