[论文解读] Towards the Detection of Diffusion Model Deepfakes

Jonas Ricker, Simon Damm+2 more|arXiv (Cornell University)|Oct 26, 2022

Digital Media Forensic Detection被引用 25

一句话总结

论文表明，预训练的GAN检测器在检测扩散模型（DM）深度伪造方面表现不佳，但在DM生成的图像上重新训练检测器可实现近乎完美的检测并且对GANs具有泛化能力；DM图像由于训练目标，呈现较少的伪影，尤其在高频区域。

ABSTRACT

In the course of the past few years, diffusion models (DMs) have reached an unprecedented level of visual quality. However, relatively little attention has been paid to the detection of DM-generated images, which is critical to prevent adverse impacts on our society. In contrast, generative adversarial networks (GANs), have been extensively studied from a forensic perspective. In this work, we therefore take the natural next step to evaluate whether previous methods can be used to detect images generated by DMs. Our experiments yield two key findings: (1) state-of-the-art GAN detectors are unable to reliably distinguish real from DM-generated images, but (2) re-training them on DM-generated images allows for almost perfect detection, which remarkably even generalizes to GANs. Together with a feature space analysis, our results lead to the hypothesis that DMs produce fewer detectable artifacts and are thus more difficult to detect compared to GANs. One possible reason for this is the absence of grid-like frequency artifacts in DM-generated images, which are a known weakness of GANs. However, we make the interesting observation that diffusion models tend to underestimate high frequencies, which we attribute to the learning objective.

研究动机与目标

• 评估最先进的GAN检测器是否能够区分真实图像与DM生成图像。
• 评估在DM生成图像上重新训练的检测器在不同模型之间的泛化能力。
• 分析DM生成图像的特征表示和频域特征。
• 调查DM扩散模型的频率行为及去噪过程中的伪影存在原因。

提出的方法

• 使用AUROC和Pd@FAR(1%)在DM和GAN图像上评估三种基于CNN的检测器（Wang2020、Gragnaniello2021、Mandelli2022）。
• 在DM生成的图像上重新训练检测器，并测试跨模型在多种DM和GAN生成器之间的泛化。
• 用t-SNE分析学习到的特征空间，并通过DFT、DCT及降维频谱研究频域伪影。
• 检查扩散训练目标如何影响高频内容及去噪过程中的频谱低估。

实验结果

研究问题

• RQ1 预训练的GAN检测器是否可以可靠地区分真实与DM生成的图像？
• RQ2 在DM生成数据上重新训练检测器是否能提高DM检测并对GAN泛化？
• RQ3 DM与GAN生成图像在特征表示和频域伪影方面存在哪些差异？
• RQ4 DM训练目标如何影响生成图像中的高频内容？
• RQ5 扩散去噪步骤数量如何影响频谱准确性和可检测性？

主要发现

• 预训练的GAN检测器在DM生成的图像上性能显著下降（平均AUROC下降约15.2点）。
• 在DM生成图像上重新训练的检测器在DM数据上实现近乎完美的检测，并对GANs泛化。
• 在DM图像上训练的检测器将GAN和DM生成的图像映射到相似的嵌入，表明DM特定伪影较少。
• DM生成的图像总体上缺乏典型的网格状频率伪影，且由于DM训练目标，往往低估高频内容。
• 增加扩散去噪步骤有助于提高高频再现和检测性能；步骤较少时会造成更大程度的低估。