Pith. sign in

REVIEW 5 major objections 6 minor 68 references

LAID: Lightweight AI-Generated Image Detection in Spatial and Spectral Domains

T0 review · 5 major / 6 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 Lightweight neural networks under 10 million parameters can detect AI-generated images with over 99% accuracy in the spectral domain, at a fraction of the compute of current detectors.

desk verdict Useful clean-domain lightweight AIGI benchmark, but the robustness claims in the abstract and conclusion are contradicted by the paper's own blur results. read the letter →

arxiv 2507.05162 v1 pith:JCQ7KKTA submitted 2025-07-07 cs.CV cs.AIcs.CR

Nicholas Chivaran , Jianbing Ni This is my paper
classification cs.CVcs.AIcs.CR
keywords AI-generatedimagedetectionlightweightneuralnetworksspectraldomainanalysisadversarialrobustnessmodelefficiencyGendatasetvisiontransformersconvolutional
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 tries to establish that heavy, high-parameter detectors are not necessary for practical AI-generated image detection. It benchmarks off-the-shelf lightweight models under 10 million parameters or 1 GFLOP per inference on spatial, spectral, and fused inputs drawn from a subset of the GenImage dataset. Reported clean accuracy exceeds 99% for most models in the spectral domain and 92-99% spatially, while spectral inputs collapse to chance under adversarial corruption and fusion recovers much of the lost robustness. The payoff, if the claims hold, is that real-time large-scale screening on social-media workloads could use small models with a small fraction of the memory and compute of current state-of-the-art detectors.

What carries the argument

The carrying mechanism is LAID's two-branch evaluation: an architecturally identical lightweight model is trained separately on raw spatial pixels and on zero-centered 2D-FFT spectral images, and at test time the two predictions are combined by decision-level OR-fusion. Model selection uses an efficiency score $E = \lambda_1 \frac{\text{Acc}}{\max \text{Acc}} + \lambda_2 \frac{\min \text{FLOPs}}{\text{FLOPs}} + \lambda_3 \frac{\min \text{Params}}{\text{Params}}$ with $\lambda = (0.5, 0.25, 0.25)$, after filtering candidates to fewer than 10 million parameters or under 1 GFLOP per inference. The spectral branch carries the clean-detection performance; the spatial branch carries most of the robustness; the fusion step combines these complementary failure modes.

What would settle it

Run the same LAID models under model-aware adversarial perturbations that a determined attacker could craft, such as small gradient-based pixel changes that preserve visual quality, and compare lightweight accuracy with heavy state-of-the-art baselines on the same GenImage subset; if lightweight accuracy falls to near chance while heavy models stay high, the claim that lightweight models remain competitive under adversarial conditions would not extend to adaptive adversaries.

Watch

Extended reading notes

Core claim

The central claim is that lightweight models are competitive with state-of-the-art AIGI detectors when given the right input representation. LAID trains each selected model twice, once on raw pixels and once on zero-centered 2D-FFT spectral images, then combines predictions by OR-fusion at test time. In clean settings, spatial accuracy ranges from 78% (MnasNet) to 99.22% for a custom lightweight CNN baseline, with most models above 92%, while spectral accuracy exceeds 99% for every model except MnasNet. Under adversarial perturbations (cropping, Gaussian blur, Gaussian noise, JPEG compression, and their combination), spectral accuracy drops to near 50% for all models, but fusion with the spatial branch yields many accuracies above 90% under noise and compression. The paper also reports that lightweight CNNs tolerate JPEG compression better than lightweight vision transformers, and that ShuffleNet gives the best accuracy per parameter and per FLOP.

Load-bearing premise

The paper's robustness claim rests on assuming that five common image corruptions—cropping, blurring, noise, JPEG compression, and their combination—stand in for the 'adversarial conditions' a deployed detector will face; these are ordinary post-processing transforms, not perturbations crafted to fool a specific model.

Editorial extensions

If this is right

  • Clean spectral detection with lightweight models exceeds 99% accuracy for almost every architecture tested, so frequency-domain input alone can make small models sufficient for clean detection.
  • Under noise and JPEG compression, fusing spatial and spectral predictions restores many accuracies above 90%, suggesting a cheap OR-fusion rule is an effective robustness strategy.
  • Lightweight CNNs such as ShuffleNet, MobileNetV3, and SqueezeNet keep 80%+ accuracy under JPEG compression while lightweight vision transformers fall to 65% or below, so architecture choice matters for compression-heavy deployments.
  • Efficiency metrics show ShuffleNet delivers the highest accuracy per parameter and per FLOP, while the two heavier baselines are dramatically less efficient, which favors lightweight models for resource-constrained platforms.
  • Spectral-only models collapse to chance under every tested adversarial perturbation, so spectral input alone is not a robust deployment choice.

