Papers › Unleashing the Potential of Pre-Trained Diffusion Models for Generalizable Person...

Unleashing the Potential of Pre-Trained Diffusion Models for Generalizable Person Re-Identification

10 Feb 2025arXiv:2502.06619archive 2025-07-28

Jiachen Li, Xiaojin Gong

Domain-generalizable re-identification (DG Re-ID) aims to train a model on one or more source domains and evaluate its performance on unseen target domains, a task that has attracted growing attention due to its practical relevance. While numerous methods have been proposed, most rely on discriminative or contrastive learning frameworks to learn generalizable feature representations. However, these approaches often fail to mitigate shortcut learning, leading to suboptimal performance. In this work, we propose a novel method called diffusion model-assisted representation learning with a correlation-aware conditioning scheme (DCAC) to enhance DG Re-ID. Our method integrates a discriminative and contrastive Re-ID model with a pre-trained diffusion model through a correlation-aware conditioning scheme. By incorporating ID classification probabilities generated from the Re-ID model with a set of learnable ID-wise prompts, the conditioning scheme injects dark knowledge that captures ID correlations to guide the diffusion process. Simultaneously, feedback from the diffusion model is back-propagated through the conditioning scheme to the Re-ID model, effectively improving the generalization capability of Re-ID features. Extensive experiments on both single-source and multi-source DG Re-ID tasks demonstrate that our method achieves state-of-the-art performance. Comprehensive ablation studies further validate the effectiveness of the proposed approach, providing insights into its robustness. Codes will be available at https://github.com/RikoLi/DCAC.

PaperPDFCode

Code

RikoLi/DCAC officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Contrastive LearningGeneralizable Person Re-identificationPerson Re-IdentificationRepresentation Learning

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Generalizable Person Re-identification CUHK03-NP (detected) DCAC MSMT17->Rank-1 34.4 #4 of 5 Archive leaderboard report
Generalizable Person Re-identification CUHK03-NP (detected) DCAC MSMT17->mAP 34.1 #4 of 5 Archive leaderboard report
Generalizable Person Re-identification CUHK03-NP (detected) DCAC Market-1501->Rank-1 33.2 #4 of 5 Archive leaderboard report
Generalizable Person Re-identification CUHK03-NP (detected) DCAC Market-1501->mAP 32.5 #4 of 5 Archive leaderboard report
Generalizable Person Re-identification DukeMTMC-reID DCAC MSMT17->Rank1 75.0 #1 of 4 Archive leaderboard report
Generalizable Person Re-identification DukeMTMC-reID DCAC MSMT17->mAP 58.4 #1 of 4 Archive leaderboard report
Generalizable Person Re-identification DukeMTMC-reID DCAC Market-1501->Rank1 69.1 #1 of 4 Archive leaderboard report
Generalizable Person Re-identification DukeMTMC-reID DCAC Market-1501->mAP 49.5 #1 of 4 Archive leaderboard report
Generalizable Person Re-identification MSMT17 DCAC Market-1501->Rank1 52.1 #3 of 4 Archive leaderboard report
Generalizable Person Re-identification MSMT17 DCAC Market-1501->mAP 23.4 #3 of 4 Archive leaderboard report
Generalizable Person Re-identification Market-1501 DCAC DukeMTMC-reID->Rank1 71.5 #4 of 5 Archive leaderboard report
Generalizable Person Re-identification Market-1501 DCAC DukeMTMC-reID->mAP 42.3 #4 of 5 Archive leaderboard report
Generalizable Person Re-identification Market-1501 DCAC MSMT17->Rank-1 77.9 #4 of 5 Archive leaderboard report
Generalizable Person Re-identification Market-1501 DCAC MSMT17->mAP 52.1 #4 of 5 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

AttentionContrastive LearningDiffusionSETSoftmax

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections