Video generative models get robust invisible watermarks for ownership verification
An Efficient and Effective Watermarking Scheme for the Protection of the Intellectual Property Rights of Video Generative Models
Computer Vision and Pattern Recognition
Summary
Videos created by advanced AI models can look very real, which raises concerns about who owns these models and their creations. The authors developed a new way to add invisible watermarks directly into synthetic videos during their creation. These watermarks are hard to remove, don’t affect video quality, and can prove if a video came from a specific AI model or if a model is a copy of another. Their tests showed the method works very well, accurately identifying watermarked videos and verifying ownership even under attempts to hide or alter the watermark.
What this means in practice
- •For video platform operators: Detect videos generated by protected AI models to enforce copyright and prevent unauthorized use.
- •For content protection teams: Verify if a suspicious AI video model is a copied or unauthorized version of a protected original.
Authors
Wenhong Huang, Jianwei Fei, Benedetta Tondi, Bin Ma, Fangjun Huang
Abstract
The rapid development of video generative models (VGMs) has enabled the generation of highly realistic synthetic videos, raising concerns about the intellectual property rights (IPR) of these models. In particular, two closely related forensic tasks remain largely unaddressed: synthetic video verification (determining whether a video was generated by a protected VGM) and model ownership verification (determining whether a suspect VGM is an unauthorized copy of a protected VGM). In this paper, we propose a new in-generation watermarking scheme that can address the two verification tasks. First, a novel video watermarking network named VidMark is presented, which incorporates a two-scale discrete wavelet transform (DWT) decomposition and a global temporal attention block (GTAB) to enhance watermark robustness and imperceptibility. Second, we present a decoder-guided fine-tuning procedure. By leveraging the frozen VidMark decoder, this process enables VGMs to synthesize videos carrying an imperceptible, robust, and model-specific watermark. Finally, two verification frameworks are established to perform synthetic video verification and model ownership verification. Extensive experiments on representative VGMs demonstrate that the proposed scheme achieves over 99% watermark extraction accuracy and 100% verification accuracy on both tasks, with negligible impact on video generation quality. Furthermore, the watermarks exhibit strong robustness against a comprehensive range of video-level and model-level attacks.