Reading between the lines

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

  • An implication the paper leaves implicit is that its robustness evidence covers common post-processing, not adaptive adversaries; gradient-based attacks optimized against the detector would be a stronger and untested challenge.
  • Because spectral input alone collapses under any spatial transform, the clean spectral advantage looks fragile, which suggests spectral information is best used as a fusion feature or with adversarial training rather than as a standalone input.
  • The efficiency score's weighting ($\lambda = (0.5, 0.25, 0.25)$) is one choice among many, and the paper itself notes that alternative weightings could reorder the models; a natural extension is to treat the score as a tunable selection tool and validate it on other datasets.
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 / 6 minor

Summary. The paper introduces LAID, a benchmark and evaluation framework for AI-generated image (AIGI) detection using off-the-shelf lightweight neural networks. The authors select ten lightweight models via an efficiency score that combines ImageNet top-1 accuracy, FLOPs, and parameter count, then fine-tune each model independently on spatial images and on zero-centered 2D FFT spectral representations from a representative subset of the GenImage dataset. They report clean detection accuracy, adversarial robustness under five perturbations (cropping, Gaussian blur, Gaussian noise, JPEG compression, and their combination), and efficiency metrics. Clean results show high performance (most models above 92% spatial accuracy and above 99% spectral accuracy), while adversarial results show near-chance accuracy under blur and a universal collapse of spectral-only models under all attacks. The paper concludes that lightweight models achieve competitive accuracy even under adversarial conditions and exhibit strong resilience to basic perturbations.

Significance. If the claims held, the paper would provide a useful practical benchmark showing that lightweight architectures are viable for large-scale, real-time AIGI detection. The study has notable strengths: the model-selection protocol is transparent, the efficiency accounting is explicit, the use of standardized model hubs improves reproducibility, the GenImage subset is clearly described, and source code is promised. However, the paper's own results contradict the robustness half of the central claim: Table III shows that all models fall to near-chance accuracy under Gaussian blur, and Section VI.B.1 explicitly acknowledges that blurring and compression cause near-random accuracy. In addition, the claim of competitiveness with state-of-the-art methods is not directly tested, because no large current SOTA detector is evaluated under the same protocol. The clean-domain efficiency results are internally consistent, but the adversarial-robustness conclusion and the SOTA-competitiveness claim are load-bearing parts of the abstract and conclusion and are not supported by the reported experiments.

