Search papers, labs, and topics across Lattice.
This paper introduces Attention-guided Noise Learning (ANL), a novel deepfake detection framework that leverages the denoising process of a pre-trained diffusion model to expose subtle artifacts in synthetic images. ANL trains a detector to predict the noise contained in an input image at a given diffusion step, using an attention mechanism to focus on global discrepancies. Experiments show ANL significantly outperforms existing methods in detecting diffusion-generated deepfakes and generalizes better to unseen forgery types without increasing inference overhead.
Diffusion models, often used to create deepfakes, also hold the key to detecting them: training a detector to predict diffusion noise reveals subtle, globally distributed artifacts that expose synthetic images.
Deepfake detectors face growing challenges in generalization as new image synthesis techniques emerge. In particular, deepfakes generated by diffusion models are highly photorealistic and often evade detectors trained on GAN-based forgeries. This paper addresses the generalization problem in deepfake detection by leveraging diffusion noise characteristics. We propose an Attention-guided Noise Learning (ANL) framework that integrates a pre-trained diffusion model into the deepfake detection pipeline to guide the learning of more robust features. Specifically, our method uses the diffusion model's denoising process to expose subtle artifacts: the detector is trained to predict the noise contained in an input image at a given diffusion step, forcing it to capture discrepancies between real and synthetic images, while an attention-guided mechanism derived from the predicted noise is introduced to encourage the model to focus on globally distributed discrepancies rather than local patterns. By harnessing the frozen diffusion model's learned distribution of natural images, the ANL method acts as a form of regularization, improving the detector's generalization to unseen forgery types. Extensive experiments demonstrate that ANL significantly outperforms existing methods on multiple benchmarks, achieving state-of-the-art accuracy in detecting diffusion-generated deepfakes. Notably, the proposed framework boosts generalization performance (e.g., improving ACC/AP by a substantial margin on unseen models) without introducing additional overhead during inference. Our results highlight that diffusion noise provides a powerful signal for generalizable deepfake detection.
