Research map / Image generation / Image restoration
Deep Robust Reversible Watermarking
Jiale Chen, Wei William Wang, Chongyang Shi, Li Dong, Yuanman Li, Xiping Hu
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 overflow penalty loss (lambda_p 0 -> 10^2) (ablation) | DRRW · Stego quality | PSNR (dB): 29.6822 -> 31.6585 | Improved |
| increase penalty weight to 10^4 (ablation) | DRRW · Stego quality | PSNR (dB): 31.6585 -> 34.1974 | Improved |
| increase penalty weight to 10^6 (ablation) | DRRW · Stego quality | PSNR (dB): 29.6822 -> 40.9473 | Improved |
| add penalty loss (ablation) | DRRW · Stego quality | SSIM: 0.9942 -> 0.9995 | Improved |
| add penalty loss (ablation) | DRRW · Robustness at max distortion | extraction accuracy: over 10% average gain | Improved |
| increase penalty weight (ablation) | DRRW · Latent z bits | bitstream length: below 400 bits at 10^6 | Hurt |
| increase penalty weight (ablation) | DRRW · Overflow map bits | bitstream length: decreases significantly | Improved |
| add regularization loss on latent z (ablation) | DRRW · Gray test images | z bitstream length: 11,292.83 -> 1,046.03 | Improved |
| add regularization loss on latent z (ablation) | DRRW · Gray test images | perfect recovery failure rate: 46.123% -> 0% | Improved |
| integer flow (DRRW) vs real-valued flow (R-DRRW) | DRRW · Irreversible comparison | PSNR (dB): 33.9180 -> 40.9473 | Improved |
| integer flow vs real-valued flow | DRRW · Median filter w=19 | extraction accuracy: 85.12 -> 90.01 | Improved |
| DRRW vs PZM | embedding speed · Embedding time | seconds per image: 248.65 -> 4.51 | Improved |
Closest papers by meaning
- END^2: Robust Dual-Decoder Watermarking Framework Against Non-Differentiable Distortions (AAAI 2025 · similarity 0.87)
- Lightweight-Mark: Rethinking Deep Learning-Based Watermarking (ICML 2025 · similarity 0.86)
- Robust Watermarking Using Generative Priors Against Image Editing: From Benchmarking to Advances (ICLR 2025 · similarity 0.85)
- An Undetectable Watermark for Generative Image Models (ICLR 2025 · similarity 0.85)
- Decoder Gradient Shield: Provable and High-Fidelity Prevention of Gradient-Based Box-Free Watermark Removal (CVPR 2025 · similarity 0.85)
- T2SMark: Balancing Robustness and Diversity in Noise-as-Watermark for Diffusion Models (NeurIPS 2025 · similarity 0.85)
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.
