Micro-Defects Expose Macro-Fakes: Detecting AI-Generated Images via Local Distributional Shifts

Hugging Face Daily Papers Papers
ai-generated-image-detection local-distribution-shifts micro-defects forensic-signature patch-wise-modeling deepfake-detection

Summary

A local distribution-aware detection framework that amplifies micro-scale statistical irregularities to identify AI-generated images with improved accuracy, outperforming baseline detectors across benchmarks.

Recent generative models can produce images that appear highly realistic, raising challenges in distinguishing real and AI-generated images. Yet existing detectors based on pre-trained feature extractors tend to over-rely on global semantics, limiting sensitivity to the critical micro-defects. In this work, we propose Micro-Defects expose Macro-Fakes (MDMF), a local distribution-aware detection framework that amplifies micro-scale statistical irregularities into macro-level distributional discrepancies. To avoid localized forensic cues being diluted by plain aggregation, we introduce a learnable Patch Forensic Signature that projects semantic patch embeddings into a compact forensic latent space. We then use Maximum Mean Discrepancy (MMD) to quantify distributional discrepancies between generated and real images. Our theory-grounded analysis shows that patch-wise modeling yields provably larger discrepancies when localized forensic signals are present in generated images, enabling more reliable separation from real images. Extensive experiments demonstrate that MDMF consistently outperforms baseline detectors across multiple benchmarks, validating its general effectiveness. Project page: https://zbox1005.github.io/MDMF-project/
Original Article
View Cached Full Text

Cached at: 05/13/26, 08:14 PM

Paper page - Micro-Defects Expose Macro-Fakes: Detecting AI-Generated Images via Local Distributional Shifts

Source: https://huggingface.co/papers/2605.09296

Abstract

A local distribution-aware detection framework that amplifies micro-scale statistical irregularities to identify AI-generated images with improved accuracy.

Recentgenerative modelscan produce images that appear highly realistic, raising challenges in distinguishing real and AI-generated images. Yet existing detectors based onpre-trained feature extractorstend to over-rely onglobal semantics, limiting sensitivity to the criticalmicro-defects. In this work, we proposeMicro-Defectsexpose Macro-Fakes (MDMF), alocal distribution-aware detectionframework that amplifies micro-scale statistical irregularities into macro-leveldistributional discrepancies. To avoid localized forensic cues being diluted by plain aggregation, we introduce a learnablePatch Forensic Signaturethat projects semantic patch embeddings into a compactforensic latent space. We then useMaximum Mean Discrepancy(MMD) to quantifydistributional discrepanciesbetween generated and real images. Our theory-grounded analysis shows thatpatch-wise modelingyields provably larger discrepancies when localized forensic signals are present in generated images, enabling more reliable separation from real images. Extensive experiments demonstrate that MDMF consistently outperforms baseline detectors across multiple benchmarks, validating its general effectiveness. Project page: https://zbox1005.github.io/MDMF-project/

View arXiv pageView PDFProject pageGitHub2Add to collection

Get this paper in your agent:

hf papers read 2605\.09296

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2605.09296 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2605.09296 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2605.09296 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

The Alpha Blending Hypothesis: Compositing Shortcut in Deepfake Detection

Hugging Face Daily Papers

This paper introduces the Alpha Blending Hypothesis, suggesting deepfake detectors primarily identify compositing artifacts rather than semantic anomalies. It proposes a method called BlenD that achieves superior cross-dataset generalization using real-only image augmentation with self-blended images.

MELD: Multi-Task Equilibrated Learning Detector for AI-Generated Text

arXiv cs.CL

This paper introduces MELD, a detector for AI-generated text that uses multi-task learning with auxiliary heads for generator family, attack type, and source domain to improve robustness. MELD achieves strong performance on the RAID benchmark and maintains low false-positive rates under adversarial attacks.