Pith. sign in

REVIEW 5 major objections 5 minor 65 references

VERITAS: Verification and Explanation of Realness in Images for Transparency in AI Systems

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 VERITAS claims that a five-stage pipeline can detect AI-generated 32x32 images and explain the detection through localized, human-readable artifact descriptions.

desk verdict A clearly written pipeline paper whose central claims—accurate detection and artifact explanation—have no quantitative support; the artifact-score link is unvalidated. read the letter →

arxiv 2507.05146 v1 pith:5BCT6IXA submitted 2025-07-07 cs.CV cs.LG

Aadi Srivastava , Vignesh Natarajkumar , Utkarsh Bheemanaboyna , Devisree Akashapu , Nagraj Gaonkar , Archit Joshi This is my paper · ORCID
classification cs.CVcs.LG
keywords syntheticimagedetectionexplainableAIartifactlocalizationvision-languagemodelsCLIPsimilarityscoringGradCAMattentionlow-resolutionanalysisCIFAKE
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

VERITAS is a five-stage framework for classifying 32x32 images as real or AI-generated and, crucially, for explaining each 'fake' verdict by pointing to visual artifacts and describing them in plain language. The authors argue that existing detectors focus on accuracy at higher resolutions and leave users with black-box decisions, whereas small images such as profile pictures, thumbnails, and messaging images are a realistic and high-risk setting for synthetic media. Their claim is that a pipeline combining super-resolution, GradCAM-based localization, patch-level CLIP scoring against hand-written artifact descriptors, and MOLMO-generated text can both detect and account for synthetic images. If correct, this would make synthetic-image detection transparent enough for ordinary users to judge why an image was flagged, not just trust that it was.

What carries the argument

The load-bearing mechanism is the artifact score $S$ computed in Step 4. A patch is encoded by CLIP's similarity to three hand-written descriptors—positive (artifact present), negative (realistic), and neutral (artifact type not applicable to that patch)—and each patch votes for the descriptor with highest similarity. The votes are averaged with weights $w_k$ equal to the summed GradCAM heatmap intensity over the patch. This is the place where a visual artifact is converted into a number, and it is also the step that ties localization (where the heatmap says to look) to semantic reasoning (what the text says to look for); the threshold on $S$ decides whether MOLMO is asked to explain.

What would settle it

Compare the per-patch votes $v_k$ and the final artifact score $S$ against human-annotated artifact masks on a held-out sample of 32x32 synthetic images; if high-$S$ patches do not overlap the regions humans mark as artifacts significantly more often than chance, then the pipeline is not explaining the true basis of its decisions. A simpler test: substitute shuffled or nonsensical artifact descriptors and check whether $S$ stays high; if it does, the score is not measuring the semantic content that the explanations depend on.

Watch

Extended reading notes

Core claim

On its own terms, the paper's contribution is a five-stage pipeline: DRCT super-resolves the 32x32 input; a binary classifier with GradCAM produces a heatmap; the super-resolved image is cut into patches weighted by heatmap intensity; each patch casts a positive, negative, or neutral vote based on which of three hand-written artifact descriptions is most CLIP-similar; and the weighted votes form an artifact score $S=(\sum_k w_k v_k)/(\sum_k w_k)$. Patches and images above a threshold are passed to MOLMO, which generates human-readable artifact descriptions such as 'legs bent at unnatural angles' or 'visible ghosting effects near the center.' The authors claim this approach is more effective than MOLMO alone at detecting and helping users understand spatial and frequency-based artifacts in low-resolution images, and they present qualitative examples as evidence.

Load-bearing premise

The entire explanation stage rests on the assumption that CLIP's similarity between a patch and a hand-written sentence like 'this patch contains a ghosting artifact' actually tracks whether the artifact is present, so if CLIP similarity and real artifact presence diverge, the score, threshold, and every downstream explanation lose their grounding.

Editorial extensions

If this is right

  • If VERITAS works as described, a low-resolution image can be flagged as synthetic with a concrete reason: the explanation names a localized artifact rather than a global probability.
  • The same patch-scoring scheme can in principle report on multiple artifact types from one image, since each patch is scored independently against many descriptors.
  • Selecting a VLM matters: the comparison indicates MOLMO's localized descriptions are better suited than Qwen2.5 VL or Pixtral 12B for fine-grained artifact explanations, so downstream users would inherit that choice.
  • Because the pipeline is patch-based and CLIP is zero-shot, the explanation stage does not need retraining on each new generative model, provided the artifact descriptors remain applicable.

