3C: Confidence-guided clustering and contrastive learning for unsupervised person re-identification

Unsupervised person re-identification (Re-ID) aims to learn a feature network with cross-camera retrieval capability in unlabelled datasets. Although pseudo-label based methods have achieved great progress in Re-ID, their effectiveness in complex scenarios is hindered by three unresolved challenges: (i) Noisy pseudo-labels due to disturbed clustering. (ii) Camera-biased clusters that lack diversity. (iii) Unreliable hard samples that amplify noise in contrastive learning. To address these issues, a Confidence-guided Clustering and Contrastive learning (3C) framework is proposed in this paper. The 3C framework presents three confidence degrees: (i) in the clustering stage, the confidence of the discrepancy between samples and clusters is proposed to implement a harmonic discrepancy clustering algorithm (HDC); (ii) in the forward-propagation training stage, the confidence of the camera diversity of a cluster is evaluated using a novel camera information entropy (CIE), and clusters with high CIE will play the leading role in model training; and (iii) in the back-propagation training stage, the confidence of the hard sample in each cluster is designed and further used in a confidence integrated harmonic discrepancy (CHD), to select the informative sample for updating the memory for contrastive learning. Extensive experiments on three popular Re-ID benchmarks demonstrated the superiority of the proposed framework. In particular, the 3C framework achieved state-of-the-art results: 86.7%/94.7%, 45.3%/73.1% and 47.1%/90.6% in term of mAP/Rank-1 accuracy on Market-1501, MSMT17 and VeRi-776, respectively.