论文

轻量级适配器让图像水印模型互相转换

Adaptable Watermarking: Transforming Image Watermarking Models via Lightweight Adapters

精选理由

做水印或内容溯源的工程师可以看看这篇:不用重训整个模型,挂个 1-2M 参数的适配器就能把 TrustMark、WAM 这类模型转成另一种水印风格,解码准确率还几乎不掉。如果担心模型升级要推倒重来,这套做法值得参考。

arXiv 论文提出一套框架,可将 TrustMark、InvisMark、RoSteALS、HiDDeN、WAM 等编码器-解码器水印架构改造成模块化、可升级的形态,无需重新训练完整模型即可支持多种水印风格。

正文

技术上,方法在模型上挂载紧凑的参数高效适配器,并以多域残差对齐损失(覆盖空间、梯度和频率三个域)进行微调,同时通过 secret-loss 项保持比特准确率。每个组件只需 1-2M 可训练参数,例如多解码器推理管线中的解码器适配器仅为 2M 参数。

实验显示,适配后的编码器能近似复现目标模型的残差,解码器在标准增强和失真条件下以接近最优的准确率恢复消息。这项跨模型适配能力可用于构建可互操作、可升级的水印体系。

原文 · arXiv cs.LG

Adaptable Watermarking: Transforming Image Watermarking Models via Lightweight Adapters

We explore the adaptability of neural image watermarking models; the extent to which one watermarking method can be transformed into another through lightweight adaptation. We introduce a framework that enables encoder-decoder watermarking architectures to become modular and upgradable, supporting multiple watermarking styles without re-training full models. Adaptation is performed by attaching a compact parameter-efficient adapter and fine-tuning under a multi-domain residual alignment loss (spatial, gradient, and frequency) while preserving bit accuracy via a secret-loss term. We evaluate this approach across a suite of recent watermarking models, including TrustMark, InvisMark, RoSteALS, HiDDeN, and WAM, demonstrating that adapted encoders closely reproduce target residuals and that adapted decoders recover messages with near-optimal accuracy under standard augmentations and distortions. We show that cross-model adaptation, with applications in interoperable and upgradable watermarking, can be achieved with as few as 1-2M trainable parameters per component (e.g., the 2M decoder adapter used in our multi-decoder inference pipeline).

产业坐标与信源核验

本条情报归属 AITOP「论文」核心追踪栏目。在跨区域产业观察坐标中,该动态直接关联「AI 开发者工具链、Coding Agent 与工程化实践」研究主线,反映出当前技术演进与商业落地的最新风向。涉及核心实体与技术议题包括:watermarking、TrustMark、HiDDeN、参数高效微调,读者可通过下方标签进入对应专题页查阅历史脉络与关联对比。

信源核验说明:本文原始线索采编自「arXiv cs.LG」(@Mehrdad Saberi, Vishal Asnani, Shruti Agarwal, Soheil Feizi, John Collomosse),经 AITOP 多维评分管线结构化提炼。