Reading between the lines

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

  • The paper does not test this, but the same weighted-vote mechanism could be inverted to rank which artifact descriptors are most discriminative across a dataset, turning the explanation system into a tool for discovering new generator fingerprints.
  • If the CLIP descriptor proxy holds, updating the descriptor list becomes a no-retraining route to adapting the detector to new generators; the paper mentions domain generalization but does not demonstrate this route.
  • The framework's usefulness for trust hinges on the reliability of MOLMO's text; pairing the artifact score with a hallucination filter would make the explanations checkable, a step the paper lists as future work.
  • Because GradCAM heatmaps come from one classifier, the localization is only as good as that classifier's attention; aggregating heatmaps from multiple classifiers would give a stronger test of whether the explanations track real image content.
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 paper proposes VERITAS, a five-stage pipeline for detecting and explaining AI-generated 32x32 images. The pipeline super-resolves the input with DRCT, computes GradCAM heatmaps on the original low-resolution image, divides the super-resolved image into patches weighted by those heatmaps, aggregates CLIP-based votes over hand-written artifact descriptors into an artifact score S (Eq. 2), and finally prompts MOLMO to generate textual explanations for patches with high S. The authors claim that VERITAS not only accurately classifies images as real or synthetic but also provides human-readable, artifact-grounded explanations. The experimental section (Section 5) presents qualitative comparisons of VLM outputs and selected pipeline-generated descriptions against MOLMO, with no quantitative metrics anywhere in the paper.

Significance. If validated, an explainable detector for low-resolution synthetic images would be a useful contribution, and the integration of super-resolution, GradCAM, CLIP, and a VLM into a single pipeline is a reasonable high-level idea. The authors provide a code repository and describe their prompts, which is helpful for reproducibility. However, the paper contains no quantitative evaluation supporting the central claim of accurate detection, and the explanation mechanism in Eq. (2) is never validated against any ground-truth artifact signal. The contribution therefore currently rests on unsupported assertions, and the qualitative tables do not establish that the pipeline explains anything beyond what MOLMO already produces from the entire image.

major comments (5)
  1. [Section 5.4; abstract] The central claim that VERITAS “accurately detects whether a small (32x32) image is AI-generated” is never quantified. The paper reports no accuracy, precision, recall, F1, AUC, confusion matrix, or statistical test for the binary classification stage, nor any detection rate for the artifact score S. Tables 2–6 contain only hand-picked examples and qualitative descriptions, so the sentence “Clearly, our proposed approach is more effective than MOLMO” is unsupported by any measured result.
  2. [Section 4.4, Eq. (2)] The artifact score S is the load-bearing link between GradCAM-attended patches and the claimed explanations, but the paper never validates that CLIP similarity to the hand-written positive/negative/neutral descriptors tracks actual artifact presence. No threshold value for S is given, the numerical encoding of the vote v_k is not specified, and there is no comparison of S between real and synthetic images, no sensitivity analysis, and no correlation with external artifact annotations or human judgments. Consequently, the MOLMO explanations generated in Step 5 are produced for regions whose artifact status is unverified.
  3. [Section 5.4, Tables 2–6] The comparison against MOLMO is not a valid evaluation of explanation quality. The tables list artifact names and generated descriptions with no metric, no blind human evaluation, no hallucination rate (a concern the authors themselves acknowledge in Section 6.2), and no measurement of false positives on real images. The claimed superiority over MOLMO therefore lacks empirical support.
  4. [Section 5.2, Table 1] The VLM selection process is described as a comparison of “accuracy and descriptiveness,” but no scoring rubric, quantitative metric, or statistical test is presented; Table 1 is purely qualitative. The subsequent claim in Section 5.3 that MOLMO has a localization-aware architecture advantage is also presented without evidence. Since this selection determines the explanatory component of the pipeline, the justification for MOLMO is unsubstantiated.
  5. [Section 3.5, Figure 2] The proposed latent-space optimization with contrastive and triplet losses is described as a contribution but is never used in the main pipeline and is not evaluated with any quantitative measure. The t-SNE visualization in Figure 2 has no legend or class labels, so the claimed improvement in feature separability cannot be checked by the reader.
minor comments (5)
  1. [Section 4.1, Step 2] The sentence “This heatmap serves two key purposes:” appears twice in succession, creating a redundant passage.
  2. [Section 3.5] The contrastive loss expression has a formatting error with an unbalanced parenthesis, and the combined loss uses α and β without any discussion of how these coefficients are chosen or normalized.
  3. [Table 4] The artifact “Improper fur direction flows” is described in the pipeline output as visible near a vehicle’s left front wheel arch, while the same row also mentions fur on the vehicle’s surface; this looks like a hallucinated caption and should be either corrected or explicitly discussed as a failure case.
  4. [Section 5.1] The experimental setup lacks details on the train/validation/test split, the number of Optuna trials, the hyperparameters for DRCT and the classifiers, and the random seed; these omissions prevent reproduction.
  5. [References] Several references are incomplete or malformed, including the FakeFormer entry which is a bare arXiv URL, and some bibliographic entries are missing venue or year information.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: detection uses external CIFAKE labels, explanation modules are external, and no central claim reduces to its own inputs by construction.

