用合成数据训练的模型能精准识别假图像，无需见过真实图像。
Detect Fake with Fake: Leveraging Synthetic Data-driven Representation for Synthetic Image Detection
- 用合成数据训练视觉变换器，无需真实图像即可学习区分真假图。
- 在未见过的GAN模型上，性能比CLIP提升10.32 mAP和4.73%准确率。
- 适合关注生成图像检测、少依赖真实数据的研究者。
仅通过合成数据获取的通用视觉表征是否有助于检测假图像？本文证明了合成数据驱动的表征在合成图像检测中有效。分析发现，使用最新视觉表征学习方法训练的视觉变换器，在预训练阶段完全未接触真实图像的情况下，仍能有效区分真实与虚假图像。特别地，将SynCLR作为骨干网络应用于当前最先进的检测方法，在未见过的GAN模型测试中，相比广泛使用的CLIP，mAP提升10.32，准确率提升4.73%。代码已公开于https://github.com/cvpaperchallenge/detect-fake-with-fake。
原文摘要 · Abstract (English)
Are general-purpose visual representations acquired solely from synthetic data useful for detecting fake images? In this work, we show the effectiveness of synthetic data-driven representations for synthetic image detection. Upon analysis, we find that vision transformers trained by the latest visual representation learners with synthetic data can effectively distinguish fake from real images without seeing any real images during pre-training. Notably, using SynCLR as the backbone in a state-of-the-art detection method demonstrates a performance improvement of +10.32 mAP and +4.73% accuracy over the widely used CLIP, when tested on previously unseen GAN models. Code is available at https://github.com/cvpaperchallenge/detect-fake-with-fake.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。