Pith. sign in

REVIEW 5 major objections 5 minor 50 references

Peccavi: Visual Paraphrase Attack Safe and Distortion Free Image Watermarking Technique for AI-Generated Images

T0 review · 5 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read PECCAVI claims the first distortion-free watermark that survives visual paraphrase attacks, by hiding in paraphrase-stable image regions.

desk verdict NMP placement is a genuinely new idea, but the circular evaluation and overclaims mean the paper needs major revision before it can support its headline claims. read the letter →

arxiv 2506.22960 v1 pith:VM5FQZWD submitted 2025-06-28 cs.CV

Shreyas Dixit , Ashhar Aziz , Shashwat Bajpai , Vasu Sharma , Aman Chadha , Vinija Jain , Amitava Das This is my paper · ORCID
classification cs.CV
keywords imagewatermarkingAI-generatedimagesvisualparaphraseattackNon-MeltingPointssaliencydetectionfrequency-domainwatermarkprobabilitycontentprovenance
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Visual paraphrase attacks regenerate an image with an image-to-image diffusion model, keeping the meaning but changing pixels enough to erase ordinary watermarks. This paper claims the first watermark that survives such attacks without visible distortion: PECCAVI finds the regions that stay put across five paraphrases, calls them Non-Melting Points, and embeds a multi-channel frequency-domain watermark only there, adding random extra patches and noise to hide the location. On 100 COCO images, it reports watermark detection probability 0.90 at paraphrase strength 0.1 and 0.85 at 0.2, above ZoDiac's 0.81/0.70 and WAM's 0.63/0.56, at PSNR around 29.8 and SSIM 0.93. If true, AI-generated images could carry provenance through mild edits and semantic rewrites that defeat current watermarks.

What carries the argument

The central object is the Non-Melting Point (NMP): a salient image region whose location survives visual paraphrase, found by intersecting saliency maps of five paraphrased copies. The watermark is carried by multi-channel Fourier-space ring patterns placed inside NMP patches; the ring spacing (channel strength) sets resistance, and the number of paraphrases in which a region appears sets the strength. Two security mechanisms protect the embedding: random patching plants an extra watermark patch at a random non-overlapping location, and noisy burnishing adds adversarial noise that blinds saliency detectors trying to find the NMPs. Adaptive enhancement blends the watermarked image with the original to minimize distortion, choosing the smallest blending factor $\gamma$ that keeps SSIM above a target.

What would settle it

Generate a PECCAVI watermarked image, paraphrase it at strength 0.5 or with a different image-to-image model than the one used to define the Non-Melting Points, and measure watermark detection probability; if detection drops below the paper's reported range while SSIM stays near 0.93, then Non-Melting Point stability does not generalize. A second test: remove or perturb only the randomly patched watermark regions and check whether detection collapses, which would show random patching, not Non-Melting Point embedding, is carrying the robustness.

Watch

Extended reading notes

Core claim

The paper's central claim is that robustness to visual paraphrase attacks and freedom from visible distortion are compatible for image watermarking, contrary to the brittleness of existing methods. The argument is constructive: a watermark should be placed precisely in the semantic core that a paraphrase preserves, not spread across the whole image. PECCAVI identifies Non-Melting Points by generating five paraphrases, computing saliency maps, and keeping the regions with highest Intersection over Union across paraphrases, then embeds Fourier-domain ring watermarks in multiple channels of those patches, with strength set by stability score ($W_s = \max(0.1, 1 - 0.25(n-1))$). Random patching and noisy burnishing are added so an attacker who reverse-engineers the saliency method cannot simply excise the watermark. The paper reports that this design beats ZoDiac and WAM on watermark detection probability after paraphrase attacks at strengths 0.1 and 0.2 while keeping SSIM at 0.93.

Load-bearing premise

The load-bearing premise is that the Non-Melting Points found from five paraphrases made with the authors' own image-to-image pipeline will also stay put when an adversary paraphrases with a stronger or different model; if those regions melt, the watermark goes with them.