full rationale

The binary classification stage is trained on CIFAKE, an externally labeled dataset, so the real/fake decision is not derived from the paper's own outputs. The explanation pipeline uses external components: DRCT super-resolution, GradCAM, CLIP, and MOLMO. GradCAM is a post-hoc method on the trained classifier, CLIP is pretrained, and MOLMO is an open-source VLM; none of these are outputs of this paper. Equation (2) defines the artifact score as a weighted vote over hand-written CLIP descriptors, but no result is fitted from, or defined in terms of, the explanation that the paper claims to predict. The paper acknowledges validity threats in Section 6.1 (GradCAM is model-dependent and may reflect classifier biases) and Section 6.2 (VLMs can hallucinate), and Section 5.4's qualitative, selected-example comparison lacks external artifact ground truth, human evaluation, threshold reporting, and false-positive measurement. These are empirical validation gaps and correctness risks, not circularity: the explanations could be wrong, but that does not make the derivation equivalent to its inputs. There are no load-bearing self-citations and no imported uniqueness theorems. Therefore no circular step meeting the quoted-reduction standard is present.

Assumptions & free parameters 3 free parameters · 5 assumptions · 1 invented entities

The central detection claim relies on standard supervised labels from CIFAKE, but the explanatory claim rests on author-defined artifact classes, unvalidated CLIP descriptors, and an unstated threshold. No free parameters are reported numerically, so the pipeline is not reproducible as specified.

free parameters (3)
  • Artifact score threshold theta = not reported
    In Section 4.4 and Algorithm 3, images whose S exceeds 'a predefined threshold' are retained; the threshold is never specified and directly controls which images get explanations.
  • Patch vote encoding v_k = not reported
    Eq. (2) sums w_k times v_k, but the numeric encoding of positive, negative, and neutral votes is not given; changing it shifts the artifact score.
  • Optuna ensemble weights = not reported
    Algorithm 1 fits w1, w2, w3 on validation data; no fitted weights or validation metrics are reported.
assumptions (5)
  • domain assumption CIFAKE real and fake labels are correct and Stable Diffusion artifacts are present in 32x32 images after super-resolution.
    Section 5.1 treats CIFAKE as ground truth; no artifact-level ground truth is used.
  • domain assumption GradCAM heatmaps from the binary classifier identify true generative artifacts rather than classifier biases.
    Section 4.2 uses heatmaps to weight patches, but no heatmap-accuracy evaluation is given.
  • domain assumption CLIP similarity against hand-written artifact descriptors is a valid proxy for artifact presence.
    Section 4.4 defines S with CLIP votes; no calibration or validation against external artifact labels is reported.
  • domain assumption DRCT super-resolution reveals artifacts without introducing new ones that mimic them.
    Section 4.1 relies on DRCT; Section 6.3 admits upscaling can introduce artifacts.
  • domain assumption MOLMO descriptions are factually grounded.
    Section 5 uses MOLMO for text; Section 6.2 acknowledges hallucinations are a risk.
invented entities (1)
  • Curated artifact descriptor taxonomy
    purpose: Enables CLIP patch scoring and provides the vocabulary for explanations.
    The positive, negative, and neutral triplets in Section 4.4 are authored for this paper and are not benchmarked against any external artifact annotation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of VERITAS: Verification and Explanation of Realness in Images for Transparency in AI Systems." pith.science (2026). https://pith.science/paper/5BCT6IXA

@misc{pith2026250705146,
  author       = {Pith},
  title        = {Pith review of: VERITAS: Verification and Explanation of Realness in Images for Transparency in AI Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5BCT6IXA}},
  note         = {Machine review of arXiv:2507.05146}
}
read the original abstract

The widespread and rapid adoption of AI-generated content, created by models such as Generative Adversarial Networks (GANs) and Diffusion Models, has revolutionized the digital media landscape by allowing efficient and creative content generation. However, these models also blur the difference between real images and AI-generated synthetic images, raising concerns regarding content authenticity and integrity. While many existing solutions to detect fake images focus solely on classification and higher-resolution images, they often lack transparency in their decision-making, making it difficult for users to understand why an image is classified as fake. In this paper, we present VERITAS, a comprehensive framework that not only accurately detects whether a small (32x32) image is AI-generated but also explains why it was classified that way through artifact localization and semantic reasoning. VERITAS produces human-readable explanations that describe key artifacts in synthetic images. We show that this architecture offers clear explanations of the basis of zero-shot synthetic image detection tasks. Code and relevant prompts can be found at https://github.com/V-i-g-n-e-s-h-N/VERITAS .

Figures

Figures reproduced from arXiv: 2507.05146 by the authors.

