Pith. sign in

REVIEW 3 major objections 5 minor 76 references

Unified Steganography via Implicit Neural Representation

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read This paper proposes U-INR, a steganography scheme that stores cover and secret data of any modality in different neurons of a single implicit neural representation, with a private key regenerating the mask that locates the secret weights.

desk verdict A genuinely new key-mask mechanism and broad INR experiments, but the missing key-to-initialization specification makes the core extraction claim unreproducible as written. read the letter →

arxiv 2505.01749 v1 pith:AT2NQX4F submitted 2025-05-03 cs.CR

Qi Song , Ziyuan Luo , Xiufeng Huang , Sheng Li , Renjie Wan This is my paper · ORCID
classification cs.CR
keywords digitalsteganographydatahidingimplicitneuralrepresentationcross-modalkey-basedselectionstegamasknetworkprivacyNeRF
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

The paper sets out to show that steganography does not need to be redesigned for every media type. Its method, U-INR, represents both the cover data and the secret payload inside the weights of a single implicit neural representation (an INR), so the same embedding and extraction procedure applies to images, video, audio, signed distance fields, and NeRF scenes alike. A private key shared between sender and receiver generates a binary stega mask that picks out the secret-carrying weights; the sender first fits the secret data to those masked weights and then fits the cover data to the rest, and the receiver regenerates the same mask to read the secret out. If the method works as claimed, steganography becomes format-agnostic and no longer needs external encoders or extractors, which are the components attackers currently exploit.

What carries the argument

The central object is the stega mask $M_e$, a binary vector over the INR's weights created by what the paper calls implicit consensus. Starting from the private key $k_e$, the network is initialized deterministically; the weights are then sorted by absolute value and a threshold $t_S$ is set at the $p$-th largest weight with $p = \lfloor S \cdot N \rfloor$, marking every weight above the threshold as a secret position. This mask, used with the element-wise product $W \odot M_e$, both tells the sender which weights to fit to the secret first and tells the receiver where to look for the secret later, without any side information being transmitted. Choosing the largest-magnitude weights is what keeps the network usable after a fraction $S$ of its parameters are devoted to the payload.

What would settle it

Train U-INR twice on identical cover and secret data with the same private key $k_e$, then try to recover the secret using a mask regenerated from a fresh run with the same key; if the two runs produce different masks or the recovered secret is unusably distorted, the claim that the key alone determines the secret's position is false.

Watch

Extended reading notes

Core claim

The central claim is that hidden data can reside directly in the parameters of an INR rather than in the decoded output. U-INR derives a stega mask $M_e$ from a shared private key $k_e$: weights initialized from the key are sorted by absolute value, and the top $S \cdot 100\%$ are designated as the secret positions, with $S$ a steganographic ratio chosen by the user. The sender optimizes those masked weights against the secret data, then freezes them while the remaining weights are optimized against the cover data; the resulting stega representation is still an ordinary INR whose full inference reproduces the cover. A receiver who holds the key can regenerate $M_e$ and run inference on the masked weights alone to recover the secret. Because the same mask logic works for any data an INR can fit, the paper claims a single unified mechanism covers images, video, audio, signed distance fields, and 3D scenes while removing the extractor components that conventional methods expose to attackers.

Load-bearing premise

The load-bearing premise is that the shared private key deterministically reproduces the same network initialization on the receiver's side; if the same key can yield different initializations, the receiver cannot reconstruct the stega mask and cannot locate the secret weights.

Editorial extensions

If this is right

  • A single U-INR pipeline hides images, video, audio, signed distance fields, and NeRF scenes, since the payload lives in network weights rather than in a format-specific carrier.
  • The stega representation appears to ordinary inference as just the cover data, so an attentive receiver can only read the secret after regenerating the mask from the private key.
  • The stega ratio $S$ is a direct dial between cover fidelity and secret fidelity, with intermediate values around 30–70% giving the most balanced trade-off.
  • Under magnitude-based pruning, the secret weights survive longer than the cover weights, so the hidden payload is more robust to compression attacks than the steganographic cover itself.
  • Because no external extractor exists, there is no separate module for an attacker to corrupt or expose, which is the concrete security advantage over format-specific steganography.

