Research map / Image generation / Autoregressive and latent generation
SemBind: Binding Diffusion Watermarks to Semantics Against Black-Box Forgery Attacks
Xin Zhang, Zijin Yang, Kejiang Chen, Linfeng Ma, Weiming Zhang, Nenghai Yu
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 |
|---|---|---|---|
| bind watermark to semantics with SemBind | Tree-Ring · Imprint attack SDP, SD 2.1 attacker | detection rate: 1.00 -> 0.01 | Improved |
| bind watermark to semantics with SemBind | Gaussian Shading · Imprint attack SDP, SD 2.1 attacker | bit accuracy: 0.9998 -> 0.4818 | Improved |
| bind watermark to semantics with SemBind | PRC · Imprint attack SDP, SD 2.1 attacker | detection rate: 1.00 -> 0.06 | Improved |
| bind watermark to semantics with SemBind | Gaussian Shading++ · Imprint attack SDP, SD 2.1 attacker, step 150 | detection rate: 1.00 -> 0.49 | Improved |
| use mismatched SD 1.5 attacker against SemBind | SemBind · Imprint attack | detection rate: PRC-S 0.06 -> 0.00 at step 50 vs SD 2.1 | Improved |
| bind watermark to semantics with SemBind | Gaussian Shading · Reprompt attack SDP, SD 2.1 attacker | detection rate: 1.00 -> 0.60 | Improved |
| bind watermark to semantics with SemBind | PRC · Reprompt attack COCO, SD 1.5 attacker | detection rate: 0.93 -> 0.11 | Improved |
| add SemBind to Tree-Ring | Tree-Ring · image quality | FID: 25.43 -> 25.27 | Improved |
| add SemBind to provably undetectable schemes | GS, PRC, GS++ · image quality | FID and CLIP | No clear effect |
| add SemBind to PRC | PRC · Robustness SDP (average over distortions) | detection rate: 0.995 -> 0.940 | Hurt |
| add SemBind to Gaussian Shading++ | Gaussian Shading++ · Robustness SDP (average over distortions) | detection rate: 0.973 -> 0.932 | Hurt |
| add SemBind to Tree-Ring | Tree-Ring · Robustness SDP (average over distortions) | detection rate: 0.997 -> 0.992 | No clear effect |
Closest papers by meaning
- Black-Box Forgery Attacks on Semantic Watermarks for Diffusion Models (CVPR 2025 · similarity 0.92)
- StableGuard: Towards Unified Copyright Protection and Tamper Localization in Latent Diffusion Models (NeurIPS 2025 · similarity 0.88)
- SleeperMark: Towards Robust Watermark against Fine-Tuning Text-to-image Diffusion Models (CVPR 2025 · similarity 0.88)
- WMAdapter: Adding WaterMark Control to Latent Diffusion Models (ICML 2025 · similarity 0.87)
- WMCopier: Forging Invisible Watermarks on Arbitrary Images (NeurIPS 2025 · similarity 0.87)
- PGID: Progressive Guided Inversion and Denoising for Robust Watermark Detection (NeurIPS 2026 · similarity 0.87)
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.