Figure 1
Figure 1. Sample image generated by Adobe Firefly, analyzed by the proposed pipeline. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Embeddings trained on contrastive and triplet loss [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Pipeline for Detecting Artifacts Using Super-Resolution and GradCAM [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Examples 1 and 2 [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]
Figure 5
Figure 5. Figure 5: Examples 3 and 4 [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

65 extracted references · 41 canonical work pages

  1. [1]

    https://arxiv.org/html/2410.21964

    FakeFormer: Efficient Vulnerability-Driven Transformers for Generalisable Deepfake Detection. https://arxiv.org/html/2410.21964. Wahidul Abir, Faria Khanam, Kazi Alam, Myriam Hadjouni, Hela Elmannai, Sami Bourouis, Rajesh Dey, and Moham- mad Khan. Detecting Deepfake Images Using Deep Learning Techniques and Explainable AI Methods.Intelligent Au- tomation ...

    arXiv

  2. [4]

    Naveed Akhtar, Ajmal Mian, Navid Kardan, and Mubarak Shah

    doi:10.1109/CoDIT62066.2024.10708200. Naveed Akhtar, Ajmal Mian, Navid Kardan, and Mubarak Shah. Advances in Adversarial Attacks and Defenses in Computer Vision: A Survey. IEEE Access , 9:155161–155196,

    arXiv 2024

  3. [8]

    doi:10.1145/3647722.3647739

    ACM. doi:10.1145/3647722.3647739. URLhttps://dl.acm.org/doi/10.1145/3647722.3647739. Rishi Bommasani, Drew A. Hudson, Ehsan Adeli, Russ Altman, Simran Arora, Sydney von Arx, Michael S. Bernstein, Jeannette Bohg, Antoine Bosselut, Emma Brunskill, Erik Brynjolfsson, Shyamal Buch, Dallas Card, Rodrigo Castellon, Niladri Chatterji, Annie Chen, Kathleen Creel,...

    work page arXiv

  4. [10]

    VisualGPT: Data-efficient Adaptation of Pre- trained Language Models for Image Captioning

    Jun Chen, Han Guo, Kai Yi, Boyang Li, and Mohamed Elhoseiny. VisualGPT: Data-efficient Adaptation of Pre- trained Language Models for Image Captioning. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 18009–18019, New Orleans, LA, USA, June

    work page 2022

  5. [11]

    ISBN 978-1-6654-6946-3

    IEEE. ISBN 978-1-6654-6946-3. doi:10.1109/CVPR52688.2022.01750. Jun Chen, Deyao Zhu, Xiaoqian Shen, Xiang Li, Zechun Liu, Pengchuan Zhang, Raghuraman Krishnamoorthi, Vikas Chandra, Yunyang Xiong, and Mohamed Elhoseiny. MiniGPT-v2: Large language model as a unified interface for vision-language multi-task learning, November

    arXiv 2022

  6. [12]

    Francesco Croce and Matthias Hein

    doi:10.1007/978-3-031-06433- 3_19. Francesco Croce and Matthias Hein. Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attacks, August

    work page doi:10.1007/978-3-031-06433-

  7. [13]

    doi:10.1007/s00354-024-00283-0

    ISSN 0288-3635, 1882-7055. doi:10.1007/s00354-024-00283-0. Chao Dong, Chen Change Loy, Kaiming He, and Xiaoou Tang. Image Super-Resolution Using Deep Convolutional Networks, July

    work page doi:10.1007/s00354-024-00283-0

  8. [14]

    ISBN 978-1-4503-6859-9

    ACM. ISBN 978-1-4503-6859-9. doi:10.1145/3340531.3411892. Ricard Durall, Margret Keuper, Franz-Josef Pfreundt, and Janis Keuper. Unmasking DeepFakes with simple Features, March

    arXiv

Show all 65 references
  1. [15]

    doi:10.1109/ACCESS.2023.3271515

    ISSN 2169-3536. doi:10.1109/ACCESS.2023.3271515. Joel Frank, Thorsten Eisenhofer, Lea Schönherr, Asja Fischer, Dorothea Kolossa, and Thorsten Holz. Leveraging Frequency Analysis for Deep Fake Image Recognition, June

    2023

  2. [17]

    Lips Don’t Lie: A Generalis- able and Robust Approach to Face Forgery Detection

    Alexandros Haliassos, Konstantinos V ougioukas, Stavros Petridis, and Maja Pantic. Lips Don’t Lie: A Generalis- able and Robust Approach to Face Forgery Detection. In 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5037–5047, Nashville, TN, USA, June

    2021

  3. [18]

    ISBN 978-1-6654-4509-2

    IEEE. ISBN 978-1-6654-4509-2. doi:10.1109/CVPR46437.2021.00500. Darryl Hannan, John Cooper, Dylan White, Timothy Doster, Henry Kvinge, and Yijing Watkins. Foundation Models for Remote Sensing: An Analysis of MLLMs for Object Localization, April

    2021

  4. [20]

    DRCT: Saving Image Super-Resolution away from Information Bottleneck

    VERITAS Chih-Chung Hsu, Chia-Ming Lee, and Yi-Shiuan Chou. DRCT: Saving Image Super-Resolution away from Information Bottleneck. In 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 6133–6142, Seattle, W A, USA, June

    2024

  5. [21]

    ISBN 979-8-3503-6547-4

    IEEE. ISBN 979-8-3503-6547-4. doi:10.1109/CVPRW63382.2024.00618. Ling Huang, Anthony D Joseph, Blaine Nelson, Benjamin I P Rubinstein, and J D Tygar. Adversarial Machine Learning. Uiwon Hwang, Jaewoo Park, Hyemi Jang, Sungroh Yoon, and Nam Ik Cho. PuV AE: A Variational Autoenc...

