Research map / Adversarial robustness and security / Attacks on text and graphs
Deep Model Intellectual Property Protection via Deep Watermarking
Jie Zhang, Dongdong Chen, Jing Liao, Weiming Zhang, Huamin Feng, Gang Hua, 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 |
|---|---|---|---|
| add adversarial training stage for the extractor (ablation) | Debone-IEEE watermark · Debone surrogate attack (UNet, L2) | SR_NC: 27% -> 100% | Improved |
| add adversarial training stage for the extractor (ablation) | Debone-IEEE watermark · Debone surrogate attack (CNet, L2) | SR_NC: 0% -> 93% | Improved |
| add adversarial training stage for the extractor (ablation) | Derain-Flower watermark · Derain surrogate attack (UNet, L2+L_adv) | SR_NC: 0% -> 99% | Improved |
| add clean loss on watermark-free images (ablation) | extractor R · watermark-free images | false extraction | Improved |
| add watermark consistent loss (ablation) | extractor R · surrogate outputs | extraction quality | Improved |
| add overwriting-aware enhanced training with re-watermark noise layer (ablation) | extractor R · Watermark overwriting (Flower) | SR_NC: 0% -> 100% | Improved |
| increase surrogate CNet channels from 8 to 256 (ablation) | Debone watermark extractor · Debone CNet surrogate | SR_NC: 3% -> 99% | Improved |
| increase lambda from 0.1 to 10 (ablation) | Debone-IEEE · Debone-IEEE | PSNR: 49.82 -> 43.21 | Hurt |
| vary watermark size from 32 to 256 (ablation) | Debone-IEEE · Debone-IEEE | PSNR: 46.89 -> 47.29 | No clear effect |
| embed multiple watermarks in one network (ablation) | Debone_IEEE · Debone-IEEE | PSNR: 47.76 -> 41.87 | Hurt |
| replace spatial invisible watermarking with deep invisible watermark network | TSIW-64 · Debone surrogate attack (CNet) | SR_NC: 0% -> 93% | Improved |
| make target model self-watermarked instead of adding barrier | Derain model · Derain | PSNR: 32.49 -> 32.13 | No clear effect |
Closest papers by meaning
- Model Watermarking for Image Processing Networks (AAAI 2020 · similarity 0.95)
- IPRemover: A Generative Model Inversion Attack against Deep Neural Network Fingerprinting and Watermarking (AAAI 2024 · similarity 0.89)
- Lightweight-Mark: Rethinking Deep Learning-Based Watermarking (ICML 2025 · similarity 0.89)
- Safe and Robust Watermark Injection with a Single OoD Image (ICLR 2024 · similarity 0.88)
- Revisiting the Information Capacity of Neural Network Watermarks: Upper Bound Estimation and Beyond (AAAI 2024 · similarity 0.87)
- SleeperMark: Towards Robust Watermark against Fine-Tuning Text-to-image Diffusion Models (CVPR 2025 · similarity 0.86)
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.