Editorial extensions

If this is right

  • A deployed PECCAVI detector can scan all patches and channels and take the highest score, so it needs no prior knowledge of where the watermark was placed.
  • Classical post-processing—brightness change, Gaussian noise, JPEG compression at quality 50—leaves watermark detection probability near 0.96–0.99 in the paper's measurements, so routine social-media re-encoding should not erase the watermark.
  • Because NMP detection uses saliency methods, the approach works across different text-to-image generators, including SD 3, SDXL, SD 2.1, DALL-E 3, and Midjourney 6 images evaluated on MS-COCO.
  • At paraphrase strengths 0.1 and 0.2, PECCAVI retains watermark detection probability of 0.90 and 0.85 with the best saliency configuration, a margin over ZoDiac and WAM that the paper attributes to embedding inside paraphrase-stable regions.
  • The random patch and noisy burnishing are meant to stop an attacker from locating the embedded regions, so the method's security does not rely on the watermark placement being secret.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper tests only paraphrase strengths 0.1 and 0.2; an untested implication is that at higher strengths or with a different image-to-image model, the Non-Melting Points may shift and the reported detection margins could shrink.
  • An adversary who reproduces the same paraphrase pipeline offline might learn where the stable regions are; the random patch adds uncertainty, but the paper does not quantify how much entropy that uncertainty provides.
  • The same Non-Melting Point idea could be extended to video watermarking by looking for patches stable across frames rather than across paraphrases, though the paper does not propose this.
  • If NMP stability is intrinsic to semantic content, similar stable-region detection could also improve the robustness of other image provenance markers beyond watermarking, but this is an editorial extrapolation.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The manuscript proposes PECCAVI, a watermarking method for AI-generated images designed to survive visual paraphrase attacks. The method detects Non-Melting Points (NMPs)—regions stable across five paraphrases generated with the Barman et al. (2024) pipeline—embeds watermarks in these regions using multi-channel Fourier-domain watermarking, adds random patching and noisy burnishing to hinder reverse engineering, and applies adaptive enhancement to reduce distortion. The authors report results on 100 COCO images, claiming PECCAVI is the first visual-paraphrase-attack-safe, distortion-free watermarking technique and that it outperforms ZoDiac and WAM under brightness, Gaussian noise, JPEG compression, and two paraphrase strengths. The paper also states that resources and code will be open-sourced.

Significance. The problem is timely and important: generative-AI image provenance and resistance to paraphrase-based de-watermarking are active concerns for policy and practice. The core idea of embedding watermarks in regions that are stable under visual paraphrasing is intuitive and worth investigating, and the paper provides useful ablations over saliency methods (XRAI, MSI-Net, Vanilla Integrated) and top-k patch choices. The authors are also explicit that PECCAVI is model-agnostic and that code will be open-sourced, which are commendable commitments. However, the current validation does not establish the headline claims: the NMP selection and the attack evaluation share the same paraphrase pipeline, the 'distortion-free' claim is contradicted by the reported PSNR/SSIM values, and statistical reliability is not demonstrated. If the evaluation were broadened and the claims appropriately qualified, the method could be a meaningful contribution.