    2024

  6. [22]

    arXiv:2009.09622 [eess]

    URL http://arxiv.org/abs/2009.09622. arXiv:2009.09622 [eess]. Samer Y . Khamaiseh, Derek Bagagem, Abdullah Al-Alaj, Mathew Mancino, and Hakam W. Alomari. Adversarial Deep Learning: A Survey on Adversarial Attacks and Defense Mechanisms on Image Classification. IEEE Access, 10:...

    2009 arXiv

  7. [23]

    doi:10.1109/ACCESS.2022.3208131

    ISSN 2169-3536. doi:10.1109/ACCESS.2022.3208131. Sohail Ahmed Khan and Duc-Tien Dang-Nguyen. CLIPping the Deception: Adapting Vision-Language Models for Universal Deepfake Detection, February

    2022

  8. [24]

    Adversarial Examples for Generative Models

    Jernej Kos, Ian Fischer, and Dawn Song. Adversarial Examples for Generative Models. In 2018 IEEE Security and Privacy Workshops (SPW), pages 36–42, May

    2018

  9. [25]

    Alexey Kurakin, Ian Goodfellow, and Samy Bengio

    doi:10.1109/SPW.2018.00014. Alexey Kurakin, Ian Goodfellow, and Samy Bengio. Adversarial examples in the physical world, February

    2018

  10. [26]

    In Ictu Oculi: Exposing AI Created Fake Videos by Detecting Eye Blinking

    Yuezun Li, Ming-Ching Chang, and Siwei Lyu. In Ictu Oculi: Exposing AI Created Fake Videos by Detecting Eye Blinking. In 2018 IEEE International Workshop on Information Forensics and Security (WIFS), pages 1–7, Hong Kong, Hong Kong, December

    2018

  11. [27]

    ISBN 978-1-5386-6536-7

    IEEE. ISBN 978-1-5386-6536-7. doi:10.1109/WIFS.2018.8630787. VERITAS Zongxia Li, Xiyang Wu, Hongyang Du, Fuxiao Liu, Huy Nghiem, and Guangyao Shi. A Survey of State of the Art Large Vision Language Models: Alignment, Benchmark, Evaluations and Challenges, April

    2018

  12. [28]

    Explainable Deep-Fake Detection Using Visual Interpretability Methods

    Badhrinarayan Malolan, Ankit Parekh, and Faruk Kazi. Explainable Deep-Fake Detection Using Visual Interpretability Methods. In 2020 3rd International Conference on Information and Computer Technologies (ICICT), pages 289–293, March

    2020

  13. [29]

    François Menet, Paul Berthier, Michel Gagnon, and José M

    doi:10.1109/ICICT50521.2020.00051. François Menet, Paul Berthier, Michel Gagnon, and José M. Fernandez. Spartan Networks: Self-feature-squeezing neural networks for increased robustness in adversarial settings. Computers & Security, 88:101537, January

    2020

  14. [30]

    doi:10.1016/j.cose.2019.05.014

    ISSN 01674048. doi:10.1016/j.cose.2019.05.014. Dongyu Meng and Hao Chen. MagNet: A Two-Pronged Defense against Adversarial Examples, September

    2019 doi

  15. [33]

    doi:10.1016/j.heliyon.2024.e25933

    ISSN 24058440. doi:10.1016/j.heliyon.2024.e25933. Utkarsh Ojha, Yuheng Li, and Yong Jae Lee. Towards Universal Fake Image Detectors that Generalize Across Generative Models, April

    2024 doi

  16. [34]

    Berkay Celik, and Ananthram Swami

    Nicolas Papernot, Patrick McDaniel, Somesh Jha, Matt Fredrikson, Z. Berkay Celik, and Ananthram Swami. The Limitations of Deep Learning in Adversarial Settings. In 2016 IEEE European Symposium on Security and Privacy (EuroS&P), pages 372–387, March 2016a. doi:10.1109/EuroSP.20...

    2016 doi

  17. [35]

    Thinking in Frequency: Face Forgery Detection by Mining Frequency-Aware Clues

    Yuyang Qian, Guojun Yin, Lu Sheng, Zixuan Chen, and Jing Shao. Thinking in Frequency: Face Forgery Detection by Mining Frequency-Aware Clues. In Andrea Vedaldi, Horst Bischof, Thomas Brox, and Jan-Michael Frahm, editors, Computer Vision – ECCV 2020, volume 12357, pages 86–103....

