1. Two-Branch Asymmetric Model With Alternately Clustering for Unsupervised Person Re-Identification
- Author
-
Yangbin Yu, Haifeng Hu, Dihu Chen, and Ying Zeng
- Subjects
business.industry ,Computer science ,Applied Mathematics ,Pattern recognition ,Field (computer science) ,Re identification ,Cluster algorithm ,Constraint (information theory) ,ComputingMethodologies_PATTERNRECOGNITION ,Signal Processing ,Cluster (physics) ,Artificial intelligence ,Electrical and Electronic Engineering ,Cluster analysis ,business - Abstract
In the field of unsupervised person re-identification (Re-ID), mainstream methods adopt cluster algorithm to generate pseudo labels for training. Despite the effectiveness, the cluster algorithm generates noisy labels, which are retained in further model updating and hinder higher performance. To solve this problem, we propose a Two-branch Asymmetric Model with Alternately Clustering. Specifically, the designed Alternately Clustering (AC) strategy leverages a two-branch model to cluster different pseudo labels for each branch, which prevents the continuous existence of identical noisy labels. To establish a mapping between the two label sets, pseudo label mapping constraint (PLMC) module is devised, which helps retain reliable pseudo labels. Our method improves the quality of generated pseudo labels by keeping noisy labels changing and retaining the reliable ones. Experimental results demonstrate that our proposed method outperforms the state-of-the-art methods.
- Published
- 2022