major comments (5)
  1. [Section VI.B.1, Table III, vs. Section VII and Abstract] The paper's central robustness claim is contradicted by its own results. Under Gaussian blur (kernel size 3-9, sigma 1-4), every evaluated model in every input domain falls to near-chance accuracy: spatial accuracies are 51-56%, spectral accuracies are approximately 50%, and the best fusion accuracy is 56.50% (FastViT). Section VI.B.1 explicitly states that 'Blurring and compression attacks lead to the greatest performance degradation with accuracies dropping to near-random levels,' yet Section VII concludes that 'LiMs exhibit strong resilience to basic adversarial perturbations' and the Abstract claims 'competitive accuracy, even under adversarial conditions.' Since Gaussian blur is one of the five basic perturbations the authors themselves selected and is a routine social-media post-processing operation, the robustness half of the central claim is not merely unverified; it is contradicted by the reported experiments. The abstract, conclusion, and contribution list should be revised to reflect the actual findings, or the adversarial evaluation should be re-designed to distinguish mild from destructive perturbation strengths.
  2. [Section VI.A and Section V.C] The claim that lightweight models are 'competitive' with state-of-the-art AIGI detectors is not directly tested. No large current SOTA detector (e.g., DIRE, LGrad, BiHPF, CNNSpot) is run under the same training and evaluation protocol. The two baselines, Ladevic et al. and SpottingDiffusion, are themselves lightweight methods, and one has 13.46 GFLOPs per sample while the other has 23M parameters, so they do not represent the heavy SOTA architectures the paper criticizes. Accuracy comparisons to numbers reported in prior literature are not controlled because datasets, splits, and preprocessing differ. To support the 'competitive with SOTA' wording, the authors should either include head-to-head experiments under a shared protocol with representative SOTA detectors or narrow the claim to competitiveness with lightweight baselines.
  3. [Section V.D, Eq. (5), and Table III] The fusion evaluation results are interpreted in a way that overstates the contribution of the spectral branch. Under the fusion success rule (yp = G) ∨ (yf = G), fusion accuracy should be at least the maximum of the two component accuracies. Many fusion results are far above that maximum (e.g., MobileViT under compression: 99.50% fusion versus 65.06% spatial and approximately 50% spectral), which indicates that the spatial and spectral errors are complementary. However, because all spectral-only models are near chance under every attack, the robust performance of fusion is entirely carried by the spatial model; the spectral branch contributes no signal of its own. The statement in Section VI.B.3 that fusion provides 'substantial improvement in adversarial robustness' and the suggestion that spectral features help in adversarial settings should be re-framed accordingly.
  4. [Section IV.A and Section V.B] The spectral representation is not specified sufficiently for reproducibility. The paper says spectral images are 'zero-centered 2D Fast Fourier Transforms' that are 'strictly normalized to [0, 255]', but it does not state how the complex FFT output is converted to a real-valued image, whether magnitude or log-magnitude is used, how the dynamic range is scaled, or how color channels are handled. Since the spectral domain is central to the paper's clean-performance results (Table II), this missing detail is load-bearing for reproducing the reported >99% spectral accuracies.
  5. [Section V and Table II/III] All experimental results come from a single training run and a single fixed split, with no error bars, confidence intervals, or multiple seeds. The paper makes comparative statements about model rankings and about differences between LiViTs and LiCNNs that are based on small accuracy margins (e.g., Table II spatial accuracies 92-99%). Without variance estimates, these comparisons are not statistically grounded. At minimum, key results should be reported over at least three random seeds or with appropriate confidence intervals.
minor comments (6)
  1. [Section V.D, Eqs. (4)-(5)] In the adversarial testing formulas, the spectral evaluation uses Mp instead of Mf: yf = Mp(Îadv) and the fusion equation also use Mp for the spectral branch. These should be Mf(Îadv) to match the text.
  2. [Throughout] The notation is inconsistent between 'FLOPs' and 'FLOPS'; the standard form 'FLOPs' should be used consistently.
  3. [Section VI.C] There is a typo in the sentence 'both the relatively narrow performance gape between LiMs'; 'gape' should be 'gap'.
  4. [Section V.E] The cropping attack description contains a duplicated word: 'resizes the image to to 256 × 256 pixels'.
  5. [Section VI.B.3] The text states that 'noising and other spectral attacks target an image's frequency distribution,' but the noising attack in Section V.E adds Gaussian noise in the spatial domain; calling it a spectral attack is imprecise.
  6. [Figure 4] The R² values in Figure 4 are computed from only eight data points (after omissions); the caption should state the number of points and the regression method so readers can judge the strength of the trends.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: LAID reports direct benchmark measurements; the efficiency score is defined from pre-existing model attributes and is not fitted to the AIGI test results.

full rationale

The paper's central claims are empirical benchmark results, not derivations from a fitted parameter. The efficiency score E in Eq. (1) combines ImageNet top-1 accuracy, FLOPs, and parameter count, all known before any AIGI detection training; it selects which off-the-shelf models to evaluate, but the reported clean and adversarial accuracies (Tables II and III) are direct measurements on held-out GenImage subsets. No equation in the paper defines the reported detection accuracy in terms of E or any other fitted quantity, so the selection criterion and the measured outcome are not circularly related. The paper does cite its own community's prior work, but all load-bearing baselines (Ladevic et al., SpottingDiffusion) are external methods with independent implementations, and no uniqueness theorem or ansatz from the authors' prior work is invoked to force the conclusions. The abstract and conclusion overstate robustness relative to the paper's own Gaussian-blur results (near-chance accuracy in Table III), and the spectral models collapse under attack; that is an internal-consistency or correctness problem, not a circularity problem, because the claims are not equivalent to the inputs by construction. The benchmark methodology is self-contained and externally checkable against the released code, so the circularity score is 0.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central claim rests on domain assumptions rather than mathematical derivation: the FFT artifact is a reliable signal, the GenImage subsample is representative, the chosen perturbations define 'adversarial' conditions, and ImageNet pretraining transfers. None are proved in the paper; all are supported by prior work or plausibility. There are no invented physical entities.