    2020

  18. [36]

    doi:10.1007/978-3-030-58610-2_6

    ISBN 978-3-030-58609-6 978-3-030-58610-2. doi:10.1007/978-3-030-58610-2_6. VERITAS Chongli Qin, James Martens, Sven Gowal, Dilip Krishnan, Krishnamurthy Dvijotham, Alhussein Fawzi, Soham De, Robert Stanforth, and Pushmeet Kohli. Adversarial Robustness through Local Linearizati...

    doi

  19. [37]

    doi:10.1007/s11633-023-1469-x

    ISSN 2731-538X, 2731-5398. doi:10.1007/s11633-023-1469-x. Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning Transferable Visual Mode...

    doi

  20. [38]

    doi:10.1609/aaai.v32i1.11504

    ISSN 2374-3468, 2159-5399. doi:10.1609/aaai.v32i1.11504. Andreas Rössler, Davide Cozzolino, Luisa Verdoliva, Christian Riess, Justus Thies, and Matthias Nießner. FaceForen- sics++: Learning to Detect Manipulated Facial Images, August

    doi

  21. [40]

    doi:10.1007/s11263-019-01228-7

    ISSN 0920-5691, 1573-1405. doi:10.1007/s11263-019-01228-7. Bo Sun, Nian-Hsuan Tsai, Fangchen Liu, Ronald Yu, and Hao Su. Adversarial Defense by Stratified Convolutional Sparse Coding. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 11439–1...

    2019 doi

  22. [41]

    ISBN 978-1-7281-3293-8

    IEEE. ISBN 978-1-7281-3293-8. doi:10.1109/CVPR.2019.01171. Na Sun and Huina Li. Super Resolution Reconstruction of Images Based on Interpolation and Full Convolutional Neural Network and Application in Medical Fields. IEEE Access, 7:186470–186479,

    2019

  23. [42]

    doi:10.1109/ACCESS.2019.2960828

    ISSN 2169-3536. doi:10.1109/ACCESS.2019.2960828. Christian Szegedy, Wojciech Zaremba, Ilya Sutskever, Joan Bruna, Dumitru Erhan, Ian Goodfellow, and Rob Fergus. Intriguing properties of neural networks, February

    2019

  24. [44]

    ISBN 979-8-3503-0129-8

    IEEE. ISBN 979-8-3503-0129-8. doi:10.1109/CVPR52729.2023.01165. Naftali Tishby and Noga Zaslavsky. Deep learning and the information bottleneck principle. In 2015 IEEE Information Theory Workshop (ITW), pages 1–5, Jerusalem, Israel, April

    2023

  25. [47]

    D2Defend: Dual-Domain based Defense against Ad- versarial Examples

    Xin Yan, Yanjie Li, Tao Dai, Yang Bai, and Shu-Tao Xia. D2Defend: Dual-Domain based Defense against Ad- versarial Examples. In 2021 International Joint Conference on Neural Networks (IJCNN), pages 1–8, July

    2021

  26. [48]

    Shukang Yin, Chaoyou Fu, Sirui Zhao, Tong Xu, Hao Wang, Dianbo Sui, Yunhang Shen, Ke Li, Xing Sun, and Enhong Chen

    doi:10.1109/IJCNN52387.2021.9533589. Shukang Yin, Chaoyou Fu, Sirui Zhao, Tong Xu, Hao Wang, Dianbo Sui, Yunhang Shen, Ke Li, Xing Sun, and Enhong Chen. Woodpecker: Hallucination Correction for Multimodal Large Language Models. Science China Information Sciences, 67(12):220105...

    2021

  27. [49]

    doi:10.1007/s11432-024-4251-x

    ISSN 1674-733X, 1869-1919. doi:10.1007/s11432-024-4251-x. Lejun Yu, Siming Cao, Jun He, Bo Sun, and Feng Dai. Single-image super-resolution based on regularization with stationary gradient fidelity. In 2017 10th International Congress on Image and Signal Processing, BioMedical...

    1919 doi

  28. [50]

    Ning Yu, Larry Davis, and Mario Fritz

    doi:10.1109/CISP-BMEI.2017.8301942. Ning Yu, Larry Davis, and Mario Fritz. Attributing Fake Images to GANs: Learning and Analyzing GAN Fingerprints. In 2019 IEEE/CVF International Conference on Computer Vision (ICCV), pages 7555–7565, Seoul, Korea (South), October

    2017

  29. [51]

    ISBN 978-1-7281-4803-8

    IEEE. ISBN 978-1-7281-4803-8. doi:10.1109/ICCV .2019.00765. Duzhen Zhang, Yahan Yu, Jiahua Dong, Chenxing Li, Dan Su, Chenhui Chu, and Dong Yu. MM-LLMs: Recent Advances in MultiModal Large Language Models. In Lun-Wei Ku, Andre Martins, and Vivek Srikumar, editors, Findings of ...

