Towards Generalizable Person Re-identification with a Bi-stream Generative Model

TL;DR

This paper introduces a bi-stream generative model that enhances person re-identification by learning camera-invariant global features and pedestrian-aligned local features, significantly improving generalization across unseen domains.

Contribution

The proposed Bi-stream Generative Model uniquely combines global and local feature learning with a part-weighted loss to address camera and pedestrian misalignment challenges in re-ID.

Findings

Outperforms state-of-the-art on large-scale generalizable re-ID benchmarks.

Effectively handles cross-camera appearance variations.

Improves pedestrian alignment through semantic part maps.

Abstract

Generalizable person re-identification (re-ID) has attracted growing attention due to its powerful adaptation capability in the unseen data domain. However, existing solutions often neglect either crossing cameras (e.g., illumination and resolution differences) or pedestrian misalignments (e.g., viewpoint and pose discrepancies), which easily leads to poor generalization capability when adapted to the new domain. In this paper, we formulate these difficulties as: 1) Camera-Camera (CC) problem, which denotes the various human appearance changes caused by different cameras; 2) Camera-Person (CP) problem, which indicates the pedestrian misalignments caused by the same identity person under different camera viewpoints or changing pose. To solve the above issues, we propose a Bi-stream Generative Model (BGM) to learn the fine-grained representations fused with camera-invariant global feature and pedestrian-aligned local feature, which contains an encoding network and two stream decoding sub-networks. Guided by original pedestrian images, one stream is employed to learn a camera-invariant global feature for the CC problem via filtering cross-camera interference factors. For the CP problem, another stream learns a pedestrian-aligned local feature for pedestrian alignment using information-complete densely semantically aligned part maps. Moreover, a part-weighted loss function is presented to reduce the influence of missing parts on pedestrian alignment. Extensive experiments demonstrate that our method outperforms the state-of-the-art methods on the large-scale generalizable re-ID benchmarks, involving domain generalization setting and cross-domain setting.