Research map / Image generation / Text-to-image diffusion
Adapter Shield: A Unified Framework with Built-in Authentication for Preventing Unauthorized Zero-Shot Image-to-Image Generation
Jun Jia, Hongyi Miao, Yingjie Zhou, Wangqiu Zhou, Jianbo Zhang, Linhan Cao, Dandan Zhu, Hua Yang, Xiongkuo Min, Wei Sun, Guangtao Zhai
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 |
|---|---|---|---|
| Adapter Shield embedding-targeted protection instead of fine-tuning-oriented defenses (Pretender, Mist, ACE, CAAT) | IP-Adapter FaceID · FFHQ 200 | ISM: 0.8691 (Pretender) -> 0.0514 | Improved |
| Adapter Shield on artwork protection | IP-Adapter · Artwork anti-plagiarism | ESM: 0.7920 (ACE) -> 0.0161 | Improved |
| general (two-encoder) version instead of specific single-encoder version (ablation) | Adapter Shield · Face identity, IP-Adapter FaceID | ISM / PSNR: ISM 0.0514 -> 0.1422; PSNR 30.01 -> 32.10 | Mixed |
| add multi-target loss Lmt (two encoders) (ablation) | Adapter Shield · Face protection, unseen encoder | Unseen similarity: 0.2946 -> 0.1054 | Improved |
| add multi-target loss Lmt (two encoders) (ablation) | Adapter Shield · Artwork protection, specific encoder | Specific similarity: 0.1276 -> 0.1944 | Hurt |
| add diversity losses Ldiv and Ldiv_s (ablation) | Adapter Shield encryptor/decryptor · Facial identity | Encrypt diversity (cos sim): 0.9849 -> 0.0232 | Improved |
| add diversity losses Ldiv and Ldiv_s (ablation) | Adapter Shield · Facial identity | Wrong decryption rate: 0.0 -> 0.015 (artwork 0.0 -> 0.05) | Hurt |
| add same-password diversity loss Ldiv_s (ablation) | Adapter Shield · Facial identity | Same-password diversity (cos sim): 0.9958 -> 0.0361 | Improved |
| JPEG compression applied to protected images | Adapter Shield · Robustness, IP-Adapter | embedding cos sim to original: 0.0107 -> 0.6870 | Hurt |
Closest papers by meaning
- IDProtector: An Adversarial Noise Encoder to Protect Against ID-Preserving Image Generation (CVPR 2025 · similarity 0.88)
- Edit Away and My Face Will not Stay: Personal Biometric Defense against Malicious Generative Editing (CVPR 2025 · similarity 0.85)
- DiffusionGuard: A Robust Defense Against Malicious Diffusion-based Image Editing (ICLR 2025 · similarity 0.85)
- Robust Secure Swap: Responsible Face Swap With Persons of Interest Redaction and Provenance Traceability (ICML 2025 · similarity 0.85)
- Enhancing Facial Privacy Protection via Weakening Diffusion Purification (CVPR 2025 · similarity 0.84)
- CGI-DM: Digital Copyright Authentication for Diffusion Models via Contrasting Gradient Inversion (CVPR 2024 · similarity 0.84)
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.
