Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

Fragile Watermarking for Image Certification Using Deep Steganographic Embedding

T0 review · 3 major / 6 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 A hidden logo inside a certified passport photo makes any modification visible and lets a classifier identify the manipulation type.

desk verdict Honest feasibility study showing deep stego embedding can leave classifiable artifacts in recovered markers, but the morphing test only covers post-issuance tampering, not the real pre-enrollment attack. read the letter →

arxiv 2504.13759 v1 pith:XL2JRZQG submitted 2025-04-18 cs.CV cs.LG

Davide Ghiani , Jefferson David Rodriguez Chivata , Stefano Lilliu , Simone Maurizio La Cava , Marco Micheletto , Giulia Orrù , Federico Lama , Gian Luca Marcialis This is my paper · ORCID
classification cs.CVcs.LG
keywords fragilewatermarkingdeepsteganographyimagecertificationmorphingattackdetectionbiometricdocumentintegritymanipulationclassificationICAO-compliantfacialimages
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 proposes certifying ICAO-compliant facial images by hiding a small known image—the ICAO logo—inside the photo with a deep steganographic encoder at issuance. At verification a decoder recovers the marker; an untouched photo yields a clean marker, while compression, resizing, blur, sharpening, noise, or morphing leaves systematic artifacts in the recovered logo. The authors show these artifacts are measurable and visually consistent across two different steganographic models, and that a ResNet-50 classifier trained on the recovered marker can identify the manipulation type with accuracy above 99.9% in the same-embedding protocols and around 72–81% when training and test use different embedding models. If correct, official identity photos could carry a proactive integrity check that both detects post-issuance tampering and names the kind of alteration applied.

What carries the argument

The central object is the recovered hidden marker, the decoded secret image obtained as $D(T(E(I_C, I_S)))$, where $E$ is the steganographic embedding of the ICAO logo $I_S$ into the face photo $I_C$, $T$ is a post-issuance transformation, and $D$ is the steganographic decoder. The argument is carried by the deliberate fragility of these deep decoders, which are trained to reconstruct the secret only when the cover is essentially unchanged, so any transformation of the stego-image injects manipulation-specific artifacts into the revealed logo. The second mechanism is the classification pairing: ResNet-50 features of the recovered marker, followed by dimensionality reduction, ReLU, dropout, and a final linear layer, map those artifacts to one of seven manipulation classes, with protocols that vary whether training and test use the same embedding model and whether all or only some manipulation strengths are seen in training.

What would settle it

Create a morphed face first, then embed the ICAO logo into the already-morphed image and decode it; if the recovered marker is clean and the classifier reports no manipulation, the morphing cue is specific to post-issuance attacks rather than the canonical pre-issuance morphing scenario.

Watch

Extended reading notes

Core claim

The central claim is that standard deep steganographic embedding, repurposed as a fragile watermark, makes any post-issuance modification of a certified facial image visible in the recovered hidden marker, and that the degradation pattern is systematic enough to classify the manipulation. Using Stegformer and SteGuz to embed a 224×224 ICAO logo into each face keeps both the certified photo and the recovered marker high-fidelity, with SSIM above 0.92 in both stages. JPEG and WebP compression, resizing, Gaussian and salt-and-pepper noise, Gaussian and median blur, sharpening, and morphing each degrade the marker in reproducible ways, with morphing and salt-and-pepper noise pushing SSIM below 0.5 in several cases. The revealed-marker artifacts are then fed to a ResNet-50 with fully connected layers, which separates seven manipulation classes with high accuracy in intra-steganography protocols and meaningful transfer in cross-steganography settings.

Load-bearing premise

The method assumes the watermark is embedded at issuance and that all tampering happens afterward; if an attacker morphed the face before the marker was embedded, the recovered logo would show no morphing artifact.

Editorial extensions

If this is right

  • An intact recovered marker would certify that the photo has not been altered after issuance, turning integrity verification into a simple comparison against the expected marker.
  • A degraded marker would not only flag tampering but also indicate the manipulation family, giving forensic examiners a starting point for further analysis.
  • The quality thresholds reported in the paper, such as SSIM below 0.75 or PSNR below 22 dB, could serve as a cheap pre-filter before running the classifier.
  • Cross-steganography accuracy of 72–81% suggests the forensic signal generalizes across embedding architectures well enough to support interoperability between certification systems.
  • Training on six of eight manipulation strengths and testing on all eight still yields reliable classification, indicating that the artifact patterns are not merely memorized from discrete parameter values.

