Manifold-based aggregation clustering for unsupervised vehicle re-identification

作者：

Highlights：

• A Manifold-based Aggregation Clustering framework is proposed for unsupervised V-reID without any annotations.

• An agglomeration-classification loss is proposed to learn aggregated features.

• A manifold-based distance and cluster diversity are formulated to determine the seeded vehicle criterion.

摘要

•A Manifold-based Aggregation Clustering framework is proposed for unsupervised V-reID without any annotations.•An agglomeration-classification loss is proposed to learn aggregated features.•A manifold-based distance and cluster diversity are formulated to determine the seeded vehicle criterion.

论文关键词：Vehicle re-identification,Aggregation clustering,Manifold distance

论文评审过程：Received 24 April 2021, Revised 17 October 2021, Accepted 19 October 2021, Available online 25 October 2021, Version of Record 29 October 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.107624