Pith. sign in

REVIEW 7 cited by

TrustMark: Universal Watermarking for Arbitrary Resolution Images

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2311.18297 v1 pith:FVFG376Q submitted 2023-11-30 cs.CV cs.AI

TrustMark: Universal Watermarking for Arbitrary Resolution Images

classification cs.CV cs.AI
keywords watermarkingarbitraryimageimagesmethodresolutiontrustmarkwatermark
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Imperceptible digital watermarking is important in copyright protection, misinformation prevention, and responsible generative AI. We propose TrustMark - a GAN-based watermarking method with novel design in architecture and spatio-spectra losses to balance the trade-off between watermarked image quality with the watermark recovery accuracy. Our model is trained with robustness in mind, withstanding various in- and out-place perturbations on the encoded image. Additionally, we introduce TrustMark-RM - a watermark remover method useful for re-watermarking. Our methods achieve state-of-art performance on 3 benchmarks comprising arbitrary resolution images.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 7 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Secure Seed-Based Multi-bit Watermarking for Diffusion Models from First Principles

    cs.CR 2026-05 unverdicted novelty 7.0

    A theoretical framework decouples diffusion model generation from watermark decisions, enabling SSB to reach any security-robustness-fidelity regime without model-specific empirical tests.

  2. ResGuard: Enhancing Robustness Against Known Original Attacks in Deep Watermarking

    cs.CV 2026-04 accept novelty 6.5

    ResGuard makes deep-watermark residuals image-specific via a contrastive loss and KOA noise layer, restoring near-perfect extraction under residual-subtraction attacks that previously collapsed accuracy to chance.

  3. Are Watermarked Images Editable? SafeMark for Watermark-Preserving Text-Guided Image Editing

    cs.CV 2026-05 unverdicted novelty 6.0

    SafeMark integrates a thresholded watermark-decoding loss into diffusion editors to enable text-guided edits that preserve embedded watermarks with high bit accuracy.

  4. DeepSignature: Digitally Signed, Content-Encoding Watermarks for Robust and Transparent Image Authentication

    cs.CR 2026-04 unverdicted novelty 6.0

    DeepSignature embeds digitally signed content-encoding watermarks via neural networks for robust image authentication, source attribution, and latent-space tamper localization.

  5. Guidance Watermarking for Diffusion Models

    cs.CR 2025-09 unverdicted novelty 6.0

    Guidance watermarking steers diffusion denoising steps via gradients from an off-the-shelf watermark decoder to embed marks during generation, converting post-hoc schemes into in-generation ones while remaining comple...

  6. LoT-Pass: Long-term-robust Image Watermarking for Image to Video Generation

    cs.CV 2025-09 conditional novelty 6.0

    I2VWM uses video-like training distortions and optical-flow frame alignment to keep image watermarks decodable in AI-generated videos made from that image.

  7. SPFM-Net: Semantic-Prior-Guided Frequency-Constrained Mamba for Invisible Watermark Attack

    cs.CV 2026-07 reject novelty 5.0

    SPFM-Net attacks invisible watermarks by reconstructing heavily masked images with a pretrained semantic model; it reports good removal on Stable Signature/HiDDeN but poor removal on Yu.