    2019

  30. [52]

    doi:10.1109/TIP.2018.2826139

    ISSN 1941-0042. doi:10.1109/TIP.2018.2826139. Yutong Zhang, Yao Li, Yin Li, and Zhichang Guo. A Review of Adversarial Attacks in Computer Vision, August

    1941

  31. [53]

    Clipped BagNet: Defending Against Sticker Attacks with Clipped Bag-of-features

    Zhanyuan Zhang, Benson Yuan, Michael McCoyd, and David Wagner. Clipped BagNet: Defending Against Sticker Attacks with Clipped Bag-of-features. In 2020 IEEE Security and Privacy Workshops (SPW), pages 55–61, May

    2020

  32. [54]

    Zhun Zhang, Yi Zeng, Qihe Liu, and Shijie Zhou

    doi:10.1109/SPW50608.2020.00026. Zhun Zhang, Yi Zeng, Qihe Liu, and Shijie Zhou. Towards a Novel Perspective on Adversarial Examples Driven by Frequency, April 2024c. Nan Zhong, Yiran Xu, Sheng Li, Zhenxing Qian, and Xinpeng Zhang. PatchCraft: Exploring Texture Patch for Effic...

    2020

  33. [55]

    Adversarial Learning With Margin-Based Triplet Embedding Regularization

    Yaoyao Zhong and Weihong Deng. Adversarial Learning With Margin-Based Triplet Embedding Regularization. In 2019 IEEE/CVF International Conference on Computer Vision (ICCV), pages 6548–6557, Seoul, Korea (South), October

    2019

  34. [56]

    artifact

    IEEE. ISBN 978-1-7281-4803-8. doi:10.1109/ICCV .2019.00665. Deyao Zhu, Jun Chen, Xiaoqian Shen, Xiang Li, and Mohamed Elhoseiny. MINIGPT-4: ENHANCING VISION- LANGUAGE UNDERSTANDING WITH ADV ANCED LARGE LANGUAGE MODELS. VERITAS Appendix A Artifact Prompting Instruction: You are...

    2019

  35. [57]

    (Zhang et al., 2023), (Dhamija and Bansal, 2024). Such attacks have been shown to exploit various vulnerabilities in models and pose serious security risks in practical settings (Goodfellow et al., 2015), (Huang et al.), (Szegedy et al., 2014), (Papernot et al., 2016a), (Kurak...

    2023

  36. [58]

    Other white-box attacks convert attacks into optimization problems, such as the Carlini and Wagner Attack (C&W) (Carlini and Wagner, 2017)

    and Jacobian-based Saliency Map Attacks (JSMA) (Papernot et al., 2016a), with more advanced attacks also being proposed such as GreedyFool (Dong et al., 2020). Other white-box attacks convert attacks into optimization problems, such as the Carlini and Wagner Attack (C&W) (Carl...

    2020

  37. [59]

    (Dhamija and Bansal, 2024). Some attacks are discussed in greater detail below: Fast Gradient Sign Method (Goodfellow et al., 2015): It exploits the gradient of the loss function with respect to the input data to craft a perturbation in the direction that maximizes the model’s...

    2024

  38. [60]

    By slightly altering the wavelet coefficients at specific scales and then reconstructing the image, it creates an adversarial example designed to mislead neural networks

    Wavelet Packet Decomposition (Zhang et al., 2024c): This method uses a wavelet transform to break down an input image into frequency components. By slightly altering the wavelet coefficients at specific scales and then reconstructing the image, it creates an adversarial exampl...

    2020

  39. [61]

    PatchGuard (Xiang et al.,

    and its derivatives, such as Clipped BAGNet (Zhang et al., 2020), limit the receptive field of the convolutional layers through dimension reduction of filters and introducing averaging methods instead. PatchGuard (Xiang et al.,

    2020

  40. [62]

    (Goodfellow et al., 2015), (Szegedy et al., 2014), (Madry et al.,

    also introduces a robust masking scheme to mask attacks on images, apart from reducing the input receptive field. (Goodfellow et al., 2015), (Szegedy et al., 2014), (Madry et al.,

    2015

  41. [63]

    teacher” model on the input data, following which the output probabilities are learned by a “student

    and image quilting (Efros and Freeman) are effective in tackling many adversarial attacks. Defensive distillation, proposed by (Papernot et al., 2016b) and built upon the idea of distillation ideated by (Ba and Caruana, 2014), uses an innovative method of training a “teacher” ...

    2014

  42. [64]

    Other forms of regularization are also explored in (Zhong and Deng, 2019), (Qin et al., 2019)

    performs input regularization, which then reduces the tendency of target models to flip decisions with perturbed inputs. Other forms of regularization are also explored in (Zhong and Deng, 2019), (Qin et al., 2019). To target imperceptible manipulations in the frequency domain...

    2019

  43. [65]

    An alternative approach to adversarial robustness is ensuring that the internal representations and feature maps of the model are modified to be resilient to perturbations

    to detect texture inconsistencies. An alternative approach to adversarial robustness is ensuring that the internal representations and feature maps of the model are modified to be resilient to perturbations. This is done by (Sun et al., 2019), which introduces a sparse transfo...