Reading between the lines

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

  • Because the mask is derived from sorted weight magnitudes, the same implicit-consensus trick could be extended to embed several independent secrets at once by assigning different key-derived masks to disjoint slices of the sorted weights.
  • The security argument implicitly assumes the key space is large enough to resist brute-force mask search; a direct extension is to measure how many candidate keys an attacker must try before a wrong mask yields a recognizable secret.
  • The approach is not limited to the modalities tested: any field an INR can fit, such as a volumetric simulation or a high-dimensional sensor reading, should be embeddable by the same procedure without changing the algorithm.
  • Since the stega mask is visible in the magnitude distribution of the weights, a white-box observer could try to detect the secret by comparing weight statistics against a normally trained INR; the paper does not evaluate this detection scenario.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes U-INR, a steganography method that embeds secret data directly into the weights of an implicit neural representation (INR). The sender and receiver share a private key k_e; according to Algorithm 1, the key initializes the network weights, and the stega mask M_e is formed by thresholding the top S% of weights by absolute magnitude. Secret weights are then optimized to represent the secret data and frozen, while the remaining weights are optimized to represent the cover data. The receiver is supposed to regenerate M_e from k_e and extract the secret by evaluating the network with the complement of the mask zeroed, Eq. (11). Experiments cover images, video, audio, SDF, and NeRF, with image comparisons against HiDDeN, Baluja, HiNet, and PUSNet, plus a pruning-based robustness analysis.

Significance. If the protocol were fully specified and the security claims validated, the paper would make a useful contribution: unifying steganography across modalities through INR parameters, removing the explicit extractor as an attack surface, and reporting competitive or better image-recovery quality than existing DNN steganography methods. The cross-modality demonstrations on video, audio, SDF, and NeRF are a genuine step beyond single-modality INR steganography. However, the paper does not release code, and the central key-to-mask regeneration protocol is not specified precisely enough to reproduce, so the extraction and security claims cannot currently be verified from the manuscript alone.

major comments (3)
  1. [§4.2, Algorithm 1 line 2; §7.1] Algorithm 1 initializes weights as W_e = I(N[·], k_e), but the initialization function I is never defined anywhere in the paper. Section 7.1 states only that weights are initialized with Xavier [15] and does not say that k_e is used to seed the random generator. This is load-bearing because the receiver must regenerate exactly the same W_e to sort and threshold it into M_e; if the receiver draws a different random initialization, the positions selected by Eq. (7) and Eq. (11) will not match the sender's mask, and secret extraction fails. The paper must specify a deterministic key-to-initialization mapping (e.g., a PRNG seeded with a hash of k_e) and state how ties at the threshold t_S are broken, since strict inequality with equal magnitudes makes the mask size and mask positions implementation-dependent.
  2. [§5.4, Table 4, Fig. A.1; §4.4 security claim] The paper's own threat analysis undermines the claim that only key holders can locate the secret weights. Table 4 shows that magnitude-based pruning removes stega weights while leaving the secret representation essentially unchanged (PSNR stays at 33.86 through 20% pruning), and Fig. A.1 explicitly states that the secret-representation weights have relatively high values. An attacker who obtains the stega INR can therefore sort the weights by magnitude and recover a mask very similar to M_e without knowing k_e, then apply Eq. (11). The paper reports no steganalysis experiment, no key-less extraction attempt, and no adversarial evaluation of detectability, so the confidentiality claim in §4.4 is not supported. Please add an explicit attack model and experiments, or substantially weaken the security claims.
  3. [§5.2, Tables 2 and 3] The claimed generalization across modalities is demonstrated only by comparing the stega and secret representations with the original INR baseline, not with any existing modality-specific steganography method. For example, StegaNeRV [4] is discussed in §2 but not compared on video, and StegaNeRF-type methods are not compared on NeRF. Since the paper's headline contribution is 'unified' cross-modal steganography, the empirical support should include quantitative comparisons with representative modality-specific methods (or a clear statement that such comparisons are out of scope), otherwise the improvement claim in §1 ('advanced improvement compared to existing steganography methods') is limited to 2D images.
minor comments (5)
  1. [Abstract] The phrase 'the practice of concealing for encrypted data transmission' is not grammatical; it should likely be 'the practice of concealing secret data for encrypted transmission' or similar.
  2. [§3.0.2, Eq. (5)] The symbol S is used both for the secret data and for the stega ratio, which is confusing; please use distinct notation, e.g., m for the secret message and S for the ratio.
  3. [§4.2.1] The sentence 'users can use pre-shared Arabic numerals as the private key k_e' is unclear: a private key in a cryptographic sense is normally a bit string, and 'Arabic numerals' does not specify an encoding; please clarify what key representation is used and how it is shared.
  4. [Figure 2 caption] The caption refers to an 'implicit neural network's weight-sharing mechanism,' but the method does not use weight sharing; the text describes masking and freezing of selected weights. Please align the caption with the actual mechanism.
  5. [§5.2.5, Table 3] The audio results report MSE mean and standard deviation over 10 evaluations, but no such variance information is given for the image, video, SDF, or NeRF results; adding variance or multiple-seed reporting would strengthen the reproducibility claims.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the mask is regenerated by direct protocol, results are empirical comparisons, and self-citations are contextual rather than load-bearing.

