AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Hiding Images in Deep Probabilistic Models

Haoyu Chen, Linqi Song, Zhenxing Qian, Xinpeng Zhang, Kede Ma

NeurIPS 2022 · Image generation › Autoregressive and latent generation · approach: Robust diffusion watermarking · 4 citations (OpenAlex)

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 changedWhereMeasuredEffect
Use probabilistic SinGAN hiding rather than autoencoder-based image-in-image methodsImage hiding · 200 test image pairsPSNR: 35.64 -> 36.84Improved
Use probabilistic SinGAN hiding rather than autoencoder-based image-in-image methodsImage hiding · 200 test image pairsSSIM: 0.952 -> 0.958Improved
Use probabilistic SinGAN hiding rather than autoencoder-based image-in-image methodsImage hiding · 200 test image pairsDISTS: 0.047 -> 0.038Improved
Hide more secret images in the same SinGAN (ablation)SinGAN image hiding · Multiple image hidingPSNR: 36.84 -> 35.91 -> 34.93 -> 34.03Hurt
Hide more secret images in the same SinGAN (ablation)SinGAN image hiding · Multiple image hidingSSIM: 0.958 -> 0.946 -> 0.935 -> 0.923Hurt
Hide more secret images in the same SinGAN (ablation)SinGAN image hiding · Multiple image hidingDISTS: 0.038 -> 0.043 -> 0.049 -> 0.055Hurt
Pixel-shuffle the secret image before hiding (ablation)SinGAN image hiding · Image obfuscationPSNR: 36.84 -> 20.53Hurt
Pixel-shuffle the secret image before hiding (ablation)SinGAN image hiding · Image obfuscationSSIM: 0.958 -> 0.726Hurt
Pixel-shuffle the secret image before hiding (ablation)SinGAN image hiding · Image obfuscationDISTS: 0.038 -> 0.172Hurt
Replace cover-image reconstruction with secret-image reconstruction during training (ablation)SinGAN · SinGAN training and testingExtraction accuracyNo clear effect
Hide secret images while preserving the cover patch distributionStego SinGAN · Cover generation with hidden imagesSIFID: 0.041 -> 0.046No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy