Delving into Probabilistic Uncertainty for Unsupervised Domain Adaptive Person Re-Identification

TL;DR

This paper introduces P²LR, a probabilistic uncertainty-guided method for unsupervised domain adaptive person re-identification, which models label uncertainty and progressively refines pseudo labels to improve accuracy.

Contribution

The paper proposes a novel probabilistic uncertainty modeling and a progressive pseudo label refinement strategy for UDA person ReID, achieving state-of-the-art results.

Findings

Significant performance improvements over baseline methods.

Achieved 6.5% mAP increase on Duke2Market benchmark.

Surpassed existing methods by 2.5% mAP on Market2MSMT.

Abstract

Clustering-based unsupervised domain adaptive (UDA) person re-identification (ReID) reduces exhaustive annotations. However, owing to unsatisfactory feature embedding and imperfect clustering, pseudo labels for target domain data inherently contain an unknown proportion of wrong ones, which would mislead feature learning. In this paper, we propose an approach named probabilistic uncertainty guided progressive label refinery (P$^2$LR) for domain adaptive person re-identification. First, we propose to model the labeling uncertainty with the probabilistic distance along with ideal single-peak distributions. A quantitative criterion is established to measure the uncertainty of pseudo labels and facilitate the network training. Second, we explore a progressive strategy for refining pseudo labels. With the uncertainty-guided alternative optimization, we balance between the exploration of target domain data and the negative effects of noisy labeling. On top of a strong baseline, we obtain significant improvements and achieve the state-of-the-art performance on four UDA ReID benchmarks. Specifically, our method outperforms the baseline by 6.5% mAP on the Duke2Market task, while surpassing the state-of-the-art method by 2.5% mAP on the Market2MSMT task.