full rationale

The paper's derivation chain is an empirical construction rather than a fitted-then-predicted loop. The stega mask M_e is generated deterministically from the private key k_e and stega ratio S by Algorithm 1; the secret weights are then optimized to fit the secret data and fixed, and the remaining weights are optimized to fit the cover data. Extraction in eq. (11) simply re-derives M_e from the same key and applies the same element-wise mask, so the extraction is a direct protocol inversion of the sender's construction, not a prediction of a quantity that was used as an input. No parameter is fit to a subset of data and then relabeled as a prediction; the reported image, video, audio, SDF, and NeRF results are measured against external baselines and against standard INR quality metrics. The self-citations [35,52] are used only to position prior INR-based steganography work as modality-limited and are not invoked as uniqueness theorems or as the justification for the central mask mechanism. The selection of 'significant' weights by magnitude is a heuristic design choice; if it is suboptimal, that would be a correctness or robustness concern, not circularity. One genuine gap exists: Algorithm 1 requires an initialization function I(N[·], k_e), but Section 7.1 states only that weights are initialized with Xavier [15] and does not specify how k_e seeds the random generator or how ties in the magnitude sort are broken. This is a reproducibility and security-completeness issue because a receiver could not regenerate the exact mask without that specification, but it is not a circular reduction of the paper's claims; a properly specified deterministic seed would make the receiver's mask an exact recomputation of the sender's mask. Overall, the central derivation is self-contained and the score is 0.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The central claim rests on three domain assumptions: the viability of magnitude-based weight splitting, the reproducibility of the key-seeded initialization, and the heuristic that large-magnitude weights can be repurposed for secret data. The stega ratio S is a free parameter set by the user, and the security model implicitly assumes key secrecy alone provides confidentiality.

free parameters (1)
  • stega ratio S = 0.3 and 0.5 in main results; 0.1 to 0.9 in trade-off study
    The fraction of weights reserved for the secret data is chosen by hand. It directly controls the quality trade-off between secret and cover representation (Fig. 7), and the paper does not provide a principled way to set it.
assumptions (3)
  • domain assumption A subset of INR weights selected by magnitude at initialization can represent secret data while the remaining weights compensate to represent cover data.
    The core mechanism assumes the masked subnetwork can fit secret data and the complement can later adapt to fit cover data without disturbing the secret. No theoretical guarantee or ablation beyond final quality metrics is provided.
  • domain assumption The receiver can deterministically reconstruct the exact weight mask from the private key.
    Algorithm 1 calls for initializing weights with I(N[·], k_e), but Section 7.1 only mentions Xavier initialization and omits how k_e seeds the random generator. This is required for extraction to work.
  • domain assumption Weight magnitude at initialization is a valid proxy for which neurons can be dedicated to secret data without destroying cover fidelity.
    The paper selects weights by absolute value at initialization, motivated by network redundancy, but this heuristic is not tested against other selection criteria such as random selection or gradient-based importance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unified Steganography via Implicit Neural Representation." pith.science (2026). https://pith.science/paper/AT2NQX4F