free parameters (2)
  • Efficiency score weights (lambda_1, lambda_2, lambda_3) = 0.5, 0.25, 0.25
    Hand-chosen in Eq. (1) to balance ImageNet accuracy, FLOPs, and parameter count. They determine which models enter the benchmark and influence the reported R^2 values, but are not fitted to AIGI detection data.
  • Adversarial attack strength ranges = Crop 5-20%; blur kernel 3-9, sigma 1-4; noise variance 5-20; JPEG quality 25-90; combination at 50% probability
    Selected from cited literature to simulate social-media transforms. These ranges define what 'adversarial conditions' means in this paper and directly shape the robustness numbers.
assumptions (5)
  • domain assumption FFT-based spectral representation exposes detectable artifacts in AI-generated images
    Invoked in Section IV.A and supported by cited prior works [10]-[14]. If the artifact is generator-specific, spectral accuracy may not generalize to new generators.
  • domain assumption The randomly subsampled GenImage subset is representative of the full dataset
    Section V.A subsamples 100,000 training images. Bias in the subset could inflate or deflate reported accuracies.
  • domain assumption The perturbation suite in Section V.E approximates real-world adversarial conditions
    The five transforms are assumed to be the relevant failure modes for social-media deployment. They are not model-specific adversarial examples, so the robustness claim is limited to these transforms.
  • domain assumption ImageNet-1K pretrained weights transfer to AIGI detection
    All models are initialized with ImageNet-1K weights (Section IV.C). If transfer is weak for some architectures, their measured performance may not reflect their capacity when trained from scratch.
  • domain assumption Binary cross-entropy with Adam and early stopping is adequate for the detection task
    The single training recipe in Section IV.C is used for all models; different hyperparameters could change the rank ordering, though likely not the main conclusion.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LAID: Lightweight AI-Generated Image Detection in Spatial and Spectral Domains." pith.science (2026). https://pith.science/paper/JCQ7KKTA

@misc{pith2026250705162,
  author       = {Pith},
  title        = {Pith review of: LAID: Lightweight AI-Generated Image Detection in Spatial and Spectral Domains},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JCQ7KKTA}},
  note         = {Machine review of arXiv:2507.05162}
}
read the original abstract

The recent proliferation of photorealistic AI-generated images (AIGI) has raised urgent concerns about their potential misuse, particularly on social media platforms. Current state-of-the-art AIGI detection methods typically rely on large, deep neural architectures, creating significant computational barriers to real-time, large-scale deployment on platforms like social media. To challenge this reliance on computationally intensive models, we introduce LAID, the first framework -- to our knowledge -- that benchmarks and evaluates the detection performance and efficiency of off-the-shelf lightweight neural networks. In this framework, we comprehensively train and evaluate selected models on a representative subset of the GenImage dataset across spatial, spectral, and fusion image domains. Our results demonstrate that lightweight models can achieve competitive accuracy, even under adversarial conditions, while incurring substantially lower memory and computation costs compared to current state-of-the-art methods. This study offers valuable insight into the trade-off between efficiency and performance in AIGI detection and lays a foundation for the development of practical, scalable, and trustworthy detection systems. The source code of LAID can be found at: https://github.com/nchivar/LAID.

Figures

Figures reproduced from arXiv: 2507.05162 by the authors.

Figure 1
Figure 1. LAID training pipeline. Testing removes binary cross-entropy (BCE) and additionally evaluates input images ( [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Image from spatial dataset and its corresponding zero [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

68 extracted references · 40 canonical work pages

  1. [1]

    Seeing is not always believing: Benchmarking human and model perception of ai-generated images,

    Z. Lu, D. Huang, L. Bai, J. Qu, C. Wu, X. Liu, and W. Ouyang, “Seeing is not always believing: Benchmarking human and model perception of ai-generated images,” pp. 25 435–25 447, 2023

    work page 2023

  2. [2]

    Crafting syn- thetic realities: Examining visual realism and misinformation potential of photorealistic ai-generated images,

    Q. Peng, Y . Lu, Y . Peng, S. Qian, X. Liu, and C. Shen, “Crafting syn- thetic realities: Examining visual realism and misinformation potential of photorealistic ai-generated images,” pp. 1–12, 2025

    work page 2025

  3. [3]

    How spammers and scammers leverage ai-generated images on facebook for audience growth,

    R. DiResta and J. A. Goldstein, “How spammers and scammers leverage ai-generated images on facebook for audience growth,” 2024

    work page 2024

  4. [4]

    99 amazing social media statistics and facts,

    R. Tsvetkova, “99 amazing social media statistics and facts,” https://www.brandwatch.com/blog/amazing-social-media-statistics-and- facts/, 2023

    work page 2023

  5. [5]

    3.2 billion images and 720,000 hours of video are shared online daily. can you sort real from fake?

