Learning global and local features using graph neural networks for person re-identification

作者：

Highlights：

• A model that exploits both global and local features.

• GNNs that model the relations of patches and represent pairwise relationships.

• A scoring function that improves alignment and learns discriminative features.

• A model that performs favorably well against state-of-the-art methods.

摘要

•A model that exploits both global and local features.•GNNs that model the relations of patches and represent pairwise relationships.•A scoring function that improves alignment and learns discriminative features.•A model that performs favorably well against state-of-the-art methods.

论文关键词：Person re-identification,Body-part,Alignment,Graph neural networks

论文评审过程：Received 21 February 2021, Revised 5 December 2021, Accepted 25 May 2022, Available online 9 June 2022, Version of Record 17 June 2022.

论文官网地址：https://doi.org/10.1016/j.image.2022.116744