@misc{pith2026250501749,
  author       = {Pith},
  title        = {Pith review of: Unified Steganography via Implicit Neural Representation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AT2NQX4F}},
  note         = {Machine review of arXiv:2505.01749}
}
read the original abstract

Digital steganography is the practice of concealing for encrypted data transmission. Typically, steganography methods embed secret data into cover data to create stega data that incorporates hidden secret data. However, steganography techniques often require designing specific frameworks for each data type, which restricts their generalizability. In this paper, we present U-INR, a novel method for steganography via Implicit Neural Representation (INR). Rather than using the specific framework for each data format, we directly use the neurons of the INR network to represent the secret data and cover data across different data types. To achieve this idea, a private key is shared between the data sender and receivers. Such a private key can be used to determine the position of secret data in INR networks. To effectively leverage this key, we further introduce a key-based selection strategy that can be used to determine the position within the INRs for data storage. Comprehensive experiments across multiple data types, including images, videos, audio, and SDF and NeRF, demonstrate the generalizability and effectiveness of U-INR, emphasizing its potential for improving data security and privacy in various applications.

Figures

Figures reproduced from arXiv: 2505.01749 by the authors.

Figure 1
Figure 1. Illustration of our scenario. (a) Previous steganog [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Framework of U-INR. Our architecture establishes secure synchronization between multimedia sender and receiver [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Implicit Consensus. The initialized weight values of [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Examples of the stega and recovered images generated using different schemes. The left is the original image, and the [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Quantitative and qualitative results of our method when applying to video data. The bike video and cat video are [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: A case study on Signed distance function. Thai statue (Secret) is extracted from the representation of Room (Stega). [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Representation performance under different stega [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: the audio than Counting Ground truth 10% 30% 50% 70% 90% Ground truth 10% 30% 50% 70% 90% [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: Qualitative results of stega representation and secret representation with different stega ratios S. S denotes the ratio of parameters used for representing secret data. The residual images (→5) are located beneath each picture. TABLE III THE AUDIOS BACH AND COUNTING […

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

76 extracted references · 69 canonical work pages

  1. [15]

    Xavier Glorot and Yoshua Bengio. 2010. Understanding the difficulty of training deep feedforward neural networks. In AISTATS

    2010

  2. [4]

    Monsij Biswal, Tong Shao, Kenneth Rose, Peng Yin, and Sean Mccarthy. 2024. StegaNeRV: Video Steganography using Implicit Neural Representation. InCVPR

    work page 2024

  3. [1]

    Eirikur Agustsson and Radu Timofte. 2017. NTIRE 2017 Challenge on Single Image Super-Resolution: Dataset and Study. In CVPRW

    work page 2017

  4. [2]

    Shumeet Baluja. 2017. Hiding images in plain sight: Deep steganography. In NeurIPS

    work page 2017

  5. [3]

    Shumeet Baluja. 2020. Hiding Images within Images. TPAMI (2020)

    work page 2020

  6. [5]

    Alexia Briassouli, Panagiotis Tsakalides, and Athanasios Stouraitis. 2005. Hidden messages in heavy-tails: DCT-domain watermark detection using alpha-stable models. TMM (2005)

    work page 2005

  7. [6]

    Yambem Jina Chanu, Kh Manglem Singh, and Themrichon Tuithung. 2012. Image steganography and steganalysis: A survey. IJCA (2012)

    work page 2012

  8. [7]

    Lifeng Chen, Jia Liu, Wenquan Sun, Weina Dong, and Fuqiang Di. 2024. NeRF in NeRF: An Implicit Representation Watermark Algorithm for NeRF. Preprint (2024)

    work page 2024

Show all 76 references
  1. [8]

    Zhiqin Chen and Hao Zhang. 2019. Learning implicit fields for generative shape modeling. In CVPR

    2019

  2. [9]

    Nedeljko Cvejic and Tapio Seppanen. 2002. Increasing the capacity of LSB-based audio steganography. In IEEE MMSP

    2002

  3. [10]