Reading between the lines

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

  • The paper's threat-model order—embed first, attack second—is not the canonical ePassport morphing order, in which the morph is created before enrollment; the authors' own protocol may therefore overstate the method's value against the headline morphing attack.
  • Because the ICAO logo is a known public image, the marker's fragility does not depend on secrecy, but an informed attacker could attempt to re-embed a clean marker into a modified photo; a per-document keyed marker would close that gap.
  • A natural testable extension is to see whether the same fragility signal appears when the attack is applied to the cover before embedding, and whether the classifier can be trained to detect morph-before-embedding cases at all.
  • The recovered-marker quality metrics could be turned into an interpretable tamper score for border officers, independent of the neural classifier, since the measured SSIM and PSNR drops are consistent per manipulation family.
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 / 6 minor

Summary. The paper proposes a fragile watermarking framework for ICAO-compliant facial images, where a known marker (the ICAO logo) is embedded using deep steganographic models (Steguz and Stegformer). Any post-issuance modification of the stego-image is hypothesized to degrade the recovered marker, and the degradation pattern is used both as a tamper-detection cue and as input to a ResNet-50 classifier that categorizes the manipulation type (compression, resize, blur, Gaussian noise, salt-and-pepper noise, sharpening, morphing). The authors report high intra-stega classification accuracy (≥99.95%), a substantial drop in cross-stega scenarios (~25%), and define SSIM/PSNR thresholds for flagging manipulated images. The central claim is that this approach detects and categorizes post-issuance manipulations, including morphing attacks.

Significance. If the central claim held, the paper would be a useful feasibility study showing that off-the-shelf deep steganographic embedding can serve as a fragile watermark for biometric document certification, with the added benefit of manipulation-type classification. The experimental design has notable strengths: a held-out identity split that avoids identity-level overfitting, two diverse steganographic architectures, systematic parameter sweeps for most manipulations, and cross-stega generalization tests. The paper is also commendably explicit about its scope limitations in Section III. However, the morphing threat model is mismatched with the standard ePassport morphing attack, and the detection component (as opposed to classification among known manipulations) is not actually evaluated with a pristine class or false-alarm analysis. These issues bear directly on the headline claims, so the current evidence does not support the paper's stated conclusions for morphing detection.

