Towards Anytime Retrieval: A Benchmark for Anytime Person Re-Identification

2025-09-20

AI Summary

Existing person re-identification (ReID) methods are constrained by fixed time periods and single-scene settings, failing to address practical retrieval challenges such as diurnal variation, long-term temporal spans, and clothing changes. To bridge this gap, we propose “Any-Time Person ReID” (AT-ReID), a new task demanding robust identification across arbitrary times and conditions. We introduce AT-USTC—the first large-scale cross-temporal multimodal dataset—comprising 403k RGB/infrared images captured over 21 months with diverse clothing variations. To achieve cross-scene robust matching, we design Uni-AT, a unified model featuring a multi-scene feature learning framework, a Mixture-of-Attribute-Experts (MoAE) module, and a Hierarchical Dynamic Weighting (HDW) strategy to mitigate scene-specific interference. Extensive experiments demonstrate that Uni-AT achieves state-of-the-art performance across all time periods and scenes, exhibiting strong generalization and balanced accuracy. This work establishes a new benchmark and viable paradigm for any-time person retrieval.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multimodal Learning

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

Abstract

In real applications, person re-identification (ReID) is expected to retrieve the target person at any time, including both daytime and nighttime, ranging from short-term to long-term. However, existing ReID tasks and datasets can not meet this requirement, as they are constrained by available time and only provide training and evaluation for specific scenarios. Therefore, we investigate a new task called Anytime Person Re-identification (AT-ReID), which aims to achieve effective retrieval in multiple scenarios based on variations in time. To address the AT-ReID problem, we collect the first large-scale dataset, AT-USTC, which contains 403k images of individuals wearing multiple clothes captured by RGB and IR cameras. Our data collection spans 21 months, and 270 volunteers were photographed on average 29.1 times across different dates or scenes, 4-15 times more than current datasets, providing conditions for follow-up investigations in AT-ReID. Further, to tackle the new challenge of multi-scenario retrieval, we propose a unified model named Uni-AT, which comprises a multi-scenario ReID (MS-ReID) framework for scenario-specific features learning, a Mixture-of-Attribute-Experts (MoAE) module to alleviate inter-scenario interference, and a Hierarchical Dynamic Weighting (HDW) strategy to ensure balanced training across all scenarios. Extensive experiments show that our model leads to satisfactory results and exhibits excellent generalization to all scenarios.

Problem

• Achieving effective person retrieval across multiple time scenarios
• Addressing limitations of existing ReID datasets constrained by time
• Developing unified model for anytime retrieval with varying conditions

Innovation

• Multi-scenario ReID framework for feature learning
• Mixture-of-Attribute-Experts module reduces interference
• Hierarchical Dynamic Weighting ensures balanced training