major comments (5)
  1. [§3.1 and §4.4] The evaluation is potentially circular. Section 3.1 defines Non-Melting Points as regions stable across five paraphrases generated with the Barman et al. (2024) pipeline, and Section 4.4 evaluates robustness using the same pipeline (stable-diffusion-xl-base-1.0, caption-conditioned, strengths 0.1 and 0.2). The watermark is therefore placed in regions that are stable under exactly the paraphrase distribution used for testing. The claim that PECCAVI is 'visual paraphrase attack-safe' requires evidence that NMPs remain stable under stronger paraphrases (s > 0.2), under different image-to-image models, under unseen prompts and seeds, and under the open-ended visual paraphrase variant described in Section 2.1. Please add held-out attack tests that do not share the NMP-selection pipeline.
  2. [Table 1 and §5] The 'distortion-free' claim is not supported by the reported numbers. Table 1 shows PECCAVI PSNR values of 29.56–31.50 dB and SSIM 0.93–0.95, whereas WAM reports PSNR 46.05 and SSIM 1.00, and even ZoDiac reports SSIM 0.92. A PSNR around 30 dB indicates visible distortion, not a distortion-free result. The abstract and Section 5 should either remove the 'distortion-free' claim or qualify it with a perceptual threshold; additionally, Section 3.5's adaptive enhancement procedure should specify how the target SSIM s* and the blending parameter gamma are chosen, since these are free parameters that directly affect the reported distortion.
  3. [§4.4 and Table 1] The statistical basis of the comparison is too weak. All results are averages over 100 images, but no error bars, confidence intervals, or significance tests are reported. Differences such as PECCAVI (XRAI, Top-50) WDP 0.85 vs. ZoDiac 0.70 at paraphrase strength s=0.2 may be meaningful, but without variance estimates the claim of superiority is not established. The paper should also define Watermark Detection Probability (WDP) explicitly, including the detection threshold and false-positive rate, since a higher WDP at an unstated threshold can be misleading.
  4. [§3.2, §3.3, and Table 1] The two security mechanisms—random patching and noisy burnishing—are central to the claimed robustness against reverse engineering, but they are not evaluated. Table 1 contains no attack scenario that attempts to locate or distort NMPs, and no experiment measures how noisy burnishing affects image quality versus watermark detectability. To support the security claims, the authors should include an adversarial evaluation, for example re-running saliency detection on the watermarked image and deleting or distorting the detected salient patches, then reporting WDP.
  5. [References and missing appendices] There are serious completeness and citation issues. The 'ZoDiac' baseline is cited inconsistently: Section 2 and Table 1 cite Zhou et al. (2024), but the reference list entry for Zhou et al. (2024) is 'Zodiac: A cardiologist-level LLM framework for multi-agent diagnostics', which is not an image-watermarking paper, while Section 3.2 cites 'ZoDiac (Zhang et al., 2024)' with a different author. In addition, the manuscript contains multiple 'Figure ??' placeholders and references an appendix ('Appendix: Visual Paraphrase', 'Appendix: Distortion vs. Detectability') that is not present. These must be corrected for the paper to be reproducible and for the baseline comparison to be credible.
minor comments (5)
  1. [Throughout] The notation 'PECCA VI' and 'PECCAVI' are used inconsistently; please unify the spelling.
  2. [Table 1] The column labeled 'λ' is not explained anywhere in the text; please either define it or remove it.
  3. [§3.1] The stability score is described as 'lower scores indicate greater consistency', which is counterintuitive; please define the score formula explicitly and explain the direction of the measure.
  4. [Footnote 3] The statement that all resources and codes will be open-sourced conflicts with the patent footnote stating that 'adoption or research involving this technology for non-profit purposes is strictly prohibited'; please clarify the intended availability.
  5. [§4.1] The description says PECCAVI was tested across SD3, SDXL, SD2.1, DALL-E 3, and Midjourney 6, but Table 1 only reports aggregate results on 'MS COCO AI' without per-model breakdowns; please indicate where per-model results are shown.

Circularity Check

1 steps flagged · score 6.0 of 10

NMPs are selected and evaluated under the same Barman et al. (2024) paraphrase pipeline, making the headline 'paraphrase attack-safe' an in-distribution claim rather than an independent test.

  1. fitted input called prediction [Section 3.1 (NMP selection) and Section 4.4 (paraphrase attack evaluation)]
    "Section 3.1: 'For a given image, we generate five automatically paraphrased versions using the method described in (Barman et al., 2024)... These stable areas are referred to as Non-Melting Points (NMPs).' Section 4.4: 'Visual Paraphrasing (Barman et al., 2024), using stable-diffusion-xl-base-1.0 with image captions and paraphrase strengths of 0.1 and 0.2.'"

