Efficient and Robust Video Defense Framework against 3D-field Personalized Talking Face

2025-12-24

AI Summary

3D-field-driven personalized talking face generation (TFG) poses severe privacy threats to portrait videos, yet existing image-level defense methods suffer from high computational overhead, significant quality degradation, and insufficient disruption of critical 3D information. Method: We propose the first efficient and robust video-level defense framework, uniquely targeting perturbation of the 3D geometry and appearance modeling process. Our approach introduces a similarity-guided parameter-sharing mechanism and a multi-scale dual-domain attention module for joint spatial-frequency optimization, coupled with lightweight 3D-aware perturbations and adversarial robustness enhancement strategies. Contribution/Results: Experiments demonstrate substantial improvements in defense success rate, with inference speed accelerated by 47× over the fastest baseline. The framework exhibits strong robustness against geometric scaling and state-of-the-art purification attacks. Ablation studies confirm the distinct and complementary contributions of each component.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Safety and Robustness

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Large-scale security measurements
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

State-of-the-art 3D-field video-referenced Talking Face Generation (TFG) methods synthesize high-fidelity personalized talking-face videos in real time by modeling 3D geometry and appearance from reference portrait video. This capability raises significant privacy concerns regarding malicious misuse of personal portraits. However, no efficient defense framework exists to protect such videos against 3D-field TFG methods. While image-based defenses could apply per-frame 2D perturbations, they incur prohibitive computational costs, severe video quality degradation, failing to disrupt 3D information for video protection. To address this, we propose a novel and efficient video defense framework against 3D-field TFG methods, which protects portrait video by perturbing the 3D information acquisition process while maintain high-fidelity video quality. Specifically, our method introduces: (1) a similarity-guided parameter sharing mechanism for computational efficiency, and (2) a multi-scale dual-domain attention module to jointly optimize spatial-frequency perturbations. Extensive experiments demonstrate that our proposed framework exhibits strong defense capability and achieves a 47x acceleration over the fastest baseline while maintaining high fidelity. Moreover, it remains robust against scaling operations and state-of-the-art purification attacks, and the effectiveness of our design choices is further validated through ablation studies. Our project is available at https://github.com/Richen7418/VDF.

Problem

• Defend portrait videos against 3D talking face generation misuse
• Protect 3D information in videos while preserving high quality
• Achieve efficient defense with minimal computational cost and degradation

Innovation

• Perturbing 3D information acquisition process for video protection
• Similarity-guided parameter sharing mechanism for computational efficiency
• Multi-scale dual-domain attention module for spatial-frequency perturbations