Detect Fake with Fake: Leveraging Synthetic Data-driven Representation for Synthetic Image Detection
This addresses the challenge of fake image detection for security and media verification, offering a novel approach but is incremental in leveraging existing methods.
The paper tackles the problem of detecting synthetic images by using vision transformers trained solely on synthetic data, achieving a performance improvement of +10.32 mAP and +4.73% accuracy over CLIP on unseen GAN models.
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.