    NMPs are chosen as the regions with highest IoU stability across five paraphrases produced by the Barman et al. (2024) SDXL-based visual paraphrase pipeline. The paper then measures 'paraphrase attack safety' by attacking with the same pipeline (stable-diffusion-xl-base-1.0, caption-conditioned, strengths 0.1 and 0.2). The embedding locations are thus fit to the very generative distribution used as the test; the reported WDP shows the watermark survives fresh draws from the same paraphrase generator that defined the NMPs, not that it survives visual paraphrase attacks in general. No independent adversary, out-of-distribution model, or stronger strength is tested, and Barman et al. (2024) shares most authors with this paper.

full rationale

The paper's central claim is that PECCAVI is 'the first visual paraphrase attack-safe, distortion-free image watermarking technique.' The load-bearing design choice is to embed watermarks in Non-Melting Points, defined as regions stable across five paraphrases from Barman et al. (2024). Section 4.4 then evaluates attack safety using the same Barman et al. pipeline at strengths 0.1 and 0.2. This is a same-family evaluation: the NMP locations are fit to the same paraphrase distribution on which they are tested, so the reported paraphrase robustness is in-sample rather than a demonstration against independent or stronger visual paraphrase attacks. This is not a by-construction tautology, because watermark detectability after paraphrasing is still an empirical outcome, but it is a partial circularity: the 'prediction' of paraphrase attack safety is not independent of the fitting procedure. Separately, the 'distortion-free' description is in tension with the reported PSNR values (29.6-29.9 for PECCAVI XRAI vs 46.05 for WAM), though this is a consistency concern rather than circularity. The self-citation to Barman et al. (2024), whose authors overlap with the present paper, is load-bearing because it supplies both the NMP definition and the attack evaluation, but the paper does include some external benchmark comparisons (ZoDiac, WAM), so the circularity is partial. Score 6 reflects that the central attack-safety result is evaluated on the same pipeline used to choose the embedding regions.

Assumptions & free parameters 6 free parameters · 5 assumptions · 1 invented entities

The central method rests on the NMP construct, a hand-chosen strength schedule, unspecified paraphrase-generation parameters, and a 100-image sample. All of these are assumed rather than derived or independently validated.

free parameters (6)
  • Watermark strength schedule constants = Ws = max(0.1, 1 - 0.25*(n-1))
    Hand-chosen formula mapping NMP frequency n to ring spacing; no derivation or sensitivity analysis (Section 3.2).
  • Number of paraphrases for NMP detection = 5
    Chosen without justification; stability estimates depend on this (Section 3.1).
  • Paraphrase strength(s) for NMP generation = not specified
    The paper does not state at what strength the five paraphrases are generated, yet this determines which regions are called NMPs.
  • Top-K patches (Top 30/40/50) = 30/40/50
    Hyperparameter scanned in Table 1; no selection criterion or variance reported.
  • Adaptive enhancement target SSIM s* = not specified
    Equation in Section 3.5 needs gamma search with threshold s*; actual threshold and search method absent.
  • Noisy burnishing noise budget = not specified
    No noise magnitude or perturbation bound is given, though the method claims to preserve the frequency watermark.
assumptions (5)
  • domain assumption NMP stability axiom: regions stable across 5 paraphrases remain stable under later paraphrase attacks.
    Section 3.1 defines NMPs from five paraphrases; Section 4.2 tests only paraphrase strengths 0.1 and 0.2, so this stability across the broader attack space is assumed.
  • domain assumption Representativeness of the authors' Barman et al. (2024) paraphrase pipeline for all visual paraphrase attacks.
    Section 2.1 adopts Barman et al. as the attack definition, and the evaluation uses the same procedure; no independent adversary or cross-model paraphrase attack is considered.
  • domain assumption XRAI saliency maps align with watermark-preserving semantic regions.
    Section 3.1 states 'empirically, we found that XRAI performed best', but no quantitative criterion or comparison is provided for this choice.
  • domain assumption MS COCO 100-image sample is representative.
    Section 4.4 reports results on 100 randomly sampled COCO images without stratification or discussion of generalizability.
  • domain assumption Brute-force patch scanning yields a reliable watermark detection probability without a false-positive baseline.
    Section 3.4 describes brute-force detection, but no false-positive rate or threshold analysis is reported.
invented entities (1)
  • Non-Melting Points (NMPs)
    purpose: Regions presumed stable under visual paraphrase, used as watermark embedding locations
    Defined operationally via IoU over paraphrases from the authors' own attack; no external validation that NMPs survive unseen paraphrase attacks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Peccavi: Visual Paraphrase Attack Safe and Distortion Free Image Watermarking Technique for AI-Generated Images." pith.science (2026). https://pith.science/paper/VM5FQZWD

