Towards Generalized Proactive Defense against Face Swapping with Contour-Hybrid Watermark

2025-05-25

AI Summary

AI-generated face-swapping poses severe privacy and security threats; existing passive detection methods—relying on forensic traces—suffer from poor generalization and strong dependence on algorithm-specific priors. Method: We propose an active defense paradigm that embeds a hybrid watermark (CMark) into the facial contour region, jointly encoding geometric contour cues and identity semantics. CMark requires no training data and achieves zero-shot generalization. Our approach integrates contour-driven watermark generation, multimodal encoding, and a lightweight, robust embedding/extraction network. Results: Evaluated across eight state-of-the-art face-swapping algorithms, our method achieves an average detection accuracy exceeding 98.6%, significantly outperforming both active and passive baselines, while preserving high visual fidelity (PSNR > 42 dB). The core contribution is the first contour-region hybrid watermarking mechanism, eliminating reliance on face-swapping algorithm priors and large-scale training datasets.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Face swapping, recognized as a privacy and security concern, has prompted considerable defensive research. With the advancements in AI-generated content, the discrepancies between the real and swapped faces have become nuanced. Considering the difficulty of forged traces detection, we shift the focus to the face swapping purpose and proactively embed elaborate watermarks against unknown face swapping techniques. Given that the constant purpose is to swap the original face identity while preserving the background, we concentrate on the regions surrounding the face to ensure robust watermark generation, while embedding the contour texture and face identity information to achieve progressive image determination. The watermark is located in the facial contour and contains hybrid messages, dubbed the contour-hybrid watermark (CMark). Our approach generalizes face swapping detection without requiring any swapping techniques during training and the storage of large-scale messages in advance. Experiments conducted across 8 face swapping techniques demonstrate the superiority of our approach compared with state-of-the-art passive and proactive detectors while achieving a favorable balance between the image quality and watermark robustness.

Problem

• Detecting face swapping with proactive watermarking
• Generalizing detection without prior swapping techniques
• Balancing image quality and watermark robustness

Innovation

• Proactive embedding of elaborate watermarks
• Contour-hybrid watermark for robust detection
• Generalizes detection without prior technique knowledge