    P. Dootson, “3.2 billion images and 720,000 hours of video are shared online daily. can you sort real from fake?” https://www.qut.edu.au/insights/business/3.2-billion-images-and- 720000-hours-of-video-are-shared-online-daily.-can-you-sort-real- from-fake, 2021, Queensland University of Technology

    work page 2021

  6. [6]

    Photo statistics: How many photos are taken every day?

    M. Broz, “Photo statistics: How many photos are taken every day?” https://photutorial.com/photos-statistics/, 2025

    work page 2025

  7. [7]

    Labeling ai-generated images on facebook, instagram and threads,

    N. Clegg, “Labeling ai-generated images on facebook, instagram and threads,” https://about.fb.com/news/2024/02/labeling-ai-generated- images-on-facebook-instagram-and-threads/, 2024

    work page 2024

  8. [8]

    New labels for disclosing ai-generated content,

    TikTok Newsroom, “New labels for disclosing ai-generated content,” https://newsroom.tiktok.com/en-us/new-labels-for-disclosing-ai- generated-content, 2023

    work page 2023

Show all 68 references
  1. [9]

    Authenticity,

    X Help Center, “Authenticity,” https://help.x.com/en/rules-and- policies/authenticity, 2025

    2025

  2. [10]

    Ai-generated image detection: Passive or watermark?

    M. Guo, Y . Hu, Z. Jiang, Z. Li, A. Sadovnik, A. Daw, and N. Gong, “Ai-generated image detection: Passive or watermark?” 2024

    2024

  3. [11]

    Watch your up-convolution: Cnn based generative deep neural networks are failing to reproduce spectral distributions,

    R. Durall, M. Keuper, and J. Keuper, “Watch your up-convolution: Cnn based generative deep neural networks are failing to reproduce spectral distributions,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 7890–7899

    2020

  4. [12]

    Spectral leakage and rethinking the kernel size in cnns,

    N. Tomen and J. C. van Gemert, “Spectral leakage and rethinking the kernel size in cnns,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 5138–5147

    2021

  5. [13]

    Beyond the spectrum: Detecting deepfakes via re-synthesis,

    Y . He, N. Yu, M. Keuper, and M. Fritz, “Beyond the spectrum: Detecting deepfakes via re-synthesis,” 2021

    2021

  6. [14]

    Bihpf: Bilateral high-pass filters for robust deepfake detection,

    Y . Jeong, D. Kim, S. Min, S. Joe, Y . Gwon, and J. Choi, “Bihpf: Bilateral high-pass filters for robust deepfake detection,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2022, pp. 48–57

    2022

  7. [15]

    A survey of defenses against ai-generated visual media: Detection, disruption, and authentication,

    J. Deng, C. Lin, Z. Zhao, S. Liu, Q. Wang, and C. Shen, “A survey of defenses against ai-generated visual media: Detection, disruption, and authentication,” 2024

    2024

  8. [16]

    Unmarker: A universal attack on defen- sive watermarking,

    A. Kassis and U. Hengartner, “Unmarker: A universal attack on defen- sive watermarking,” arXiv preprint arXiv:2405.08363 , 2024

    2024 arXiv

  9. [17]

    Hidden: Hiding data with deep networks,

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

    2018

  10. [18]

    The stable signature: Rooting watermarks in latent diffusion models,

    P. Fernandez, G. Couairon, H. Jégou, M. Douze, and T. Furon, “The stable signature: Rooting watermarks in latent diffusion models,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 22 466–22 477

    2023

  11. [19]

    Wouaf: Weight modulation for user attribution and fingerprinting in text-to-image diffu- sion models,

    C. Kim, K. Min, M. Patel, S. Cheng, and Y . Yang, “Wouaf: Weight modulation for user attribution and fingerprinting in text-to-image diffu- sion models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 8974–8983

    2024

  12. [20]

    Gaussian shading: Provable performance-lossless image watermarking for diffu- sion models,

    Z. Yang, K. Zeng, K. Chen, H. Fang, W. Zhang, and N. Yu, “Gaussian shading: Provable performance-lossless image watermarking for diffu- sion models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 12 162–12 171

    2024

  13. [21]

