Unsupervised Video Person Re-identification via Noise and Hard frame Aware Clustering

TL;DR

This paper introduces NHAC, a clustering method for unsupervised video person re-ID that effectively handles noise and hard frames in tracklets, improving clustering accuracy and model training.

Contribution

It proposes a novel Noise and Hard frame Aware Clustering (NHAC) method with graph trimming and node re-sampling modules for better unsupervised video re-ID.

Findings

NHAC outperforms existing methods on two datasets.

Graph trimming reduces noise influence in clustering.

Node re-sampling enhances hard frame learning.

Abstract

Unsupervised video-based person re-identification (re-ID) methods extract richer features from video tracklets than image-based ones. The state-of-the-art methods utilize clustering to obtain pseudo-labels and train the models iteratively. However, they underestimate the influence of two kinds of frames in the tracklet: 1) noise frames caused by detection errors or heavy occlusions exist in the tracklet, which may be allocated with unreliable labels during clustering; 2) the tracklet also contains hard frames caused by pose changes or partial occlusions, which are difficult to distinguish but informative. This paper proposes a Noise and Hard frame Aware Clustering (NHAC) method. NHAC consists of a graph trimming module and a node re-sampling module. The graph trimming module obtains stable graphs by removing noise frame nodes to improve the clustering accuracy. The node re-sampling module enhances the training of hard frame nodes to learn rich tracklet information. Experiments conducted on two video-based datasets demonstrate the effectiveness of the proposed NHAC under the unsupervised re-ID setting.