Problem
Research questions and friction points this paper is trying to address.
Mixed Reality
Gait Privacy
Security Issues
GaitGuard: Towards Private Gait in Mixed Reality
🤖 AI Summary
In mixed reality (MR) environments, gait—acting as a sensitive biometric—is vulnerable to adversarial identification, posing significant privacy risks. To address this, we propose the first lightweight, real-time gait privacy protection framework tailored for MR scenarios. Our method integrates multi-threaded video stream processing, YOLOv8-based human detection and tracking, and a privacy-mitigation module that jointly applies dynamic occlusion and pose perturbation. We further design a holistic evaluation metric balancing identification risk suppression, visual fidelity, and system efficiency. Experiments demonstrate a 68% reduction in gait identity recognition accuracy, real-time inference at 29 FPS on edge devices, and controlled degradation in video quality. A user study confirms seamless usability with negligible perceptual impact. This work establishes a practical, deployable technical pathway for real-time biometric privacy preservation on MR edge platforms.
📝 Abstract
Augmented/Mixed Reality (AR/MR) technologies usher in a new era of immersive, collective experiences, distinctly differentiating them from traditional mobile systems. As these technologies evolve, prioritizing privacy and security is critical. This paper centers on gait privacy, a distinctive biometric vulnerable to revealing sensitive data. We introduce GaitGuard, a real-time system to safeguard gait privacy within MR environments. GaitGuard leverages a multi-threaded framework to efficiently process video frames, incorporating dedicated modules for stream capture, body detection and tracking, and privacy mitigation. This study includes a user analysis involving 20 participants to evaluate the risk of gait information exposure captured by video feeds in MR devices. Through thorough examination, we provide a comparative assessment of different mitigation techniques, analyzing their impact on privacy, video quality, and system efficiency. Our results indicate that GaitGuard significantly diminishes identification risks by up to $68%$, while sustaining a robust streaming frame rate of $29$ FPS and preserving video clarity. GaitGuard offers a real-time approach to support privacy in MR applications, delivering a holistic solution to mitigate gait information exposure without affecting user experience.
Innovation
Methods, ideas, or system contributions that make the work stand out.
GaitGuard
Mixed Reality Privacy
Real-time Gait Protection
🔎 Similar Papers
No similar papers found.