    Tree-rings watermarks: Invisible fingerprints for diffusion images,

    Y . Wen, J. Kirchenbauer, J. Geiping, and T. Goldstein, “Tree-rings watermarks: Invisible fingerprints for diffusion images,” vol. 36, 2023, pp. 58 047–58 063

    2023

  14. [22]

    Deeptag: Robust image tagging for deepfake provenance. arxiv preprint arxiv,

    R. Wang, F. Juefei-Xu, Q. Guo, Y . Huang, X. Xie, L. Ma, Y . Liu, and L. Wang, “Deeptag: Robust image tagging for deepfake provenance. arxiv preprint arxiv,” 2020

    2020

  15. [23]

    Stegastamp: Invisible hyperlinks in physical photographs,

    M. Tancik, B. Mildenhall, and R. Ng, “Stegastamp: Invisible hyperlinks in physical photographs,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 2117–2126

    2020

  16. [24]

    Faceforensics++: Learning to detect manipulated facial images,

    A. Rossler, D. Cozzolino, L. Verdoliva, C. Riess, J. Thies, and M. Nießner, “Faceforensics++: Learning to detect manipulated facial images,” in Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 1–11

    2019

  17. [25]

    Cnn- generated images are surprisingly easy to spot... for now,

    S.-Y . Wang, O. Wang, R. Zhang, A. Owens, and A. A. Efros, “Cnn- generated images are surprisingly easy to spot... for now,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 8695–8704

    2020

  18. [26]

    Learning on gradients: Generalized artifacts representation for gan-generated images detection,

    C. Tan, Y . Zhao, S. Wei, G. Gu, and Y . Wei, “Learning on gradients: Generalized artifacts representation for gan-generated images detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 12 105–12 114

    2023

  19. [27]

    Towards universal fake image detec- tors that generalize across generative models,

    U. Ojha, Y . Li, and Y . J. Lee, “Towards universal fake image detec- tors that generalize across generative models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 24 480–24 489

    2023

  20. [28]

    Dire for diffusion-generated image detection,

    Z. Wang, J. Bao, W. Zhou, W. Wang, H. Hu, H. Chen, and H. Li, “Dire for diffusion-generated image detection,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 22 445–22 455

    2023

  21. [29]

    Patchcraft: Exploring texture patch for efficient ai-generated image detection,

    N. Zhong, Y . Xu, S. Li, Z. Qian, and X. Zhang, “Patchcraft: Exploring texture patch for efficient ai-generated image detection,” arXiv preprint arXiv:2311.12397, 2023

    2023 arXiv

  22. [30]

    Leveraging frequency analysis for deep fake image recogni- tion,

    J. Frank, T. Eisenhofer, L. Schönherr, A. Fischer, D. Kolossa, and T. Holz, “Leveraging frequency analysis for deep fake image recogni- tion,” in International conference on machine learning . PMLR, 2020, pp. 3247–3258

    2020

  23. [31]

    Attributing fake images to gans: Learn- ing and analyzing gan fingerprints,

    N. Yu, L. S. Davis, and M. Fritz, “Attributing fake images to gans: Learn- ing and analyzing gan fingerprints,” in Proceedings of the IEEE/CVF international conference on computer vision , 2019, pp. 7556–7566

    2019

  24. [32]

    A sanity check for ai-generated image detection,

    S. Yan, O. Li, J. Cai, Y . Hao, X. Jiang, Y . Hu, and W. Xie, “A sanity check for ai-generated image detection,” arXiv preprint arXiv:2406.19435, 2024

    2024 arXiv

  25. [33]

    De-fake: Detection and attribution of fake images generated by text-to-image generation models,

    Z. Sha, Z. Li, N. Yu, and Y . Zhang, “De-fake: Detection and attribution of fake images generated by text-to-image generation models,” in Proceedings of the 2023 ACM SIGSAC conference on computer and communications security, 2023, pp. 3418–3432

    2023

  26. [34]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark et al., “Learning transferable visual models from natural language supervision,” in International conference on machine learning . PmLR, 2021, pp. 8748–8763

    2021

  27. [35]

    Rethinking the inception architecture for computer vision,

    C. Szegedy, V . Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, “Rethinking the inception architecture for computer vision,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 2818–2826

    2016

  28. [36]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929 , 2020

    2010 arXiv

  29. [37]

    Fakespotter: A simple yet robust baseline for spotting ai-synthesized fake faces,

