WaveGuard: Robust Deepfake Detection and Source Tracing via Dual-Tree Complex Wavelet and Graph Neural Networks

2025-05-13

AI Summary

To address privacy leakage and identity theft risks posed by Deepfakes, this paper proposes WaveGuard—the first proactive watermarking framework specifically designed for deepfake content. WaveGuard innovatively integrates dual-tree complex wavelet transform (DT-CWT)-based high-frequency subband watermark embedding with a structural consistency graph neural network (SC-GNN), enabling robust watermark injection in the frequency domain while preserving facial topological structure before and after embedding. An attention-guided watermark refinement module is further introduced to enhance visual fidelity and adversarial robustness. Extensive experiments demonstrate that WaveGuard outperforms state-of-the-art methods in detection accuracy, source attribution precision, and perceptual quality across face-swapping and facial reenactment tasks. The implementation is publicly available.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)

Application Category

• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Security and Privacy: Data transparency and provenance

Abstract

Deepfake technology poses increasing risks such as privacy invasion and identity theft. To address these threats, we propose WaveGuard, a proactive watermarking framework that enhances robustness and imperceptibility via frequency-domain embedding and graph-based structural consistency. Specifically, we embed watermarks into high-frequency sub-bands using Dual-Tree Complex Wavelet Transform (DT-CWT) and employ a Structural Consistency Graph Neural Network (SC-GNN) to preserve visual quality. We also design an attention module to refine embedding precision. Experimental results on face swap and reenactment tasks demonstrate that WaveGuard outperforms state-of-the-art methods in both robustness and visual quality. Code is available at https://github.com/vpsg-research/WaveGuard.

Problem

• Detect deepfakes robustly using dual-tree wavelet transforms
• Trace deepfake sources via graph neural networks
• Enhance watermark robustness while preserving visual quality

Innovation

• Uses Dual-Tree Complex Wavelet Transform
• Employs Structural Consistency GNN
• Integrates attention module for precision