Learn by Guessing: Multi-Step Pseudo-Label Refinement for Person Re-Identification

TL;DR

This paper introduces a multi-step pseudo-label refinement technique for unsupervised domain adaptation in person re-identification, improving cluster quality and reducing training iterations, leading to state-of-the-art results.

Contribution

It proposes a novel cluster refinement method with camera-based normalization, enhancing cluster quality without label knowledge in UDA for person Re-ID.

Findings

Achieved 3.4% improvement on Market1501-DukeMTMC datasets.

State-of-the-art UDA results on DukeMTMC-Market1501.

Significant reduction in training iterations due to normalization.

Abstract

Unsupervised Domain Adaptation (UDA) methods for person Re-Identification (Re-ID) rely on target domain samples to model the marginal distribution of the data. To deal with the lack of target domain labels, UDA methods leverage information from labeled source samples and unlabeled target samples. A promising approach relies on the use of unsupervised learning as part of the pipeline, such as clustering methods. The quality of the clusters clearly plays a major role in methods performance, but this point has been overlooked. In this work, we propose a multi-step pseudo-label refinement method to select the best possible clusters and keep improving them so that these clusters become closer to the class divisions without knowledge of the class labels. Our refinement method includes a cluster selection strategy and a camera-based normalization method which reduces the within-domain variations caused by the use of multiple cameras in person Re-ID. This allows our method to reach state-of-the-art UDA results on DukeMTMC-Market1501 (source-target). We surpass state-of-the-art for UDA Re-ID by 3.4% on Market1501-DukeMTMC datasets, which is a more challenging adaptation setup because the target domain (DukeMTMC) has eight distinct cameras. Furthermore, the camera-based normalization method causes a significant reduction in the number of iterations required for training convergence.