Seeing Through the Blur: Unlocking Defocus Maps for Deepfake Detection

Authors: Minsun Jeon, Simon S. Woo

Published: 2025-09-27 13:02:53+00:00

Comment: 12 pages, 9 figures

AI Summary

This paper proposes using defocus blur as a physically interpretable forensic signal for deepfake detection, leveraging its natural occurrence in camera-captured images but often inconsistent presence in synthetic content. The approach constructs defocus blur maps as discriminative features, providing a robust and generalizable method to distinguish real from manipulated visual media. Experimental results confirm its reliability and interpretability as an effective deepfake detection cue.

Abstract

The rapid advancement of generative AI has enabled the mass production of photorealistic synthetic images, blurring the boundary between authentic and fabricated visual content. This challenge is particularly evident in deepfake scenarios involving facial manipulation, but also extends to broader AI-generated content (AIGC) cases involving fully synthesized scenes. As such content becomes increasingly difficult to distinguish from reality, the integrity of visual media is under threat. To address this issue, we propose a physically interpretable deepfake detection framework and demonstrate that defocus blur can serve as an effective forensic signal. Defocus blur is a depth-dependent optical phenomenon that naturally occurs in camera-captured images due to lens focus and scene geometry. In contrast, synthetic images often lack realistic depth-of-field (DoF) characteristics. To capture these discrepancies, we construct a defocus blur map and use it as a discriminative feature for detecting manipulated content. Unlike RGB textures or frequency-domain signals, defocus blur arises universally from optical imaging principles and encodes physical scene structure. This makes it a robust and generalizable forensic cue. Our approach is supported by three in-depth feature analyses, and experimental results confirm that defocus blur provides a reliable and interpretable cue for identifying synthetic images. We aim for our defocus-based detection pipeline and interpretability tools to contribute meaningfully to ongoing research in media forensics. The implementation is publicly available at: https://github.com/irissun9602/Defocus-Deepfake-Detection

Key findings

The proposed defocus-based method achieved state-of-the-art performance with an average AUC of 0.998 on the FaceForensics++ dataset using only defocus blur maps. It also consistently improved AIGC detection performance on the SYNDOF dataset when fused with RGB features. Feature and model-level analyses confirmed that defocus blur patterns significantly differ between real and fake images, and the detection model actively leverages these interpretable physical cues.

Approach

The method extracts a defocus blur map from input images, quantifying pixel-wise blur intensity based on optical effects, a phenomenon often inconsistently rendered in synthetic content. This defocus map then serves as the primary discriminative feature for a deep learning classifier, either as a standalone input for deepfake detection or fused with RGB features for enhanced AI-generated content (AIGC) detection.

Datasets

FaceForensics++, SYNDOF, LFDOF, CUHK, Flickr, Middlebury, SYNTHIA, MPI Sintel, RTF

Model(s)

XceptionNet, ResNet-50, EfficientNet-B4, ViT-16

Author countries

Republic of Korea

← Previous