What Makes You Unique? Attribute Prompt Composition for Object Re-Identification

2025-09-23

AI Summary

Existing person re-identification (ReID) models suffer from overfitting in single-domain settings and diminished identity discriminability across domains due to domain-sensitive normalization strategies. To address this, we propose the Attribute-Prompt Composition Framework (APCF), a vision-language协同 approach tailored for cross-camera ReID. Our method introduces two key innovations: (1) an over-complete semantic attribute dictionary coupled with a composable prompt module, enabling fine-grained, adaptive attribute-aware feature generation; and (2) a dual-stream training paradigm—comprising fast (identity-discriminative) and slow (robust representation-preserving) branches—that jointly optimizes discriminability and generalization. APCF requires no additional annotations and is compatible with mainstream backbone architectures. Extensive experiments demonstrate state-of-the-art performance on standard benchmarks (Market-1501, DukeMTMC) and domain generalization ReID (DG-REID), achieving significant improvements in cross-domain generalization and identity discrimination.

Technology Category

• Computer Vision: Multi-modal Vision
• Application Domains: Humanities & Computational Social Science
• Cognitive Modeling & Cognitive Systems: Affective Computing

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Object Re-IDentification (ReID) aims to recognize individuals across non-overlapping camera views. While recent advances have achieved remarkable progress, most existing models are constrained to either single-domain or cross-domain scenarios, limiting their real-world applicability. Single-domain models tend to overfit to domain-specific features, whereas cross-domain models often rely on diverse normalization strategies that may inadvertently suppress identity-specific discriminative cues. To address these limitations, we propose an Attribute Prompt Composition (APC) framework, which exploits textual semantics to jointly enhance discrimination and generalization. Specifically, we design an Attribute Prompt Generator (APG) consisting of a Semantic Attribute Dictionary (SAD) and a Prompt Composition Module (PCM). SAD is an over-complete attribute dictionary to provide rich semantic descriptions, while PCM adaptively composes relevant attributes from SAD to generate discriminative attribute-aware features. In addition, motivated by the strong generalization ability of Vision-Language Models (VLM), we propose a Fast-Slow Training Strategy (FSTS) to balance ReID-specific discrimination and generalizable representation learning. Specifically, FSTS adopts a Fast Update Stream (FUS) to rapidly acquire ReID-specific discriminative knowledge and a Slow Update Stream (SUS) to retain the generalizable knowledge inherited from the pre-trained VLM. Through a mutual interaction, the framework effectively focuses on ReID-relevant features while mitigating overfitting. Extensive experiments on both conventional and Domain Generalized (DG) ReID datasets demonstrate that our framework surpasses state-of-the-art methods, exhibiting superior performances in terms of both discrimination and generalization. The source code is available at https://github.com/AWangYQ/APC.

Problem

• Overcoming domain-specific overfitting in object re-identification models
• Balancing identity discrimination with cross-domain generalization capabilities
• Preserving discriminative cues while adapting to diverse camera views

Innovation

• Attribute Prompt Composition using textual semantics
• Semantic Attribute Dictionary with Prompt Composition Module
• Fast-Slow Training Strategy balancing discrimination and generalization