Abstract
Deepfake technology poses increasing risks such as privacy invasion andidentity theft. To address these threats, we propose WaveGuard, a proactivewatermarking framework that enhances robustness and imperceptibility viafrequency-domain embedding and graph-based structural consistency.Specifically, we embed watermarks into high-frequency sub-bands using Dual-TreeComplex Wavelet Transform (DT-CWT) and employ a Structural Consistency GraphNeural Network (SC-GNN) to preserve visual quality. We also design an attentionmodule to refine embedding precision. Experimental results on face swap andreenactment tasks demonstrate that WaveGuard outperforms state-of-the-artmethods in both robustness and visual quality. Code is available athttps://github.com/vpsg-research/WaveGuard.