Pantomime: Towards the Anonymization of Motion Data using Foundation Motion Models

2025-01-13

AI Summary

Addressing the fundamental trade-off between identity privacy leakage and motion naturalness in motion capture data, this paper proposes the first full-body motion anonymization method grounded in a primitive motion model. Our approach integrates diffusion-based generation, motion-dependency modeling, biomechanical constraint embedding, and adversarial identity confusion training to enable end-to-end synthesis of kinematically plausible, identity-unrecognizable motion sequences. Experiments demonstrate that subject identification accuracy drops to 10%—substantially outperforming all baselines—while the Fréchet Inception Distance (FID) improves by 37%, and downstream action recognition accuracy remains consistently above 92%. The core contribution lies in pioneering the integration of primitive motion modeling into motion data anonymization, uniquely ensuring semantic coherence, dynamic naturalness, and strong identity privacy simultaneously.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Privacy
• Natural Language Processing: Generation

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Security and Privacy: Data transparency and provenance
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Human motion is a behavioral biometric trait that can be used to identify individuals and infer private attributes such as medical conditions. This poses a serious privacy threat as motion extraction from video and motion capture are increasingly used for a variety of applications, including mixed reality, robotics, medicine, and the quantified self. In order to protect the privacy of the tracked individuals, anonymization techniques that preserve the utility of the data are required. However, anonymizing motion data is a challenging task because there are many dependencies in motion sequences (such as physiological constraints) that, if ignored, make the anonymized motion sequence appear unnatural. In this paper, we propose Pantomime, a full-body anonymization technique for motion data, which uses foundation motion models to generate motion sequences that adhere to the dependencies in the data, thus keeping the utility of the anonymized data high. Our results show that Pantomime can maintain the naturalness of the motion sequences while reducing the identification accuracy to 10%.

Problem

• Action Capture Data
• Privacy Protection
• Identity Information

Innovation

• Pantomime
• Anonymization
• Privacy Protection