Research map / Adversarial robustness and security / Attacks on text and graphs
Anti-Tamper Protection for Unauthorized Individual Image Generation
Zelin Li, Ruohan Zong, Yifan Liu, Ruichen Yao, Yaokun Liu, Yang Zhang, Dong Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add ATP to MetaCloak | MetaCloak · CelebA-HQ (no purification) | FDFR: 0.4850 -> 0.7038 | Improved |
| add ATP to Anti-DB | Anti-DB · VGGFace2 (no purification) | FDFR: 0.6788 -> 0.8188 | Improved |
| add ATP to CAAT | CAAT · VGGFace2 (no purification) | FDFR: 0.7825 -> 0.7863 | No clear effect |
| add ATP to MetaCloak | MetaCloak · CelebA-HQ (no purification) | ISM: 0.4911 -> 0.4048 | Improved |
| add mask guidance in pixel domain (ablation) | Anti-DB+ATP · CelebA-HQ bit-error | Bit-error (x1e-3): 349.84 -> 42.031 | Improved |
| add BDCT frequency domain with mask and Improved-PGD (ablation) | Anti-DB+ATP · CelebA-HQ bit-error | Bit-error (x1e-3): 42.031 -> 0.4688 | Improved |
| add Improved-PGD algorithm to BDCT+mask (ablation) | Anti-DB+ATP · CelebA-HQ bit-error | Bit-error (x1e-3): 81.719 -> 0.4688 | Improved |
| add BDCT without mask (ablation) | Anti-DB+ATP · CelebA-HQ bit-error | Bit-error (x1e-3): 349.84 -> 360.31 | Hurt |
| raise mask ratio from 0.5 to 0.75 (ablation) | ATP · CelebA-HQ | Bit-error (x1e-3): 0.4688 -> 0.3125 | Mixed |
| lower mask ratio from 0.5 to 0.25 (ablation) | ATP · CelebA-HQ | FDFR: 0.5213 -> 0.6400 | Mixed |
| add protection perturbation to authorization perturbation (ablation) | Authorization alone · CelebA-HQ | FDFR: 0.1075 -> 0.6338 | Improved |
| add ATP tamper-proof mechanism under purification | MetaCloak · CelebA-HQ and VGGFace2 (purification) | PSR | Improved |
Closest papers by meaning
- Edit Away and My Face Will not Stay: Personal Biometric Defense against Malicious Generative Editing (CVPR 2025 · similarity 0.86)
- Can Protective Perturbation Safeguard Personal Data from Being Exploited by Stable Diffusion? (CVPR 2024 · similarity 0.86)
- Perturb a Model, Not an Image: Towards Robust Privacy Protection via Anti-Personalized Diffusion Models (NeurIPS 2025 · similarity 0.84)
- Targeted Attack Improves Protection against Unauthorized Diffusion Customization (ICLR 2025 · similarity 0.84)
- Prompt-Agnostic Adversarial Perturbation for Customized Diffusion Models (NeurIPS 2024 · similarity 0.83)
- ImageSentinel: Protecting Visual Datasets from Unauthorized Retrieval-Augmented Image Generation (NeurIPS 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