    R. Wang, F. Juefei-Xu, L. Ma, X. Xie, Y . Huang, J. Wang, and Y . Liu, “Fakespotter: A simple yet robust baseline for spotting ai-synthesized fake faces,” 2019

    2019

  30. [38]

    keras-vggface: Vggface implementation with keras frame- work,

    R. C. Malli, “keras-vggface: Vggface implementation with keras frame- work,” https://github.com/rcmalli/keras-vggface, 2017

    2017

  31. [39]

    Detec- tion of ai-generated synthetic images with a lightweight cnn,

    A. L. La ¯devi´c, T. Kramberger, R. Kramberger, and D. Vlahek, “Detec- tion of ai-generated synthetic images with a lightweight cnn,” AI, vol. 5, no. 3, p. 1575, 2024

    2024

  32. [40]

    Spottingdiffusion:re using transfer learning to detect latent diffusion model-synthesized images,

    M. Mulki and S. Mulki, “Spottingdiffusion:re using transfer learning to detect latent diffusion model-synthesized images,” Journal of Emerging Investigators, 01 2024

    2024

  33. [41]

    High- resolution image synthesis with latent diffusion models,

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, “High- resolution image synthesis with latent diffusion models,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 10 684–10 695

    2022

  34. [42]

    Midjourney,

    MidJourney, Inc., “Midjourney,” https://www.midjourney.com/, 2022

    2022

  35. [43]

    Zero-shot text-to-image generation,

    A. Ramesh, M. Pavlov, G. Goh, S. Gray, C. V oss, A. Radford, M. Chen, and I. Sutskever, “Zero-shot text-to-image generation,” in International conference on machine learning . Pmlr, 2021, pp. 8821–8831

    2021

  36. [44]

    Genimage: A million-scale benchmark for detecting ai- generated image,

    M. Zhu, H. Chen, Q. Yan, X. Huang, G. Lin, W. Li, Z. Tu, H. Hu, J. Hu, and Y . Wang, “Genimage: A million-scale benchmark for detecting ai- generated image,” Advances in Neural Information Processing Systems , vol. 36, pp. 77 771–77 782, 2023

    2023

  37. [45]

    Diffusion models beat gans on image synthesis,

    P. Dhariwal and A. Nichol, “Diffusion models beat gans on image synthesis,” Advances in neural information processing systems , vol. 34, pp. 8780–8794, 2021

    2021

  38. [46]

    Glide: Towards photorealistic image gen- eration and editing with text-guided diffusion models,

    A. Nichol, P. Dhariwal, A. Ramesh, P. Shyam, P. Mishkin, B. McGrew, I. Sutskever, and M. Chen, “Glide: Towards photorealistic image gen- eration and editing with text-guided diffusion models,” arXiv preprint arXiv:2112.10741, 2021

    2021 arXiv

  39. [47]

    MindSpore ModelZoo, “Wukong,” https://xihe.mindspore.cn/modelzoo/wukong, 2022

    2022

  40. [48]

    Accurate compression of text-to-image diffu-sion models via vector quantization,

    V . Egiazarian, D. Kuznedelev, A. V oronov, R. Svirschevski, M. Goin, D. Pavlov, D. Alistarh, and D. Baranchuk, “Accurate compression of text-to-image diffu-sion models via vector quantization,” arXiv preprint arXiv:2409.00492, 2024

    2024 arXiv

  41. [49]

    Large scale gan training for high fidelity natural image synthesis,

    A. Brock, J. Donahue, and K. Simonyan, “Large scale gan training for high fidelity natural image synthesis,” arXiv preprint arXiv:1809.11096, 2018

    2018 arXiv

  42. [50]

    Models and pre-trained weights,

    “Models and pre-trained weights,” https://pytorch.org/vision/stable/models.html

  43. [51]

    Pytorch image models,

    R. Wightman, “Pytorch image models,” https://github.com/rwightman/pytorch-image-models, 2019

    2019

  44. [52]

    Imagenet classification with deep convolutional neural networks,

    A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet classification with deep convolutional neural networks,” vol. 60, no. 6. AcM New York, NY , USA, 2017, pp. 84–90

    2017

  45. [53]

    Densely connected convolutional networks,

    G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger, “Densely connected convolutional networks,” in Proceedings of the IEEE confer- ence on computer vision and pattern recognition , 2017, pp. 4700–4708

    2017

  46. [54]

    Efficientnetv2: Smaller models and faster training,

    M. Tan and Q. Le, “Efficientnetv2: Smaller models and faster training,” in International conference on machine learning . PMLR, 2021, pp. 10 096–10 106

