Hybrid Contrastive Learning with Cluster Ensemble for Unsupervised Person Re-Identification

EasyChair Preprint 7081, version 1

Abstract

Unsupervised person re-identification (ReID) aims to match a query image of a pedestrian to the images in gallery set without supervision labels. The most popular approaches to tackle unsupervised person ReID are usually performing a clustering algorithm to yield pseudo labels at first and then exploit the pseudo labels to train a deep neural network. However, the pseudo labels are noisy and sensitive to selected hyper-parameter(s) in the used clustering algorithm. In this paper, we propose a Hybrid Contrastive Learning (HCL) approach for unsupervised person ReID, which is based on a hybrid between instance-level and cluster-level contrastive losses. Moreover, we present a multi-granularity clustering ensemble based hybrid contrastive learning (MGCE-HCL) approach, which adopts a multi-granularity clustering ensemble strategy to mine priority information among the pseudo positive sample pairs and defines a priority weighted hybrid contrastive loss for better tolerating the noises in the pseudo positive samples. We conduct extensive experiments on two benchmark datasets Market-1501 and DukeMTMC-reID and experimental results validate the effectiveness of our proposals.

Keyphrases: Cluster Ensemble, Contrastive Learning, Unsupervised Person ReID, multi-granularity

@booklet{EasyChair:7081,
  author    = {He Sun and Mingkun Li and Chun-Guang Li},
  title     = {Hybrid Contrastive Learning with Cluster Ensemble for Unsupervised Person Re-Identification},
  howpublished = {EasyChair Preprint 7081},
  year      = {EasyChair, 2021}}