Hierarchical Anti-Aesthetics: Protecting Facial Privacy against Customized Diffusion Models

2026-07-02

AI Summary

This work addresses the critical privacy risks posed by customized diffusion models capable of generating highly realistic human faces. To counteract this threat, the study introduces aesthetic perception into face privacy protection and proposes a hierarchical anti-aesthetic perturbation framework that jointly degrades identity recognizability at both global and local levels. By designing reward-based global and local anti-aesthetic loss functions and generating adversarial perturbations that deliberately deviate from authentic facial aesthetic attributes, the method effectively reduces both the aesthetic quality and identity similarity of maliciously synthesized images. Experimental results demonstrate that the proposed approach significantly outperforms existing techniques in identity removal, establishing a novel paradigm for face privacy preservation.

Abstract

The rise of customized diffusion models has fueled a boom in personalized visual content creation, but it also introduces serious risks of malicious misuse, thereby posing threats to personal privacy. Image aesthetics are strongly correlated with human perception of image quality. Motivated by this observation, we address facial privacy protection from a novel aesthetic perspective by degrading the generation quality of maliciously customized models, thus reducing facial identity leakage. Specifically, we propose a Hierarchical Anti-Aesthetics (HAA) framework that exploits aesthetic cues at multiple perceptual levels. HAA consists of two key branches: (1) Global Anti-Aesthetics, which degrades overall aesthetics and generation quality by constructing a global anti-aesthetic reward mechanism and a corresponding loss; and (2) Local Anti-Aesthetics, which disrupts facial identity by using a local anti-aesthetic reward mechanism and loss to guide adversarial perturbations toward facial regions. By integrating both branches, HAA achieves anti-aesthetic degradation from a global to a local level during customized generation. Extensive experiments show that HAA outperforms existing methods in identity removal, providing an effective tool for protecting facial privacy.

Problem

• facial privacy
• customized diffusion models
• identity leakage
• anti-aesthetics
• privacy protection

Innovation

• Hierarchical Anti-Aesthetics
• facial privacy protection
• customized diffusion models
• adversarial perturbations
• aesthetic degradation