    2021

  47. [55]

    Going deeper with convolutions,

    C. Szegedy, W. Liu, Y . Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V . Vanhoucke, and A. Rabinovich, “Going deeper with convolutions,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 1–9

    2015

  48. [56]

    Mnasnet: Platform-aware neural architecture search for mobile,

    M. Tan, B. Chen, R. Pang, V . Vasudevan, M. Sandler, A. Howard, and Q. V . Le, “Mnasnet: Platform-aware neural architecture search for mobile,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2019, pp. 2820–2828

    2019

  49. [57]

    Mobilenetv2: Inverted residuals and linear bottlenecks,

    M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, “Mobilenetv2: Inverted residuals and linear bottlenecks,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2018, pp. 4510–4520

    2018

  50. [58]

    Searching for mobilenetv3,

    A. Howard, M. Sandler, G. Chu, L.-C. Chen, B. Chen, M. Tan, W. Wang, Y . Zhu, R. Pang, V . Vasudevan et al. , “Searching for mobilenetv3,” in Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 1314–1324

    2019

  51. [59]

    Regnet: self-regulated network for image classification,

    J. Xu, Y . Pan, X. Pan, S. Hoi, Z. Yi, and Z. Xu, “Regnet: self-regulated network for image classification,” IEEE Transactions on Neural Net- works and Learning Systems , vol. 34, no. 11, pp. 9562–9567, 2022

    2022

  52. [60]

    Shufflenet: An extremely effi- cient convolutional neural network for mobile devices,

    X. Zhang, X. Zhou, M. Lin, and J. Sun, “Shufflenet: An extremely effi- cient convolutional neural network for mobile devices,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2018, pp. 6848–6856

    2018

  53. [61]

    Squeezenet: Alexnet-level accuracy with 50x fewer parameters and< 0.5 mb model size,

    F. N. Iandola, S. Han, M. W. Moskewicz, K. Ashraf, W. J. Dally, and K. Keutzer, “Squeezenet: Alexnet-level accuracy with 50x fewer parameters and< 0.5 mb model size,” arXiv preprint arXiv:1602.07360, 2016

    2016 arXiv

  54. [62]

    Co-scale conv-attentional image transformers,

    W. Xu, Y . Xu, T. Chang, and Z. Tu, “Co-scale conv-attentional image transformers,” in Proceedings of the IEEE/CVF international conference on computer vision , 2021, pp. 9981–9990

    2021

  55. [63]

    Edgenext: efficiently amalgamated cnn- transformer architecture for mobile vision applications,

    M. Maaz, A. Shaker, H. Cholakkal, S. Khan, S. W. Zamir, R. M. Anwer, and F. Shahbaz Khan, “Edgenext: efficiently amalgamated cnn- transformer architecture for mobile vision applications,” in European conference on computer vision . Springer, 2022, pp. 3–20

    2022

  56. [64]

    Efficientformer: Vision transformers at mobilenet speed,

    Y . Li, G. Yuan, Y . Wen, J. Hu, G. Evangelidis, S. Tulyakov, Y . Wang, and J. Ren, “Efficientformer: Vision transformers at mobilenet speed,” Ad- vances in Neural Information Processing Systems , vol. 35, pp. 12 934– 12 949, 2022

    2022

  57. [65]

    Fastvit: A fast hybrid vision transformer using structural reparameterization,

    P. K. A. Vasu, J. Gabriel, J. Zhu, O. Tuzel, and A. Ranjan, “Fastvit: A fast hybrid vision transformer using structural reparameterization,” in Proceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 5785–5795

    2023

  58. [66]

    Levit: a vision transformer in convnet’s clothing for faster inference,

    B. Graham, A. El-Nouby, H. Touvron, P. Stock, A. Joulin, H. Jégou, and M. Douze, “Levit: a vision transformer in convnet’s clothing for faster inference,” in Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 12 259–12 269

    2021

  59. [67]

    Mobilevit: light-weight, general- purpose, and mobile-friendly vision transformer,

    S. Mehta and M. Rastegari, “Mobilevit: light-weight, general- purpose, and mobile-friendly vision transformer,” arXiv preprint arXiv:2110.02178, 2021

    2021 arXiv

  60. [68]

    Separable self-attention for mobile vision transformers,

    ——, “Separable self-attention for mobile vision transformers,” arXiv preprint arXiv:2206.02680, 2022

    2022 arXiv

Pith tools

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