Diffusion-based Synthetic Data Generation for Visible-Infrared Person Re-Identification

2025-03-16

AI Summary

To address the scarcity of real RGB-IR paired data, high annotation costs, and privacy constraints in Visible-Infrared Person Re-Identification (VI-ReID), this paper proposes DiVE—the first identity-aware diffusion model for VI-ReID. DiVE decouples identity representations from modality-specific features, enabling controllable cross-modal image synthesis without requiring real paired samples. It innovatively adapts text-to-image diffusion models (e.g., Stable Diffusion) into identity-preserving RGB-IR pair generators, integrating identity clustering-based representation learning and modality-conditional control mechanisms. When integrated with the CAJ model on the LLCM dataset, training on DiVE-synthesized data improves mAP by 9%, substantially alleviating data dependency. This demonstrates both the efficacy and practicality of diffusion-generated synthetic data for VI-ReID.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Humans and AI: Voting
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

The performance of models is intricately linked to the abundance of training data. In Visible-Infrared person Re-IDentification (VI-ReID) tasks, collecting and annotating large-scale images of each individual under various cameras and modalities is tedious, time-expensive, costly and must comply with data protection laws, posing a severe challenge in meeting dataset requirements. Current research investigates the generation of synthetic data as an efficient and privacy-ensuring alternative to collecting real data in the field. However, a specific data synthesis technique tailored for VI-ReID models has yet to be explored. In this paper, we present a novel data generation framework, dubbed Diffusion-based VI-ReID data Expansion (DiVE), that automatically obtain massive RGB-IR paired images with identity preserving by decoupling identity and modality to improve the performance of VI-ReID models. Specifically, identity representation is acquired from a set of samples sharing the same ID, whereas the modality of images is learned by fine-tuning the Stable Diffusion (SD) on modality-specific data. DiVE extend the text-driven image synthesis to identity-preserving RGB-IR multimodal image synthesis. This approach significantly reduces data collection and annotation costs by directly incorporating synthetic data into ReID model training. Experiments have demonstrated that VI-ReID models trained on synthetic data produced by DiVE consistently exhibit notable enhancements. In particular, the state-of-the-art method, CAJ, trained with synthetic images, achieves an improvement of about $9%$ in mAP over the baseline on the LLCM dataset. Code: https://github.com/BorgDiven/DiVE

Problem

• Addresses scarcity of training data in VI-ReID tasks.
• Proposes synthetic data generation to reduce collection costs.
• Enhances VI-ReID model performance using identity-preserving synthetic images.

Innovation

• Diffusion-based synthetic data generation for VI-ReID
• Decouples identity and modality for image synthesis
• Reduces data collection and annotation costs significantly