GROW：面向扩散模型的渐进引导水印生成
GROW: Watermark Generation with Progressive Guidance for Diffusion Models
把水印嵌入从一次性写入初始噪声改为沿扩散采样过程用频域梯度逐步引导生成，从而省去 DDIM 反转，为扩散模型版权保护提供近 100 倍加速的免训练检测方案。
原文依据 reframes watermarking from a one-shot "embedding" to a progressive "growth"
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论数字水印是版权保护的基石。随着扩散模型等生成模型的快速发展，生成中且免训练的水印技术因其内源性与便捷性而受到更多关注。这类方法通常将水印嵌入初始噪声，水印提取依赖去噪扩散隐式模型（DDIM）反转。然而，计算开销巨大的提取过程严重阻碍了其走向实际部署。为克服这一关键瓶颈，我们提出 GROW，一种新颖的免训练范式，将水印从一次性的“嵌入”重构为渐进的“生长”。通过使用频域梯度进行渐进引导，GROW 自然地将水印编织进图像，从而实现无反转提取。在多个数据集上的全面实验表明，GROW 不仅取得了更优的鲁棒性与不可感知性，而且检测速度比基于反转的技术快近 100 倍。代码将公开。
英文原文摘要 1099 字符
Digital watermarking is a cornerstone for copyright protection. With the rapid advancement of generative models like diffusion models, in-generation and training-free watermarking techniques have garnered more attention for their endogeneity and convenience. These methods typically embed a watermark into the initial noise, where watermark extraction relies on Denoising Diffusion Implicit Models (DDIM) inversion. However, the computationally intensive extraction process severely hinders their path toward practical deployment. To overcome this critical bottleneck, we propose GROW, a novel training-free paradigm that reframes watermarking from a one-shot "embedding" to a progressive "growth". By progressively guiding using frequency-domain gradients, GROW naturally weaves the watermark into the image, which enables inversion-free extraction. Comprehensive experiments on multiple datasets show that GROW not only achieves superior robustness and imperceptibility but also offers a detection speed nearly 100x faster than inversion-based techniques. The code will be made publicly available.