    Fatiha Djebbar, Beghdad Ayad, Karim Abed Meraim, and Habib Hamam. 2012. Comparative study of digital audio steganography techniques. Eurasip. J. Audio Spee. (2012)

    2012

  4. [11]

    Weina Dong, Jia Liu, Lifeng Chen, Wenquan Sun, Xiaozhong Pan, and Yan Ke

  5. [12]

    Emilien Dupont, Adam Goliński, Milad Alizadeh, Yee Whye Teh, and Arnaud Doucet. 2021. Coin: Compression with implicit neural representations. InICLRW

    2021

  6. [13]

    Jonathan Frankle, Gintare Karolina Dziugaite, Daniel M Roy, and Michael Carbin

  7. [14]

    S Geetha, S Subburam, S Selvakumar, Seifedine Kadry, and Robertas Damasevicius

  8. [16]

    Song Han, Jeff Pool, John Tran, and William Dally. 2015. Learning both weights and connections for efficient neural network. NeurIPS (2015)

    2015

  9. [17]

    Stefan Hetzl and Petra Mutzel. 2005. A graph–theoretic approach to steganogra- phy. In CMS

    2005

  10. [18]

    Dongdong Hou, Weiming Zhang, Kejiang Chen, Sian-Jheng Lin, and Nenghai Yu

  11. [19]

    Shoko Imaizumi and Kei Ozawa. 2014. Multibit embedding algorithm for steganog- raphy of palette-based images. In PSIVT

    2014

  12. [20]

    Jun Jia, Zhongpai Gao, Dandan Zhu, Xiongkuo Min, Menghan Hu, and Guangtao Zhai. 2022. RIVIE: Robust inherent video information embedding. TMM (2022)

    2022

  13. [21]

    Yue Jiang, Kejiang Chen, Wei Yan, Xuehu Yan, Guozheng Yang, and Kai Zeng

  14. [22]

    Junpeng Jing, Xin Deng, Mai Xu, Jianyi Wang, and Zhenyu Guan. 2021. HiNet: Deep Image Hiding by Invertible Network. In CVPR

    2021

  15. [23]

    Inas Jawad Kadhim, Prashan Premaratne, Peter James Vial, and Brendan Halloran

  16. [24]

    Jaeho Lee, Jihoon Tack, Namhoon Lee, and Jinwoo Shin. 2021. Meta-learning sparse implicit neural representations. In NeurIPS

    2021

  17. [25]

    Bin Li, Ming Wang, Jiwu Huang, and Xiaolong Li. 2014. A new cost function for spatial image steganography. In ICIP

    2014

  18. [26]

    TMM (2024)

    Robust Secret Image Sharing Resistant to JPEG Recompression Based on Stable Block Condition. TMM (2024)

    2024

  19. [27]

    Guobiao Li, Sheng Li, Meiling Li, Zhenxing Qian, and Xinpeng Zhang. 2023. Towards deep network steganography: from networks to networks.arXiv preprint arXiv:2307.03444 (2023)

    2023 arXiv

  20. [28]

    Guobiao Li, Sheng Li, Meiling Li, Xinpeng Zhang, and Zhenxing Qian. 2023. Steganography of steganographic networks. In AAAI

    2023

  21. [29]

    Guobiao Li, Sheng Li, Zicong Luo, Zhenxing Qian, and Xinpeng Zhang. 2024. Purified and Unified Steganographic Network. In CVPR

    2024

  22. [30]

    Guobiao Li, Sheng Li, Zhenxing Qian, and Xinpeng Zhang. 2024. Cover-separable Fixed Neural Network Steganography via Deep Generative Models. InACM MM

    2024

  23. [31]

    Ruiqi Li and Yiu-ming Cheung. 2024. Variational Multi-scale Representation for Estimating Uncertainty in 3D Gaussian Splatting. NeurIPS (2024)

    2024

  24. [32]

    Chenxin Li, Brandon Y Feng, Zhiwen Fan, Panwang Pan, and Zhangyang Wang

  25. [33]

    Jia Liu, Peng Luo, and Yan Ke. 2023. Hiding functions within functions: Steganog- raphy by implicit neural representations. arXiv preprint arXiv:2312.04743 (2023)

