Curvelet-Based Frequency-Aware Feature Enhancement for Deepfake Detection

2026-04-13

AI Summary

This work addresses the significant performance degradation of existing deepfake detection methods under image compression, which struggle to balance robustness and discriminative power. To overcome this limitation, the study introduces Curvelet transform for the first time in this domain, leveraging its multi-scale and directional properties to enhance frequency-domain feature representation. Specifically, a wedge-level attention mechanism and a scale-aware spatial mask are designed to capture salient forensic traces. These features are integrated with a modified Xception network that incorporates reconstructed inputs for final classification. The proposed method achieves 98.48% accuracy and 99.96% AUC on FaceForensics++ under low-compression settings, while maintaining strong performance even under high compression, thereby substantially improving both robustness and interpretability.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Learning on the Edge & Model Compression
• Data Mining & Knowledge Management: Data Compression

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

The proliferation of sophisticated generative models has significantly advanced the realism of synthetic facial content, known as deepfakes, raising serious concerns about digital trust. Although modern deep learning-based detectors perform well, many rely on spatial-domain features that degrade under compression. This limitation has prompted a shift toward integrating frequency-domain representations with deep learning to improve robustness. Prior research has explored frequency transforms such as Discrete Cosine Transform (DCT), Fast Fourier Transform (FFT), and Wavelet Transform, among others. However, to the best of our knowledge, the Curvelet Transform, despite its superior directional and multiscale properties, remains entirely unexplored in the context of deepfake detection. In this work, we introduce a novel Curvelet-based detection approach that enhances feature quality through wedge-level attention and scale-aware spatial masking, both trained to selectively emphasize discriminative frequency components. The refined frequency cues are reconstructed and passed to a modified pretrained Xception network for classification. Evaluated on two compression qualities in the challenging FaceForensics++ dataset, our method achieves 98.48% accuracy and 99.96% AUC on FF++ low compression, while maintaining strong performance under high compression, demonstrating the efficacy and interpretability of Curvelet-informed forgery detection.

Problem

• Deepfake Detection
• Curvelet Transform
• Frequency-domain Features
• Compression Robustness
• Digital Forgery

Innovation

• Curvelet Transform
• frequency-domain feature enhancement
• wedge-level attention
• scale-aware spatial masking
• deepfake detection