Research map / Image generation / Autoregressive and latent generation
Generating steganographic images via adversarial training
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use adversarially trained Alice for embedding instead of HUGO, WOW, or S-UNIWARD | Steganographic images · BOSS, 0.4bpp | steganalyzer accuracy: ATS: 0.66 -> 0.83 vs HUGO; 0.75 -> 0.83 vs WOW; 0.77 -> 0.83 vs S-UNIWARD. Eve: 0.59 -> 0.79 vs HUGO; 0.74 -> 0.79 vs WO | Mixed |
| Use adversarially trained Alice for embedding instead of HUGO, WOW, or S-UNIWARD | Steganographic images · CelebA, 0.4bpp | steganalyzer accuracy: ATS: 0.94 -> 0.95 vs HUGO; 0.89 -> 0.95 vs WOW; 0.91 -> 0.95 vs S-UNIWARD. Eve: 0.89 -> 0.90 vs HUGO; 0.85 -> 0.90 vs WO | Mixed |
| Train Eve with supervised data | Eve steganalyzer · BOSS and CelebA, 0.4bpp | steganalyzer accuracy: BOSS ATS/Eve: Alice 0.83/0.79, HUGO 0.66/0.59, WOW 0.75/0.74, S-UNIWARD 0.77/0.72; CelebA ATS/Eve: Alice 0.95/0.90, HUGO | Mixed |
| Increase the embedded message rate from 0.1bpp to 0.4bpp | Alice and Bob · CelebA | message decoding success: 100% at 0.1bpp; 100% at 0.4bpp | No clear effect |
| Train and transfer models between workstations rather than decode with models on the training machine | Alice and Bob · CelebA, 0.1bpp, 100 trials | message recovery: 100% -> 99.1% | Hurt |
Closest papers by meaning
- Neural Cover Selection for Image Steganography (NeurIPS 2024 · similarity 0.87)
- StegoGAN: Leveraging Steganography for Non-Bijective Image-to-Image Translation (CVPR 2024 · similarity 0.86)
- Focus on Hiders: Exploring Hidden Threats for Enhancing Adversarial Training (CVPR 2024 · similarity 0.86)
- Hiding Images in Diffusion Models by Editing Learned Score Functions (CVPR 2025 · similarity 0.86)
- Watermark-embedded Adversarial Examples for Copyright Protection against Diffusion Models (CVPR 2024 · similarity 0.85)
- StegaStyleGAN: Towards Generic and Practical Generative Image Steganography (AAAI 2024 · 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.