major comments (3)
  1. [Section IV-C, Eq. (2)] The morphing experiments apply FaceMorpher to the already-certified stego-image Istego, i.e., after embedding. In the standard ePassport morphing attack, the morph is created before enrollment and the authority embeds the marker into an already-morphed face; in that timeline the morph is part of the cover image and the decoder recovers the same clean ICAO logo as from an un-morphed image, so the proposed fragile cue and the classifier cannot detect the attack. This is not a minor detail: the abstract and Introduction explicitly cite morphing as a key threat, and Section III's own limitation statement admits that a missing or mismatched marker requires a separate detection stage that is out of scope. Therefore the paper's claim of morphing-attack detection is unsupported for the standard attack model; the authors should either test pre-issuance morphing (embedding into an already-morphed image) or narrow the claimed contribution to post-issuance modifications.
  2. [Section V-A and Section IV-E] The operational thresholds (SSIM < 0.75, PSNR < 22 dB) are read off the unaltered certified samples used in the same evaluation, and the classifier is trained and tested on seven manipulation classes with no pristine 'no manipulation' class. As a result, the paper demonstrates classification among pre-known manipulation categories but does not provide evidence of binary tamper detection (e.g., false-alarm rate on unaltered stego-images, ROC analysis, or threshold tuning on a validation split). This is load-bearing because the proposed mechanism is described as an integrity-verification tool that detects 'any modification'; without a false-alarm analysis and an unaltered test class, the detection claim is not supported.
  3. [Section IV-C, Table I] The morphing class is evaluated at a single blending factor α_M = 0.9, and the classifier is never tested on other morphing strengths or on morphing algorithms other than FaceMorpher. Since the morphing cue is the paper's headline forensic signal, the lack of any generalization test for this manipulation class leaves the robustness claim for morphing detection essentially unexamined. A minimal addition would be a second morphing tool or a sweep over α_M values, with the classifier trained on held-out identities and unseen morphing parameters.
minor comments (6)
  1. [General / Title] The paper title promises 'Image Certification' but the method assumes the marker is already present and matching; the authors should consider adding 'Post-Issuance' or 'Post-Embedding' to the title or abstract to align with the actual scope.
  2. [Section V-A, Table II] The row label 'Certifyng' is misspelled; it should be 'Certifying'.
  3. [Figure 3 caption] The caption says 'salt & paper noise' but the correct term is 'salt-and-pepper noise' (also in Figure 4).
  4. [Section IV-E] The sentence 'The first one, called P8-8, involves training and testing on the same eight variations of each manipulation type (e.g., levels of noise or compression)' has a double comma and should be rephrased for clarity.
  5. [Section VI] The conclusions section contains an exact duplicate sentence: 'To our knowledge, this is the first study to assess the use of standard deep steganographic models for this purpose in the context of biometric documents.' and 'To our knowledge, this is the first study to assess the use of standard deep steganographic models for this purpose in the context of document integrity.' The duplicate should be removed.
  6. [Section III] The claim that 'any modification to the stego-image ... inevitably affects the extracted content' is stated as a certainty, but the paper provides only empirical evidence for the tested manipulations; a softer phrasing such as 'is expected to affect' would be more accurate.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the empirical pipeline is self-contained and no derived claim reduces to its inputs.

full rationale

The paper's derivation chain is empirical rather than analytic. A known marker (ICAO logo) is embedded with external steganographic models (Stegformer, SteGuz), manipulations are applied to the stego-image, and the recovered marker is analyzed with image-quality metrics and a ResNet-50 classifier. No equation or result is defined in terms of the quantity it purports to predict. The classifier is trained on held-out identities, and the P6-8 protocol tests on unseen manipulation strengths, so the reported accuracies are not forced by construction. The only threshold heuristic (SSIM < 0.75, PSNR < 22 dB) is calibrated on unaltered samples and presented as an operational rule, not as a measured detection result, so it does not constitute a fitted input renamed as a prediction. The morphing experiment applies morphing after embedding, whereas standard pre-enrollment morphing would occur before embedding; this is a threat-model limitation acknowledged in Section III, not a circular step. Self-citations ([1], [10]) are contextual and not load-bearing. Overall, no claim in the paper reduces to its own inputs.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The empirical claim rests on these assumptions and hand-chosen values rather than on invented entities. The free parameters are the two operational thresholds (data-derived) and the morph alpha (hand-chosen). The axioms cover dataset compliance, post-embedding usability, marker availability, and the morphing threat model.

free parameters (3)
  • SSIM tamper threshold = 0.75
    Section V-A defines images with SSIM below 0.75 as potentially manipulated; the threshold is read from the observed unaltered certified case, with no held-out validation.
  • PSNR tamper threshold = 22 dB
    Section V-A defines images with PSNR below 22 dB as potentially manipulated, with the same lack of validation.
  • Morphing blending factor alpha_M = 0.9
    Section IV-C fixes alpha_M at 0.9 to favor the stego identity and maximize attack success; the choice shapes the morph artifacts seen by the classifier and is not varied.
assumptions (4)
  • domain assumption The Chicago Face Database provides ICAO-compliant facial images suitable for ePassport enrollment.
    Section IV-A states each subject has a single ICAO-compliant image per [23]; the whole evaluation context depends on this and no independent compliance check is run.
  • domain assumption Embedding the marker does not break ICAO compliance or facial recognition usability of the certified image.
    Section V-A reports SSIM, MSE, and PSNR between original and certified images, but never tests whether the stego image still satisfies ICAO quality requirements or preserves face recognition performance.
  • domain assumption At verification time the integrity marker is present and matches the expected reference.
    Section III states this assumption explicitly and excludes missing, mismatched, or unrecoverable markers from the current work.
  • ad hoc to paper Morphing attacks are modeled as post-embedding modifications of the certified image.
    Section IV-C creates morphs by blending the stego-image after embedding. The canonical enrollment attack creates the morph before certification, which would leave the recovered marker intact.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Fragile Watermarking for Image Certification Using Deep Steganographic Embedding." pith.science (2026). https://pith.science/paper/XL2JRZQG

