What does this research mean for the field?

Frequency-domain models, specifically FreqNet, outperform CNN-based models like EfficientNetAutoAttB4 in deepfake detection across various metrics, highlighting their superior generalization capabilities. Novelty: ClaimNovelty.NOVEL_FINDING. Consensus alignment: ConsensusAlignment.CHALLENGES_CONSENSUS.

What question did this study set out to answer?

This research aims to compare the generalization capabilities of CNN-based and frequency-domain models in deepfake detection.

March 14, 2026Open Access

Generalization in Deepfake Detection: A Comparative Analysis of Frequency-Domain and CNN Approaches

Key Points

This research aims to compare the generalization capabilities of CNN-based and frequency-domain models in deepfake detection.
Conducted cross-database evaluation on the GANGen-Detection dataset
Employed EfficientNetAutoAttB4 as a CNN approach
Used FreqNet to represent frequency-domain models
Measured performance metrics including Accuracy, AUC-ROC, and F1-score
FreqNet consistently outperformed EfficientNet in all performance metrics
EfficientNet demonstrated strong performance on in-domain data but overfitted
FreqNet effectively captured spectral artifacts, improving robustness to unseen manipulations

Abstract

This study investigates deepfake detection, a rapidly evolving class of synthetic media with major financial and ethical implications. We evaluate the generalization of two detection approaches: Convolutional Neural Network (CNN)-based models, represented by EfficientNetAutoAttB4, and frequency-domain models, exemplified by FreqNet. While EfficientNetAutoAttB4 achieves strong performance on in-domain data, it tends to overfit, whereas FreqNet captures spectral artifacts that improve robustness to unseen manipulations. Cross-database evaluation on the GANGen-Detection dataset shows that FreqNet consistently outperforms EfficientNet across Accuracy, AUC-ROC, and F1-score. These results highlight the importance of frequency-domain representations for building more generalizable deepfake detectors and expose the limitations of purely spatial CNNs when facing out-of-distribution data. Code and example images to reproduce the experiments are available in the associated repository https://github.com/NathFarinha/deepfake-detection-generalization-efficientnet-freqnet.git.

Generalization in Deepfake Detection: A Comparative Analysis of Frequency-Domain and CNN Approaches

Key Points

Abstract

Cite This Study