AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Your Text Encoder Can Be An Object-Level Watermarking Controller

Naresh Kumar Devulapally, Mingzhen Huang, Vishal Asnani, Shruti Agarwal, Siwei Lyu, Vishnu Suresh Lokhande

ICCV 2025 · Image generation › Autoregressive and latent generation · approach: Robust diffusion watermarking · 2 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 text-token watermark embedding instead of AquaLoRA UNet fine-tuning (same detector)AquaLoRA · WikiArt/MS-COCO (48 bits)PSNR: +7 dB PSNR, >20% bit accuracyImproved
plug watermark token into SD with Textual Inversion personalization (ablation)Ours + SD · WikiArt (48 bits)Bit accuracy under DiffPure: 0.88 -> 0.86Hurt
plug watermark token into styled SD UNet (ablation)Ours + SD · WikiArt (48 bits)Bit accuracy under DiffPure: 0.88 -> 0.84Hurt
choose timestep tau*=8 for training instead of large timesteps (ablation)Ours · T2I watermarkingimage quality vs bit accuracyMixed
add latent matching loss L_z to the watermark BCE loss (ablation)Ours · T2I watermarkingimperceptibilityImproved
use SAM segmentation masks instead of attention maps for localization (ablation)Ours · Object-level watermarkingwatermark localization accuracyImproved
increase message length to 128 bits (ablation)Ours · bit-length ablationbit accuracy: > 91% at 128 bitsNo clear effect
add object overlap of at least 40% between two watermarked objects (ablation)Ours · Object-level watermarkingbit accuracy (no attack): 0.94 -> 0.79Hurt
shrink watermarked object crop from 0.8x to 0.4x size (ablation)Ours · Object-level watermarkingbit accuracy (no attack): 0.96 -> 0.90Hurt
increase number of watermarked objects from 1 to 3 (no overlap) (ablation)Ours · Object-level watermarkingbit accuracy (no attack): 0.99 -> 0.90Hurt
use style background instead of white background for segmented object (ablation)Ours · Object-level watermarkingbit accuracy (no attack): 0.99 -> 0.95Hurt
in-generation token watermark instead of post-generation HiDDeNHiDDeN · WikiArtJPEG bit accuracy: 0.88 -> 0.96Improved

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