@misc{pith2026250413759,
  author       = {Pith},
  title        = {Pith review of: Fragile Watermarking for Image Certification Using Deep Steganographic Embedding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XL2JRZQG}},
  note         = {Machine review of arXiv:2504.13759}
}
read the original abstract

Modern identity verification systems increasingly rely on facial images embedded in biometric documents such as electronic passports. To ensure global interoperability and security, these images must comply with strict standards defined by the International Civil Aviation Organization (ICAO), which specify acquisition, quality, and format requirements. However, once issued, these images may undergo unintentional degradations (e.g., compression, resizing) or malicious manipulations (e.g., morphing) and deceive facial recognition systems. In this study, we explore fragile watermarking, based on deep steganographic embedding as a proactive mechanism to certify the authenticity of ICAO-compliant facial images. By embedding a hidden image within the official photo at the time of issuance, we establish an integrity marker that becomes sensitive to any post-issuance modification. We assess how a range of image manipulations affects the recovered hidden image and show that degradation artifacts can serve as robust forensic cues. Furthermore, we propose a classification framework that analyzes the revealed content to detect and categorize the type of manipulation applied. Our experiments demonstrate high detection accuracy, including cross-method scenarios with multiple deep steganography-based models. These findings support the viability of fragile watermarking via steganographic embedding as a valuable tool for biometric document integrity verification.

Figures

Figures reproduced from arXiv: 2504.13759 by the authors.

Figure 1
Figure 1. Overview of the proposed methodology, structured in [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Overview of the steganographic certification proces [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Recovered secret images examples using Steguz after [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Recovered secret images examples using Stegformer [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Effect of transformations on image recovery. Error b [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Confusion matrices for Stegformer and Steguz models [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Deep Data Hiding for ICAO-Compliant Face Images: A Survey

    cs.CV 2025-08 conditional novelty 4.0 of 10

    A survey of deep data hiding methods for ICAO-compliant face images concludes that only a subset of current deep watermarking and steganography models meet the combined requirements of imperceptibility, selective robu...

Reference graph

Works this paper leans on

31 extracted references · 25 canonical work pages · cited by 1 Pith paper

  1. [1]

    3d face reconstruction: the road to forensics,

    S. M. La Cava, G. Orr` u, M. Drahansky, G. L. Marcialis, and F. Roli, “3d face reconstruction: the road to forensics,” ACM Computing Surveys, vol. 56, no. 3, pp. 1–38, 2023

    work page 2023

  2. [2]

    Doc 9303, Part 9, 2015, vol

    International Civil Aviation Organization (ICAO), 9303-machine read- able travel documents-part 9: Deployment of biometric iden tification and electronic storage of data in emrtds , ser. Doc 9303, Part 9, 2015, vol. 123

    work page 2015

  3. [3]

    Face image quality assessment: A literature surv ey,

    T. Schlett, C. Rathgeb, O. Henniger, J. Galbally, J. Fier rez, and C. Busch, “Face image quality assessment: A literature surv ey,” ACM Comput. Surv. , vol. 54, no. 10s, Sep. 2022. [Online]. Available: https://doi.org/10.1145/3507901

    doi:10.1145/3507901 2022

  4. [4]

    Tolosana, R

    R. Tolosana, R. V era-Rodriguez, J. Fierrez, A. Morales, and J. Ortega- Garcia, An Introduction to Digital Face Manipulation . Cham: Springer International Publishing, 2022, pp. 3–26. [Onlin e]. Available: https://doi.org/10.1007/978-3-030-87664-7 1

    work page doi:10.1007/978-3-030-87664-7 2022

  5. [5]

    Face recognition systems under morphing attacks: A survey,

    U. Scherhag, C. Rathgeb, J. Merkle, R. Breithaupt, and C. Busch, “Face recognition systems under morphing attacks: A survey,” IEEE Access , vol. 7, pp. 23 012–23 026, 2019

