RCDN: Real-Centered Detection Network for Robust Face Forgery Identification

2026-01-17

AI Summary

Existing methods for face forgery detection suffer significant performance degradation in cross-domain scenarios due to the diversity of forgery techniques and distributional shifts. To address this, this work proposes a Realness-Centric Detection Network (RCDN) that leverages a dual-branch architecture to jointly model frequency- and spatial-domain features. Central to RCDN is the introduction of a realness-centered loss, which anchors the representation space to the distribution of genuine faces, prioritizing consistency in authentic images over learning specific forgery patterns. This paradigm substantially enhances cross-domain generalization and narrows the performance gap between in-domain and out-of-domain settings. Evaluated on the DiFF dataset, RCDN achieves state-of-the-art in-domain accuracy across three forgery types—Face Editing (FE), Image-to-Image (I2I), and Text-to-Image (T2I)—while demonstrating superior cross-domain generalization and the highest stability ratio between cross-domain and in-domain performance.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Application Domains: Misinformation & Fake News

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Image forgery has become a critical threat with the rapid proliferation of AI-based generation tools, which make it increasingly easy to synthesize realistic but fraudulent facial content. Existing detection methods achieve near-perfect performance when training and testing are conducted within the same domain, yet their effectiveness deteriorates substantially in crossdomain scenarios. This limitation is problematic, as new forgery techniques continuously emerge and detectors must remain reliable against unseen manipulations. To address this challenge, we propose the Real-Centered Detection Network (RCDN), a frequency spatial convolutional neural networks(CNN) framework with an Xception backbone that anchors its representation space around authentic facial images. Instead of modeling the diverse and evolving patterns of forgeries, RCDN emphasizes the consistency of real images, leveraging a dual-branch architecture and a real centered loss design to enhance robustness under distribution shifts. Extensive experiments on the DiFF dataset, focusing on three representative forgery types (FE, I2I, T2I), demonstrate that RCDN achieves both state-of-the-art in-domain accuracy and significantly stronger cross-domain generalization. Notably, RCDN reduces the generalization gap compared to leading baselines and achieves the highest cross/in-domain stability ratio, highlighting its potential as a practical solution for defending against evolving and unseen image forgery techniques.

Problem

• face forgery detection
• cross-domain generalization
• image manipulation
• distribution shift
• deepfake

Innovation

• Real-Centered Detection
• Cross-Domain Generalization
• Face Forgery Identification
• Frequency-Spatial CNN
• Distribution Shift Robustness