Realism to Deception: Investigating Deepfake Detectors Against Face Enhancement

2025-09-08

AI Summary

This study identifies a critical vulnerability in deepfake detection: facial enhancement techniques—including conventional filters and GAN-based generative methods—improve visual quality but inadvertently distort biologically grounded forensic traces, thereby severely degrading detector robustness. We systematically evaluate three mainstream detection paradigms—naïve (pixel-level), spatial-domain, and frequency-domain—and find that basic enhancement reduces detection accuracy to 64.63%, while GAN-based enhancement further degrades performance to 75.12%. Crucially, we provide the first empirical evidence that facial enhancement functions as a stealthy anti-forensic attack. To counter this threat, we propose an adversarial training–based defense framework, demonstrating its effectiveness in enhancing model resilience against such enhancements. Our findings underscore a previously overlooked trade-off between aesthetic enhancement and forensic integrity, offering both a cautionary insight and a practical mitigation pathway for trustworthy deepfake detection deployment.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Privacy-enhancing technologies
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Face enhancement techniques are widely used to enhance facial appearance. However, they can inadvertently distort biometric features, leading to significant decrease in the accuracy of deepfake detectors. This study hypothesizes that these techniques, while improving perceptual quality, can degrade the performance of deepfake detectors. To investigate this, we systematically evaluate whether commonly used face enhancement methods can serve an anti-forensic role by reducing detection accuracy. We use both traditional image processing methods and advanced GAN-based enhancements to evaluate the robustness of deepfake detectors. We provide a comprehensive analysis of the effectiveness of these enhancement techniques, focusing on their impact on Naïve, Spatial, and Frequency-based detection methods. Furthermore, we conduct adversarial training experiments to assess whether exposure to face enhancement transformations improves model robustness. Experiments conducted on the FaceForensics++, DeepFakeDetection, and CelebDF-v2 datasets indicate that even basic enhancement filters can significantly reduce detection accuracy achieving ASR up to 64.63%. In contrast, GAN-based techniques further exploit these vulnerabilities, achieving ASR up to 75.12%. Our results demonstrate that face enhancement methods can effectively function as anti-forensic tools, emphasizing the need for more resilient and adaptive forensic methods.

Problem

• Face enhancement techniques reduce deepfake detector accuracy
• Study evaluates enhancement methods as anti-forensic tools
• GAN-based techniques exploit detector vulnerabilities most effectively

Innovation

• Evaluating face enhancement methods on deepfake detectors
• Using traditional and GAN-based techniques for robustness assessment
• Conducting adversarial training to improve model resilience