    2019

  6. [6]

    Fraudule nt id using face morphs: Experiments on human and automatic recognitio n,

    D. J. Robertson, R. S. Kramer, and A. M. Burton, “Fraudule nt id using face morphs: Experiments on human and automatic recognitio n,” PLoS One, vol. 12, no. 3, p. e0173319, 2017

    work page 2017

  7. [7]

    The magic passpor t,

    M. Ferrara, A. Franco, and D. Maltoni, “The magic passpor t,” in IEEE international joint conference on biometrics . IEEE, 2014, pp. 1–7

    work page 2014

  8. [8]

    Face recognition vendor test (frvt) part 4: Morph-performance o f automated face morph detection,

    M. Ngan, M. Ngan, P . Grother, K. Hanaoka, and J. Kuo, “Face recognition vendor test (frvt) part 4: Morph-performance o f automated face morph detection,” 2020

    work page 2020

Show all 31 references
  1. [9]

    Fac e morphing attack generation and detection: A comprehensive survey,

    S. V enkatesh, R. Ramachandra, K. Raja, and C. Busch, “Fac e morphing attack generation and detection: A comprehensive survey,” IEEE trans- actions on technology and society , vol. 2, no. 3, pp. 128–145, 2021

    2021

  2. [10]

    Evaluating the integration of morph attack detection in automated face recognition systems,

    A. Panzino, S. M. La Cava, G. Orr` u, and G. L. Marcialis, “ Evaluating the integration of morph attack detection in automated face recognition systems,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 3827–3836

    2024

  3. [11]

    A survey on dee p- learning-based image steganography,

    B. Song, P . Wei, S. Wu, Y . Lin, and W. Zhou, “A survey on dee p- learning-based image steganography,” Expert Systems with Applications , p. 124390, 2024

    2024

  4. [12]

    Robust image stegano graphy against jpeg compression based on dct residual modulation,

    Y . Huang, Z. Liu, Q. Wu, and X. Liu, “Robust image stegano graphy against jpeg compression based on dct residual modulation, ” Signal Processing, vol. 219, p. 109431, 2024

    2024

  5. [13]

    Data hiding with deep learning: A survey unifying di gital watermarking and steganography,

    Z. Wang, O. Byrnes, H. Wang, R. Sun, C. Ma, H. Chen, Q. Wu, a nd M. Xue, “Data hiding with deep learning: A survey unifying di gital watermarking and steganography,” IEEE Transactions on Computational Social Systems , vol. 10, no. 6, pp. 2985–2999, 2023. (a) P6-8 - Trained on ...

    2023

  6. [14]

    Proact ive deepfake defence via identity watermarking,

    Y . Zhao, B. Liu, M. Ding, B. Liu, T. Zhu, and X. Y u, “Proact ive deepfake defence via identity watermarking,” in Proceedings of the IEEE/CVF winter conference on applications of computer vision , 2023, pp. 4602– 4611

    2023

  7. [15]

    Invis ible adversarial watermarking: A novel security mechanism for e nhancing copyright protection,

    J. Wang, H. Wang, J. Zhang, H. Wu, X. Luo, and B. Ma, “Invis ible adversarial watermarking: A novel security mechanism for e nhancing copyright protection,” ACM Transactions on Multimedia Computing, Communications and Applications , vol. 21, no. 2, pp. 1–22, 2024

    2024

  8. [16]

    Editgua rd: V ersatile image watermarking for tamper localization and copyright p rotection,

    X. Zhang, R. Li, J. Y u, Y . Xu, W. Li, and J. Zhang, “Editgua rd: V ersatile image watermarking for tamper localization and copyright p rotection,” in Proceedings of the IEEE/CVF conference on computer vision a nd pattern recognition, 2024, pp. 11 964–11 974

    2024

  9. [17]

    Hidden: H iding data with deep networks,

    J. Zhu, R. Kaplan, J. Johnson, and L. Fei-Fei, “Hidden: H iding data with deep networks,” in Proceedings of the European conference on computer vision (ECCV) , 2018, pp. 657–672

    2018

  10. [18]