@misc{pith2026250622960,
  author       = {Pith},
  title        = {Pith review of: Peccavi: Visual Paraphrase Attack Safe and Distortion Free Image Watermarking Technique for AI-Generated Images},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VM5FQZWD}},
  note         = {Machine review of arXiv:2506.22960}
}
read the original abstract

A report by the European Union Law Enforcement Agency predicts that by 2026, up to 90 percent of online content could be synthetically generated, raising concerns among policymakers, who cautioned that "Generative AI could act as a force multiplier for political disinformation. The combined effect of generative text, images, videos, and audio may surpass the influence of any single modality." In response, California's Bill AB 3211 mandates the watermarking of AI-generated images, videos, and audio. However, concerns remain regarding the vulnerability of invisible watermarking techniques to tampering and the potential for malicious actors to bypass them entirely. Generative AI-powered de-watermarking attacks, especially the newly introduced visual paraphrase attack, have shown an ability to fully remove watermarks, resulting in a paraphrase of the original image. This paper introduces PECCAVI, the first visual paraphrase attack-safe and distortion-free image watermarking technique. In visual paraphrase attacks, an image is altered while preserving its core semantic regions, termed Non-Melting Points (NMPs). PECCAVI strategically embeds watermarks within these NMPs and employs multi-channel frequency domain watermarking. It also incorporates noisy burnishing to counter reverse-engineering efforts aimed at locating NMPs to disrupt the embedded watermark, thereby enhancing durability. PECCAVI is model-agnostic. All relevant resources and codes will be open-sourced.

Figures

Figures reproduced from arXiv: 2506.22960 by the authors.

Figure 1
Figure 1. Some examples of the negative impacts of AI-generated images: (a) [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Illustration of multipatch watermarking, where watermark patterns are embedded in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Noisy burnishing disrupts saliency detection in watermarked images, hindering attackers [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: The PECCAVI pipeline for image watermarking encompasses NMP detection, multi [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Comparison of two sets of images before and after adaptive enhancement, which optimizes [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: The cumulative impact of successive paraphrasing attacks on PECCAVI (green) and ZoDiac [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 43 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    Stable diffusion 2.1, 2023

    Stability AI. Stable diffusion 2.1, 2023. Available at: https://huggingface.co/stabilityai/stable-diffusion-2-1

    work page 2023

  3. [3]

    Stable diffusion 3: Research paper

    Stability AI. Stable diffusion 3: Research paper. 2024

    work page 2024

  4. [4]

    Moinuddin, and E

    Md Amir Baig, Athar A. Moinuddin, and E. Khan. Psnr of highest distortion region: An effective image quality assessment method. In 2019 International Conference on Electrical, Electronics and Computer Engineering (UPCON), pages 1--4, 2019

    work page 2019

  5. [5]

    Variational image compression with a scale hyperprior, 2018

    Johannes Ballé, David Minnen, Saurabh Singh, Sung Jin Hwang, and Nick Johnston. Variational image compression with a scale hyperprior, 2018

    work page 2018

  6. [6]

    The brittleness of ai-generated image watermarking techniques: Examining their robustness against visual paraphrasing attacks, 2024

