FedBPrompt: Federated Domain Generalization Person Re-Identification via Body Distribution Aware Visual Prompts

TL;DR

FedBPrompt introduces a novel federated learning approach for person re-identification that uses visual prompts to focus Transformer attention on pedestrian regions, improving cross-domain generalization with reduced communication costs.

Contribution

The paper proposes FedBPrompt, which employs body distribution aware visual prompts and a prompt-based fine-tuning strategy to enhance federated person re-identification performance and efficiency.

Findings

Improved cross-domain generalization in FedDG-ReID tasks.

Significant reduction in communication overhead.

Enhanced feature discrimination with visual prompts.

Abstract

Federated Domain Generalization for Person Re-Identification (FedDG-ReID) learns domain-invariant representations from decentralized data. While Vision Transformer (ViT) is widely adopted, its global attention often fails to distinguish pedestrians from high similarity backgrounds or diverse viewpoints -- a challenge amplified by cross-client distribution shifts in FedDG-ReID. To address this, we propose Federated Body Distribution Aware Visual Prompt (FedBPrompt), introducing learnable visual prompts to guide Transformer attention toward pedestrian-centric regions. FedBPrompt employs a Body Distribution Aware Visual Prompts Mechanism (BAPM) comprising: Holistic Full Body Prompts to suppress cross-client background noise, and Body Part Alignment Prompts to capture fine-grained details robust to pose and viewpoint variations. To mitigate high communication costs, we design a Prompt-based Fine-Tuning Strategy (PFTS) that freezes the ViT backbone and updates only lightweight prompts, significantly reducing communication overhead while maintaining adaptability. Extensive experiments demonstrate that BAPM effectively enhances feature discrimination and cross-domain generalization, while PFTS achieves notable performance gains within only a few aggregation rounds. Moreover, both BAPM and PFTS can be easily integrated into existing ViT-based FedDG-ReID frameworks, making FedBPrompt a flexible and effective solution for federated person re-identification. The code is available at https://github.com/leavlong/FedBPrompt.