Deep learning-based person re-identification methods: A survey and outlook of recent works
作者：
Highlights：
• The main contributions of person Re-ID surveys are summarized and discussed in recent years.
• A metric and representation learning-based taxonomy is provided for recent person Re-ID methods.
• The above main categories are subdivided based on their methodologies and motivations.
• The advantages and limitations of part subcategories are summarized.
• Furthermore, some challenges and possible research directions for person Re-ID are discussed.
摘要
•The main contributions of person Re-ID surveys are summarized and discussed in recent years.•A metric and representation learning-based taxonomy is provided for recent person Re-ID methods.•The above main categories are subdivided based on their methodologies and motivations.•The advantages and limitations of part subcategories are summarized.•Furthermore, some challenges and possible research directions for person Re-ID are discussed.
论文关键词：Person re-identification,Deep metric learning,Local feature learning,Generative adversarial learning,Sequence feature learning
论文评审过程：Received 26 October 2021, Revised 12 January 2022, Accepted 20 January 2022, Available online 25 January 2022, Version of Record 3 February 2022.
论文官网地址：https://doi.org/10.1016/j.imavis.2022.104394