    Niyar R Barman, Krish Sharma, Ashhar Aziz, Shashwat Bajpai, Shwetangshu Biswas, Vasu Sharma, Vinija Jain, Aman Chadha, Amit Sheth, and Amitava Das. The brittleness of ai-generated image watermarking techniques: Examining their robustness against visual paraphrasing attacks, 2024

    work page 2024

  7. [7]

    Improving image generation with better captions

    James Betker, Gabriel Goh, Li Jing, † TimBrooks, Jianfeng Wang, Linjie Li, † LongOuyang, † JuntangZhuang, † JoyceLee, † YufeiGuo, † WesamManassra, † PrafullaDhariwal, † CaseyChu, † YunxinJiao, and Aditya Ramesh. Improving image generation with better captions

    work page

  8. [8]

    Openai mulls watermarking chatgpt generated text, but treads with caution

    Business Standard . Openai mulls watermarking chatgpt generated text, but treads with caution. Business Standard, 2024

    work page 2024

Show all 50 references
  1. [9]

    Ab-3211 california digital content provenance standards

    california legislature. Ab-3211 california digital content provenance standards. 2023

    2023

  2. [10]

    Counter T uring test ( CT 2): AI -generated text detection is not as easy as you may think - introducing AI detectability index ( ADI )

    Megha Chakraborty, S.M Towhidul Islam Tonmoy, S M Mehedi Zaman, Shreya Gautam, Tanay Kumar, Krish Sharma, Niyar Barman, Chandan Gupta, Vinija Jain, Aman Chadha, Amit Sheth, and Amitava Das. Counter T uring test ( CT 2): AI -generated text detection is not as easy as you may th...

    2023

  3. [11]

    Learned image compression with discretized gaussian mixture likelihoods and attention modules, 2020

    Zhengxue Cheng, Heming Sun, Masaru Takeuchi, and Jiro Katto. Learned image compression with discretized gaussian mixture likelihoods and attention modules, 2020

    2020

  4. [12]

    Bill gates says this is the no

    CNBC. Bill gates says this is the no. 1 unsolvable problem facing today’s young people: ‘the harm is done’. 2024

    2024

  5. [13]

    Eu code of conduct against online hate speech: latest evaluation shows slowdown in progress

    European Commission. Eu code of conduct against online hate speech: latest evaluation shows slowdown in progress. 2022

    2022

  6. [14]

    Secure spread spectrum watermarking for multimedia

    Ingemar J Cox, Joe Kilian, F Thomson Leighton, and Talal Shamoon. Secure spread spectrum watermarking for multimedia. IEEE Transactions on Image Processing, 6 0 (12): 0 1673--1687, 1997

    1997

  7. [15]

    Scaling rectified flow transformers for high-resolution image synthesis, 2024

    Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek, and Robin Rombach. Scaling rectified flow transf...

    2024

  8. [16]

    Facing reality?: Law enforcement and the challenge of deepfakes

    EUROPOL. Facing reality?: Law enforcement and the challenge of deepfakes. 2022

    2022

  9. [17]

    The stable signature: Rooting watermarks in latent diffusion models, 2023

    Pierre Fernandez, Guillaume Couairon, Hervé Jégou, Matthijs Douze, and Teddy Furon. The stable signature: Rooting watermarks in latent diffusion models, 2023

    2023

  10. [18]

    Pause giant ai experiments: An open letter, 2023

    Future of Life Institute . Pause giant ai experiments: An open letter, 2023

    2023

  11. [19]

    Can you spot the chameleon? adversarially camouflaging images from co-salient object detection

    Ruijun Gao, Qing Guo, Felix Juefei-Xu, Hongkai Yu, Huazhu Fu, Wei Feng, Yang Liu, and Song Wang. Can you spot the chameleon? adversarially camouflaging images from co-salient object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition...