    2019

  44. [1992]

    doi:10.1016/0167-2789(92)90242-F

    ISSN 01672789. doi:10.1016/0167-2789(92)90242-F. Ramprasaath R. Selvaraju, Michael Cogswell, Abhishek Das, Ramakrishna Vedantam, Devi Parikh, and Dhruv Batra. Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization. International Journal of Computer Vi...

    doi

  45. [2007]

    doi:10.1007/s11045-007-0023-2

    ISSN 0923-6082, 1573-0824. doi:10.1007/s11045-007-0023-2. Laurens van der Maaten and Geoffrey Hinton. Visualizing data using t-SNE. Journal of Machine Learning Research, 9 (86):2579–2605,

    doi

  46. [2014]

    Learning on Gradients: Generalized Artifacts Representation for GAN-Generated Images Detection

    Chuangchuang Tan, Yao Zhao, Shikui Wei, Guanghua Gu, and Yunchao Wei. Learning on Gradients: Generalized Artifacts Representation for GAN-Generated Images Detection. In 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 12105–12114, Vancouver, BC...

    2023

  47. [2015]

    ISBN 978-1-4799-5524-4 978-1-4799- 5526-8

    IEEE. ISBN 978-1-4799-5524-4 978-1-4799- 5526-8. doi:10.1109/ITW.2015.7133169. C. S. Tong and K. T. Leung. Super-resolution reconstruction based on linear interpolation of wavelet coefficients. Multidimensional Systems and Signal Processing, 18(2-3):153–171, September

    2015

  48. [2016]

    Gourab Naskar, Sk Mohiuddin, Samir Malakar, Erik Cuevas, and Ram Sarkar

    doi:10.1109/CVPR.2016.282. Gourab Naskar, Sk Mohiuddin, Samir Malakar, Erik Cuevas, and Ram Sarkar. Deepfake detection us- ing deep feature stacking and meta-learning. Heliyon, 10(4):e25933, February

    2016 doi

  49. [2017]

    DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks

    Seyed-Mohsen Moosavi-Dezfooli, Alhussein Fawzi, and Pascal Frossard. DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 2574–2582, June

    2016

  50. [2018]

    doi:10.1109/LSP.2018.2815003

    ISSN 1558-2361. doi:10.1109/LSP.2018.2815003. Hanting Chen, Yunhe Wang, Tianyu Guo, Chang Xu, Yiping Deng, Zhenhua Liu, Siwei Ma, Chunjing Xu, Chao Xu, and Wen Gao. Pre-Trained Image Processing Transformer, November

    2018

  51. [2019]

    URL https://www.science.org/doi/abs/10.1126/scirobotics

    doi:10.1126/scirobotics.aay7120. URL https://www.science.org/doi/abs/10.1126/scirobotics. aay7120. _eprint: https://www.science.org/doi/pdf/10.1126/scirobotics.aay7120. Chuan Guo, Mayank Rana, Moustapha Cisse, and Laurens van der Maaten. Countering Adversarial Images using Inp...

    doi

  52. [2020]

    doi:10.1109/ACCESS.2020.3007337

    ISSN 2169-3536. doi:10.1109/ACCESS.2020.3007337. Saeed Anwar and Nick Barnes. Densely Residual Laplacian Super-Resolution, July

    2020

  53. [2021]

    doi:10.1109/ACCESS.2021.3127960

    ISSN 2169-3536. doi:10.1109/ACCESS.2021.3127960. Mohammadreza Amirian and Friedhelm Schwenker. Radial Basis Function Networks for Convolutional Neural Networks to Learn Similarity Distance Metric and Improve Interpretability. IEEE Access, 8:123087–123097,

    2021

  54. [2022]

    doi:10.32604/iasc.2023.029653

    ISSN 1079-8587, 2326-005X. doi:10.32604/iasc.2023.029653. Pravesh Agrawal, Szymon Antoniak, Emma Bou Hanna, Baptiste Bout, Devendra Chaplot, Jessica Chudnovsky, Diogo Costa, Baudouin De Monicault, Saurabh Garg, Theophile Gervet, Soham Ghosh, Amélie Héliou, Paul Jacob, Albert Q...

    2023

  55. [2023]

    Ensemble Model of Lanczos and Bicubic Interpolation with Neural Network and Resampling for Image Enhancement

    VERITAS Ronie C Bituin and Ronielle Antonio. Ensemble Model of Lanczos and Bicubic Interpolation with Neural Network and Resampling for Image Enhancement. In Proceedings of the 2024 7th International Conference on Software Engineer- ing and Information Management, pages 110–11...

    2024

  56. [2024]

    Exploring the Impact of Real and Synthetic Data in Image Classification: A Comprehensive Investigation Using CIFAKE Dataset

    Amila Akagic, Emir Buza, Medina Kapo, and Mahdi Bohlouli. Exploring the Impact of Real and Synthetic Data in Image Classification: A Comprehensive Investigation Using CIFAKE Dataset. In 2024 10th Interna- tional Conference on Control, Decision and Information Technologies (CoD...

    2024

  57. [2025]

    arXiv:2504.10727 [cs]

    URL http://arxiv.org/abs/ 2504.10727. arXiv:2504.10727 [cs]. Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep Residual Learning for Image Recognition, December

    arXiv

Pith tools

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