    Steganogan: High capacity image steganography with gans,

    K. A. Zhang, A. Cuesta-Infante, L. Xu, and K. V eeramacha neni, “Steganogan: High capacity image steganography with gans, ” arXiv preprint arXiv:1901.03892, 2019

    1901 arXiv

  11. [19]

    Stegformer: rebuilding the glo ry of autoencoder-based steganography,

    X. Ke, H. Wu, and W. Guo, “Stegformer: rebuilding the glo ry of autoencoder-based steganography,” in Proceedings of the AAAI Confer- ence on Artificial Intelligence , vol. 38, no. 3, 2024, pp. 2723–2731

    2024

  12. [20]

    The chicago fa ce database: A free stimulus set of faces and norming data,

    D. S. Ma, J. Correll, and B. Wittenbrink, “The chicago fa ce database: A free stimulus set of faces and norming data,” Behavior research methods, vol. 47, pp. 1122–1135, 2015

    2015

  13. [21]

    Chicago face d atabase: Multiracial expansion,

    D. S. Ma, J. Kantner, and B. Wittenbrink, “Chicago face d atabase: Multiracial expansion,” Behavior Research Methods , vol. 53, pp. 1289– 1300, 2021

    2021

  14. [22]

    Th e india face set: International and cultural boundaries impact face imp ressions and perceptions of category membership,

    A. Lakshmi, B. Wittenbrink, J. Correll, and D. S. Ma, “Th e india face set: International and cultural boundaries impact face imp ressions and perceptions of category membership,” Frontiers in psychology , vol. 12, p. 627678, 2021

    2021

  15. [23]

    Accur ate and robust neural networks for face morphing attack detection,

    C. Seibold, W. Samek, A. Hilsmann, and P . Eisert, “Accur ate and robust neural networks for face morphing attack detection, ” Journal of Information Security and applications , vol. 53, p. 102526, 2020

    2020

  16. [24]

    Imperceptible image stegano graphy using symmetry-adapted deep learning techniques,

    A. Khalifa and A. Guzman, “Imperceptible image stegano graphy using symmetry-adapted deep learning techniques,” Symmetry, vol. 14, no. 7, p. 1325, 2022

    2022

  17. [25]

    U-net: Convol utional networks for biomedical image segmentation,

    O. Ronneberger, P . Fischer, and T. Brox, “U-net: Convol utional networks for biomedical image segmentation,” in Medical image computing and computer-assisted intervention–MICCAI 2015: 18th intern ational con- ference, Munich, Germany, October 5-9, 2015, proceedings, part III ...

    2015

  18. [26]

    Scope of validity of psnr in im- age/video quality assessment,

    Q. Huynh-Thu and M. Ghanbari, “Scope of validity of psnr in im- age/video quality assessment,” Electronics letters , vol. 44, no. 13, pp. 800–801, 2008

    2008

  19. [27]

    Statistica l evaluation of image quality measures,

    I. s. Avcıbas ¸ , B. l. Sankur, and K. Sayood, “Statistica l evaluation of image quality measures,” Journal of Electronic imaging , vol. 11, no. 2, pp. 206–223, 2002

    2002

  20. [28]

    Image quality assessment: from error visibility to structural si milarity,

    Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P . Simoncelli, “ Image quality assessment: from error visibility to structural si milarity,” IEEE transactions on image processing , vol. 13, no. 4, pp. 600–612, 2004

    2004

  21. [29]

    Using face morphing to pr otect privacy,

    P . Korshunov and T. Ebrahimi, “Using face morphing to pr otect privacy,” in 2013 10th IEEE international conference on advanced video a nd signal based surveillance . IEEE, 2013, pp. 208–213

    2013

  22. [30]

    Resnet 50,

    B. Koonce and B. Koonce, “Resnet 50,” Convolutional neural networks with swift for tensorflow: image recognition and dataset cat egorization, pp. 63–72, 2021

    2021

  23. [31]

    Imagenet: A large-scale hierarchical image database,

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fe i, “Imagenet: A large-scale hierarchical image database,” in 2009 IEEE conference on computer vision and pattern recognition . Ieee, 2009, pp. 248–255. This figure "fig1.png" is available in "png" format from: http://...

    2009 arXiv

Pith tools

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