Towards Sustainable Universal Deepfake Detection with Frequency-Domain Masking

2025-12-08

AI Summary

To address the challenges of poor generalization, high computational overhead, and limited adaptability to emerging generative models—particularly unseen diffusion models—in deepfake detection, this paper proposes a lightweight, universal detection framework based on frequency-domain masked training. Our method jointly applies random frequency masking and geometric transformations to reduce reliance on spatial texture cues and large pretrained models, while incorporating structured pruning to significantly compress model size without compromising robustness. Evaluated on a multi-source deepfake dataset encompassing both GAN- and diffusion-based forgeries, our approach achieves state-of-the-art cross-model generalization accuracy. It reduces parameter count by 67%, accelerates inference by 3.2×, and enables scalable, energy-efficient deployment aligned with green AI principles.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Universal deepfake detection aims to identify AI-generated images across a broad range of generative models, including unseen ones. This requires robust generalization to new and unseen deepfakes, which emerge frequently, while minimizing computational overhead to enable large-scale deepfake screening, a critical objective in the era of Green AI. In this work, we explore frequency-domain masking as a training strategy for deepfake detectors. Unlike traditional methods that rely heavily on spatial features or large-scale pretrained models, our approach introduces random masking and geometric transformations, with a focus on frequency masking due to its superior generalization properties. We demonstrate that frequency masking not only enhances detection accuracy across diverse generators but also maintains performance under significant model pruning, offering a scalable and resource-conscious solution. Our method achieves state-of-the-art generalization on GAN- and diffusion-generated image datasets and exhibits consistent robustness under structured pruning. These results highlight the potential of frequency-based masking as a practical step toward sustainable and generalizable deepfake detection. Code and models are available at: [https://github.com/chandlerbing65nm/FakeImageDetection](https://github.com/chandlerbing65nm/FakeImageDetection).

Problem

• Develops universal deepfake detection for diverse AI-generated images
• Enhances generalization to unseen models while reducing computational costs
• Proposes frequency-domain masking for scalable and sustainable detection

Innovation

• Frequency-domain masking enhances deepfake detection generalization
• Random and geometric transformations improve detector robustness
• Method maintains accuracy under model pruning for scalability