Towards Real-World Deepfake Detection: A Diverse In-the-wild Dataset of Forgery Faces

2025-10-09

AI Summary

Existing deepfake detection benchmarks lack real-world representativeness due to narrow forgery techniques, insufficient sample diversity, and inadequate modeling of black-box attack scenarios. Method: We introduce RedFace, a realistic benchmark comprising over 60,000 forged images and 1,000 manipulated videos. It is the first to systematically integrate nine commercial platform APIs and custom algorithms, capturing multi-source, multi-granularity, cross-domain, and socially propagated forgery patterns. We propose a “high-diversity–continuously-evolving” paradigm for synthetic data generation to better model real-world black-box attacks. Contribution/Results: Extensive experiments show that state-of-the-art detectors suffer significant performance degradation on RedFace, confirming its strong adversarial efficacy and challenge level. This exposes critical limitations of prior benchmarks and establishes RedFace as a more realistic, rigorous, and practically meaningful benchmark for developing robust deepfake detection models.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Application Domains: Misinformation & Fake News
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Deepfakes, leveraging advanced AIGC (Artificial Intelligence-Generated Content) techniques, create hyper-realistic synthetic images and videos of human faces, posing a significant threat to the authenticity of social media. While this real-world threat is increasingly prevalent, existing academic evaluations and benchmarks for detecting deepfake forgery often fall short to achieve effective application for their lack of specificity, limited deepfake diversity, restricted manipulation techniques.To address these limitations, we introduce RedFace (Real-world-oriented Deepfake Face), a specialized facial deepfake dataset, comprising over 60,000 forged images and 1,000 manipulated videos derived from authentic facial features, to bridge the gap between academic evaluations and real-world necessity. Unlike prior benchmarks, which typically rely on academic methods to generate deepfakes, RedFace utilizes 9 commercial online platforms to integrate the latest deepfake technologies found "in the wild", effectively simulating real-world black-box scenarios.Moreover, RedFace's deepfakes are synthesized using bespoke algorithms, allowing it to capture diverse and evolving methods used by real-world deepfake creators. Extensive experimental results on RedFace (including cross-domain, intra-domain, and real-world social network dissemination simulations) verify the limited practicality of existing deepfake detection schemes against real-world applications. We further perform a detailed analysis of the RedFace dataset, elucidating the reason of its impact on detection performance compared to conventional datasets. Our dataset is available at: https://github.com/kikyou-220/RedFace.

Problem

• Addressing limited diversity in existing deepfake detection datasets
• Bridging the gap between academic benchmarks and real-world applications
• Simulating black-box scenarios using commercial deepfake platforms

Innovation

• Introduces RedFace dataset with 60,000 forged images
• Uses 9 commercial platforms for real-world deepfake simulation
• Employs bespoke algorithms to capture diverse manipulation methods