DART$^3$: Leveraging Distance for Test Time Adaptation in Person Re-Identification

2025-05-23

AI Summary

Pedestrian re-identification (ReID) suffers from severe camera bias in cross-camera deployment: learned representations are overly sensitive to viewpoint variations rather than identity-discriminative cues, leading to sharp performance degradation when deploying to new cameras. Existing test-time adaptation (TTA) methods—designed for classification—rely on entropy minimization and are ill-suited for retrieval tasks. This paper proposes the first source-free, model-agnostic TTA framework tailored specifically for ReID. Its core innovation is a distance-aware objective that jointly models the correlation between nearest-neighbor distances and prediction errors. The framework supports black-box and hybrid deployment scenarios, with a lightweight variant (DART$^3$-Lite) and an unsupervised online adaptation mechanism. Extensive experiments on multiple benchmarks demonstrate significant improvements over state-of-the-art TTA methods, effectively mitigating camera bias and enhancing cross-camera retrieval accuracy.

Technology Category

• Computer Vision: Image and Video Retrieval
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Person re-identification (ReID) models are known to suffer from camera bias, where learned representations cluster according to camera viewpoints rather than identity, leading to significant performance degradation under (inter-camera) domain shifts in real-world surveillance systems when new cameras are added to camera networks. State-of-the-art test-time adaptation (TTA) methods, largely designed for classification tasks, rely on classification entropy-based objectives that fail to generalize well to ReID, thus making them unsuitable for tackling camera bias. In this paper, we introduce DART$^3$, a TTA framework specifically designed to mitigate camera-induced domain shifts in person ReID. DART$^3$ (Distance-Aware Retrieval Tuning at Test Time) leverages a distance-based objective that aligns better with image retrieval tasks like ReID by exploiting the correlation between nearest-neighbor distance and prediction error. Unlike prior ReID-specific domain adaptation methods, DART$^3$ requires no source data, architectural modifications, or retraining, and can be deployed in both fully black-box and hybrid settings. Empirical evaluations on multiple ReID benchmarks indicate that DART$^3$ and DART$^3$ LITE, a lightweight alternative to the approach, consistently outperforms state-of-the-art TTA baselines, making for a viable option to online learning to mitigate the adverse effects of camera bias.

Problem

• Addresses camera bias in person re-identification models
• Improves test-time adaptation for domain shifts in ReID
• Leverages distance-based objectives without source data

Innovation

• Distance-based objective for ReID adaptation
• No source data or retraining required
• Works in black-box and hybrid settings