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

TL;DR

This paper introduces an efficient video defense framework that protects against 3D-field personalized talking face generation by perturbing 3D information, achieving high fidelity, robustness, and significant computational acceleration.

Contribution

The paper proposes a novel defense method that efficiently disrupts 3D information in videos, maintaining quality and robustness against attacks, with innovative similarity-guided sharing and multi-scale attention modules.

Findings

Achieves 47x faster defense compared to baseline methods.

Maintains high video fidelity while disrupting 3D information.

Remains robust against scaling and purification attacks.

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.