Detect Fake with Fake: Leveraging Synthetic Data-driven Representation for Synthetic Image Detection

Hina Otake, Yoshihiro Fukuhara, Yoshiki Kubotani, Shigeo Morishima
9/13/2024
cs.CVcs.LG

Official code: cvpaperchallenge/detect-fake-with-fake (8 stars, MIT, last commit 2024-09-18).

Abstract

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.

View on arXivView PDF

Code Implementations(1)

cvpaperchallenge/detect-fake-with-fakeOfficial100%
MIT

[ECCV 2024 TWYN Workshop] Detect Fake with Fake: Leveraging Synthetic Data-driven Representation for Synthetic Image Detection

80Python, MakefileAug 20, 20242 years agoMIT

Similar Papers

Towards Universal Fake Image Detectors that Generalize Across Generative Models

88% match1 repo3762023

RealStats: A Rigorous Real-Only Statistical Framework for Fake Image Detection

84% match2026

ForgeryTTT: Zero-Shot Image Manipulation Localization with Test-Time Training

84% match2024

Few-Shot Learner Generalizes Across AI-Generated Image Detection

84% match1 repo102026

Synthetic Image Detection with CLIP: Understanding and Assessing Predictive Cues

84% match2026

FakeInversion: Learning to Detect Images from Unseen Text-to-Image Models by Inverting Stable Diffusion

83% match1 repo62024

Cite this paper

@article{otake2024detect,
  title  = {Detect Fake with Fake: Leveraging Synthetic Data-driven Representation for Synthetic Image Detection},
  author = {Hina Otake and Yoshihiro Fukuhara and Yoshiki Kubotani and Shigeo Morishima},
  year   = {2024},
  eprint = {2409.08884},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2409.08884v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2409.08884.svg)](https://codeofpaper.com/paper/2409.08884)

Discussion