    2023 arXiv

  26. [34]

    Qin Liu, Jiamin Yang, Hongbo Jiang, Jie Wu, Tao Peng, Tian Wang, and Guojun Wang. 2022. When deep learning meets steganography: Protecting inference privacy in the dark. In INFOCOM

    2022

  27. [35]

    Ziyuan Luo, Qing Guo, Ka Chun Cheung, Simon See, and Renjie Wan. 2023. CopyrNeRF: Protecting the copyright of neural radiance fields. In ICCV

    2023

  28. [36]

    Zicong Luo, Sheng Li, Guobiao Li, Zhenxing Qian, and Xinpeng Zhang. 2023. Securing Fixed Neural Network Steganography. In ACM MM

    2023

  29. [37]

    Srinivasan, Rodrigo Ortiz-Cayon, Nima Khademi Kalan- tari, Ravi Ramamoorthi, Ren Ng, and Abhishek Kar

    Ben Mildenhall, Pratul P. Srinivasan, Rodrigo Ortiz-Cayon, Nima Khademi Kalan- tari, Ravi Ramamoorthi, Ren Ng, and Abhishek Kar. 2019. Local Light Field Fusion: Practical View Synthesis with Prescriptive Sampling Guidelines. TOG (2019)

    2019

  30. [38]

    Srinivasan, Matthew Tancik, Jonathan T

    Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, and Ren Ng. 2020. NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis. In ECCV

    2020

  31. [39]

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. 2014. Microsoft coco: Common objects in context. In ECCV

    2014

  32. [40]

    Chong Mou, Youmin Xu, Jiechong Song, Chen Zhao, Bernard Ghanem, and Jian Zhang. 2023. Large-capacity and flexible video steganography via invertible neural network. In CVPR

    2023

  33. [41]

    Ramadhan J Mstafa, Younis Mohammed Younis, Haval Ismael Hussein, and Muhsin Atto. 2020. A new video steganography scheme based on Shi-Tomasi corner detector. IEEE Access (2020)

    2020

  34. [42]

    Bui Cong Nguyen, Sang Moon Yoon, and Heung-Kyu Lee. 2006. Multi bit plane image steganography. In IWDW

    2006

  35. [43]

    Michael Niemeyer, Lars Mescheder, Michael Oechsle, and Andreas Geiger. 2019. Occupancy flow: 4D reconstruction by learning particle dynamics. In ICCV

    2019

  36. [44]

    Michiharu Niimi, Hideki Noda, Eiji Kawaguchi, and Richard O Eason. 2002. High capacity and secure digital steganography to palette-based images. In ICIP

    2002

  37. [45]

    Feng Pan, Jun Li, and Xiaoyuan Yang. 2011. Image steganography method based on PVD and modulus function. In ICECC

    2011

  38. [46]

    Tayana Morkel, Jan HP Eloff, and Martin S Olivier. 2005. An overview of image steganography.. In ISSA

    2005

  39. [47]

    Niels Provos and Peter Honeyman. 2003. Hide and seek: An introduction to steganography. IEEE S & P (2003)

    2003

  40. [48]

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, et al

  41. [49]

    Haichao Shi, Jing Dong, Wei Wang, Yinlong Qian, and Xiaoyu Zhang. 2018. SSGAN: Secure steganography based on generative adversarial networks. In PCM

    2018

  42. [50]

    Yun-Qing Shi, Xiaolong Li, Xinpeng Zhang, Hao-Tian Wu, and Bin Ma. 2016. Reversible data hiding: Advances in the past two decades. IEEE access (2016)

    2016

  43. [51]

    Vincent Sitzmann, Julien Martel, Alexander Bergman, David Lindell, and Gor- don Wetzstein. 2020. Implicit neural representations with periodic activation functions. In NeurIPS

    2020

  44. [52]

    Qi Song, Ziyuan Luo, Ka Chun Cheung, Simon See, and Renjie Wan. 2024. Protect- ing NeRFs’ Copyright via Plug-And-Play Watermarking Base Model. In ECCV

    2024

  45. [53]