    2022

  12. [20]

    Fourier image watermarking: Print-cam application

    Khadija Gourrame, Frederic Ros, Hassan Douzi, Rachid Harba, and Rabia Riad. Fourier image watermarking: Print-cam application. Electronics, 11 0 (2): 0 266, 2022

    2022

  13. [21]

    Godfather of ai’ geoffrey hinton quits google and warns over dangers of misinformation

    The Guardian. Godfather of ai’ geoffrey hinton quits google and warns over dangers of misinformation. 2023

    2023

  14. [22]

    The rapid rise of generative ai: Assessing risks to safety and security, 2023

    Ardi Janjeva, Alexander Harris, Sarah Mercer, Alexander Kasprzyk, and Anna Gausen. The rapid rise of generative ai: Assessing risks to safety and security, 2023

    2023

  15. [23]

    Xrai: Better attributions through regions

    Andrei Kapishnikov, Tolga Bolukbasi, Fernanda Viegas, and Michael Terry. Xrai: Better attributions through regions. In 2019 IEEE/CVF International Conference on Computer Vision (ICCV), pages 4947--4956, 2019

    2019

  16. [24]

    Google ceo sundar pichai: Fears about artificial intelligence are very legitimate, he says in post interview

    Arjun Kharpal. Google ceo sundar pichai: Fears about artificial intelligence are very legitimate, he says in post interview. The Washington Post, 2018

    2018

  17. [25]

    A watermark for large language models

    John Kirchenbauer, Jonas Geiping, Yuxin Wen, Jonathan Katz, Ian Miers, and Tom Goldstein. A watermark for large language models. In International Conference on Machine Learning, 2023

    2023

  18. [26]

    Contextual encoder-decoder network for visual saliency prediction

    Alexander Kroner, Mario Senden, Kurt Driessens, and Rainer Goebel. Contextual encoder-decoder network for visual saliency prediction. Neural Networks, 129: 0 261--270, 2020

    2020

  19. [27]

    Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Doll \'a r, and C

    Tsung-Yi Lin, Michael Maire, Serge J. Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Doll \'a r, and C. Lawrence Zitnick. Microsoft coco: Common objects in context. In European Conference on Computer Vision, 2014

    2014

  20. [28]

    Francis G. Loch. Image processing algorithms part 4: Brightness adjustment. The Crypt Mag, 2010

    2010

  21. [29]

    Towards blind watermarking: Combining invertible and non-invertible mechanisms

    Rui Ma, Mengxi Guo, Yi Hou, Fan Yang, Yuan Li, Huizhu Jia, and Xiaodong Xie. Towards blind watermarking: Combining invertible and non-invertible mechanisms. In Proceedings of the 30th ACM International Conference on Multimedia, pages 1532--1542, 2022

    2022

  22. [30]

    Midjourney: A new era of ai-generated art, 2024

    Midjourney. Midjourney: A new era of ai-generated art, 2024

    2024

  23. [31]

    Navas et al

    A. Navas et al. Digital watermarking techniques: A review. Journal of Information Technology, 23 0 (4): 0 345--360, 2008

    2008

  24. [32]

    Diffusion models for adversarial purification

    Weili Nie, Brandon Guo, Yujia Huang, Chaowei Xiao, Arash Vahdat, and Animashree Anandkumar. Diffusion models for adversarial purification. In Proceedings of the 39th International Conference on Machine Learning, pages 16805--16827. PMLR, 2022

    2022

  25. [33]

    Sdxl: Improving latent diffusion models for high-resolution image synthesis, 2023

    Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna, and Robin Rombach. Sdxl: Improving latent diffusion models for high-resolution image synthesis, 2023

    2023

  26. [34]

    Podilchuk and K

    C. Podilchuk and K. Ramchandran. Image watermarking: Capacity issues and applications. In Proceedings of the International Conference on Image Processing (ICIP), pages 445--448. IEEE, 1998

    1998

  27. [35]

    What is the max seed you can set up?, 2022

    PyTorch Discussion Forum . What is the max seed you can set up?, 2022. Accessed: 2024-11-13

    2022

  28. [36]

    Zero-shot text-to-image generation, 2021

    Aditya Ramesh, Mikhail Pavlov, Gabriel Goh, Scott Gray, Chelsea Voss, Alec Radford, Mark Chen, and Ilya Sutskever. Zero-shot text-to-image generation, 2021

    2021

  29. [37]

    Hierarchical text-conditional image generation with clip latents, 2022

    Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. Hierarchical text-conditional image generation with clip latents, 2022

    2022

  30. [38]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj\"orn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10684--10695, 2022

    2022

  31. [39]

    Can ai-generated text be reliably detected?, 2024

    Vinu Sankar Sadasivan, Aounon Kumar, Sriram Balasubramanian, Wenxiao Wang, and Soheil Feizi. Can ai-generated text be reliably detected?, 2024

    2024

  32. [40]

    Sara Mahdavi, Rapha Gontijo Lopes, Tim Salimans, Jonathan Ho, David J Fleet, and Mohammad Norouzi

    Chitwan Saharia, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily Denton, Seyed Kamyar Seyed Ghasemipour, Burcu Karagol Ayan, S. Sara Mahdavi, Rapha Gontijo Lopes, Tim Salimans, Jonathan Ho, David J Fleet, and Mohammad Norouzi. Photorealistic text-to-image diffusion mod...

    2022

  33. [41]

    Watermark anything with localized messages, 2024

    Tom Sander, Pierre Fernandez, Alain Durmus, Teddy Furon, and Matthijs Douze. Watermark anything with localized messages, 2024

    2024

  34. [42]

    Stegastamp: Invisible hyperlinks in physical photographs

    Matthew Tancik, Ben Mildenhall, and Ren Ng. Stegastamp: Invisible hyperlinks in physical photographs. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2020

    2020

  35. [43]

    Thomson, Daniel Angus, and Paula Dootson

    T.J. Thomson, Daniel Angus, and Paula Dootson. 3.2 billion images and 720,000 hours of video are shared online daily: Can you sort real from fake? The Conversation, 2020. Accessed: 2024-11-04

    2020

  36. [44]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: from error visibility to structural similarity. IEEE transactions on image processing, 13 0 (4): 0 600--612, 2004

    2004

  37. [45]

    Tree-rings watermarks: Invisible fingerprints for diffusion images

    Yuxin Wen, John Kirchenbauer, Jonas Geiping, and Tom Goldstein. Tree-rings watermarks: Invisible fingerprints for diffusion images. In Thirty-seventh Conference on Neural Information Processing Systems, 2023

    2023

  38. [46]

    Yadav et al

    Rajesh K. Yadav et al. Study and analysis of wavelet based image compression techniques. International Journal of Engineering, Science and Technology, 4 0 (1): 0 1--7, 2012

    2012

  39. [47]

    Attack-resilient image watermarking using stable diffusion, 2024

    Lijun Zhang, Xiao Liu, Antoni Viros Martin, Cindy Xiong Bearfield, Yuriy Brun, and Hui Guan. Attack-resilient image watermarking using stable diffusion, 2024

    2024

  40. [48]

    Generative autoencoders as watermark attackers: Analyses of vulnerabilities and threats

    Xuandong Zhao, Kexun Zhang, Yu-Xiang Wang, and Lei Li. Generative autoencoders as watermark attackers: Analyses of vulnerabilities and threats. arXiv preprint arXiv:2306.01953, 2023

    2023 arXiv

  41. [49]

    Invisible image watermarks are provably removable using generative ai, 2024

    Xuandong Zhao, Kexun Zhang, Zihao Su, Saastha Vasan, Ilya Grishchenko, Christopher Kruegel, Giovanni Vigna, Yu-Xiang Wang, and Lei Li. Invisible image watermarks are provably removable using generative ai, 2024

    2024

  42. [50]

    Zodiac: A cardiologist-level llm framework for multi-agent diagnostics, 2024

    Yuan Zhou, Peng Zhang, Mengya Song, Alice Zheng, Yiwen Lu, Zhiheng Liu, Yong Chen, and Zhaohan Xi. Zodiac: A cardiologist-level llm framework for multi-agent diagnostics, 2024

    2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.