    Tomáš Pevn`y, Tomáš Filler, and Patrick Bas. 2010. Using high-dimensional image models to perform highly undetectable steganography. In IH

    2010

  46. [54]

    Weixuan Tang, Bin Li, Shunquan Tan, Mauro Barni, and Jiwu Huang. 2019. CNN- based adversarial embedding for image steganography. TIFS (2019)

    2019

  47. [55]

    Weixuan Tang, Shunquan Tan, Bin Li, and Jiwu Huang. 2017. Automatic stegano- graphic distortion learning using a generative adversarial network. Singal Proc. Let. (2017)

    2017

  48. [56]

    Jinyuan Tao, Sheng Li, Xinpeng Zhang, and Zichi Wang. 2018. Towards robust image steganography. TCSVT (2018)

    2018

  49. [57]

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. 2004. Image quality assessment: from error visibility to structural similarity. TIP (2004)

    2004

  50. [58]

    Xinyu Weng, Yongzhi Li, Lu Chi, and Yadong Mu. 2019. High-capacity convolu- tional video steganography with temporal residual modeling. In ICMR

    2019

  51. [59]

    Jianhua Yang, Danyang Ruan, Jiwu Huang, Xiangui Kang, and Yun-Qing Shi

  52. [60]

    Zijin Yang, Kejiang Chen, Kai Zeng, Weiming Zhang, and Nenghai Yu. 2023. Provably secure robust image steganography. TMM (2023)

    2023

  53. [61]

    Sojeong Song, Seoyun Yang, Chang D Yoo, and Junmo Kim. 2024. Implicit Steganography Beyond the Constraints of Modality. In ECCV

    2024

  54. [62]

    Xinpeng Zhang. 2011. Reversible data hiding in encrypted image. IEEE Sign. process. letters (2011)

    2011

  55. [63]

    Xiang Zhang, Fei Peng, and Min Long. 2018. Robust coverless image steganogra- phy based on DCT and LDA topic classification. TMM (2018)

    2018

  56. [64]

    Xinpeng Zhang and Shuozhong Wang. 2006. Efficient steganographic embedding by exploiting modification direction. IEEE Comm. letters (2006)

    2006

  57. [65]

    Jiren Zhu, Russell Kaplan, Justin Johnson, and Li Fei-Fei. 2018. Hidden: Hiding data with deep networks. In ECCV. 9

    2018

  58. [66]

    stega” and “secret representation

    Zhiying Zhu, Sheng Li, Zhenxing Qian, and Xinpeng Zhang. 2021. Destroying robust steganography in online social networks. Inform. Sciences (2021). 7 Appendix 7.1 Implementation details In our experiment, we implement SIREN 1 [51] for image, video, audio, SDF data. We implement...

    2021

  59. [68]

    TIFS (2019)

    An embedding cost learning framework using GAN. TIFS (2019)

    2019

  60. [70]

    Kevin Alex Zhang, Alfredo Cuesta-Infante, Lei Xu, and Kalyan Veeramachaneni

  61. [71]

    arXiv preprint arXiv:1901.03892 (2019)

    SteganoGAN: High capacity image steganography with GANs. arXiv preprint arXiv:1901.03892 (2019)

    2019 arXiv

  62. [2015]

    IJCV (2015)

    Imagenet large scale visual recognition challenge. IJCV (2015)

    2015

  63. [2018]

    TCSVT (2018)

    Reversible data hiding in color image with grayscale invariance. TCSVT (2018)

    2018

  64. [2019]

    Neurocomputing (2019)

    Comprehensive survey of image steganography: Techniques, Evaluations, and trends in future research. Neurocomputing (2019)

    2019

  65. [2020]

    Pruning neural networks at initialization: Why are we missing the mark?. In ICLR

  66. [2021]

    Steganogram removal using multidirectional diffusion in fourier domain while preserving perceptual image quality. Pattern. Recogn. Lett. (2021)

    2021

  67. [2023]

    StegaNeRF: Embedding invisible information within neural radiance fields. In ICCV

  68. [2024]

    Implicit Neural Representation Steganography by Neuron Pruning.Preprint (2024)

    2024

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.