Enhancing Deepfake Detection using SE Block Attention with CNN

Janelle Mason; Kaushik Roy; Olusola Odeyomi; Subhram Dasgupta; Xiaohong Yuan

REVIEW 4 major objections 6 minor 21 references

Enhancing Deepfake Detection using SE Block Attention with CNN

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A small CNN with squeeze-and-excitation attention spots deepfakes at 94.14% accuracy, matching the AUC of a model 30 times its size.

desk verdict A plausible but irreproducible small SE+CNN deepfake detector; the evaluation protocol is contradictory and the cross-dataset comparison is invalid. read the letter →

arxiv 2506.10683 v1 pith:VYEYYL5V submitted 2025-06-12 cs.CV

Subhram Dasgupta , Janelle Mason , Xiaohong Yuan , Olusola Odeyomi , Kaushik Roy This is my paper · ORCID

classification cs.CV

keywords DeepfakedetectionSqueeze-and-ExcitationattentionmechanismlightweightCNNStyleGANDiverseFakeFaceDatasetchannel-wisefeaturerecalibrationforgery

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

This paper claims that a deliberately small convolutional network, upgraded with squeeze-and-excitation (SE) attention blocks, can detect AI-generated fake faces almost as reliably as much larger deepfake detectors. On a subset of the Diverse Fake Face Dataset built from StyleGAN images and Flickr-Faces-HQ real faces, the model reports 94.14% accuracy and a 0.985 AUC-ROC score. The design point is efficiency: at 56.18 MB, the model is roughly 30 times smaller than a competing detector with the same AUC. The paper also reports that the SE version outperforms its own plain-CNN baseline by about three percentage points of accuracy, and the authors argue that channel-wise recalibration is what lets a small model focus on subtle generation artifacts. That combination of near-state-of-the-art accuracy in a deployable-size model is the reason a sympathetic reader would care.

What carries the argument

The load-bearing component is the Squeeze-and-Excitation (SE) block, a channel-attention module borrowed from image classification. It computes a per-channel descriptor via global average pooling, then uses two dense layers to produce a set of channel weights, and rescales the input feature maps by those weights. The authors place the block after batch normalization in the four deepest convolution stages, so the network can emphasize informative channels and suppress irrelevant ones at each scale. The rest of the architecture is a conventional sequential CNN with 3x3 kernels, ReLU activation, He initialization, max pooling, and a final dense classifier.

What would settle it

Train the exact CNN+SE architecture and the plain CNN on the same 38,000-image StyleGAN/FFHQ subset under a fixed 80/20 split with multiple seeds. If the SE model's mean accuracy does not exceed the plain CNN's by roughly three percentage points in non-overlapping confidence intervals, or if the 94.14% accuracy and 0.985 AUC do not reproduce, the central claim is refuted.

Watch

Extended reading notes

Core claim

The central claim is that inserting SE blocks into a five-layer sequential CNN turns a modest deepfake detector into one that matches the AUC of a much heavier solution. The authors' model processes 224x224 face images through five convolutional layers with 32 to 512 filters, batch normalization, max pooling, and an SE block after normalization in four of the layers; the SE block squeezes each channel to a scalar by global average pooling, passes it through two dense layers (ReLU then sigmoid), and rescales the original feature map channel by channel. On 38,000 balanced StyleGAN/FFHQ images, this pipeline yields 94.14% accuracy and per-class AUC-ROC of 0.99 (fake) and 0.98 (real), versus 91.13% accuracy for the same CNN without SE. The authors interpret the gain as evidence that channel-wise recalibration captures manipulation-specific texture cues, and that a lightweight architecture can thereby achieve results comparable to existing cutting-edge methodologies.

Load-bearing premise

The performance claim rests on the assumption that the reported evaluation is well-defined and reproducible; since the paper states both an 80/20 split and 10-fold cross-validation without saying which produced the numbers, and reports no variance, the 94.14% accuracy and the three-point gain over the baseline cannot be verified as stated.

Editorial extensions

If this is right

A 56 MB model reaching 0.985 AUC on StyleGAN faces suggests that lightweight detectors can be competitive with much larger models, making on-device and real-time deepfake screening feasible.
The three-point accuracy gain over the plain CNN attributes a concrete, reproducible benefit to SE blocks in this detection task, not just in image classification.
If the result transfers across generators, channel-wise recalibration could become a standard cheap addition to forensic classifiers.
The reported model size reduction from 1722.85 MB to 56.18 MB at equal AUC implies that most parameters in large detectors are unnecessary for this data distribution.

Reading between the lines

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

The headline comparison is cross-dataset: the 0.985 AUC for the proposed model comes from StyleGAN/DFFD, while the WM baseline's 0.985 comes from the DFDC video dataset; matching numbers do not imply matching performance on a shared benchmark, and a head-to-head test on DFDC or FaceForensics++ would be needed to validate the efficiency claim.
The paper reports both an 80/20 split and 10-fold cross-validation without stating which produced the reported numbers; if the 94.14% figure is from a single split, the variance across folds could easily exceed the three-point gap over the baseline, so the superiority of the SE version is not yet established.
The authors' stated plan to test CBAM and scaled dot-product attention on more datasets suggests the same experimental setup could be used to identify which attention mechanism gives the best accuracy-per-parameter trade-off for deepfake detection.

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.

Desk Editor's Note

read the letter

Quick read: this is a small empirical study that puts SE blocks into a simple CNN and tests it on the StyleGAN subset of DFFD. The headline numbers are plausible, but as written the evaluation protocol is internally contradictory, so I can't tell which split produced the 94.14% accuracy and 0.985 AUC. That alone makes the central claim unreproducible.

What's good: The paper is honest about its building blocks. It cites Hu et al. for SE, and it explicitly mentions Roy et al.'s prior use of SE blocks for deepfake detection on FaceForensics++. The architecture is described in enough detail to reimplement: five conv layers with 32–512 filters, SE block after batch norm, max pooling, Adam, categorical cross-entropy. The baseline comparison to a plain CNN without SE is the right kind of ablation. The model-size comparison in Table II is at least a real number.

The soft spots are bigger. Section IV.A says both '80% for training and 20% for testing' and 'trained for 100 epochs with 10-fold cross-validation.' Those are different protocols. No per-fold or per-seed numbers are given, so the 3-point gap between CNN (91.13%) and CNN+SE (94.14%) is within normal stochasticity for a 100-epoch CNN on 38k images. I can't attribute it to the SE block. Table II compares AUC numbers across datasets—DFDC, 140k, and the StyleGAN subset of DFFD—as if they were interchangeable. They are not. Matching WM's 98.5 AUC on DFDC with your 98.5 on StyleGAN/DFFD is not 'comparable performance'; it's a cross-dataset coincidence. Also, no code or data is released, and the learning rate is reported as 0.001 in Section III and 0.0001 in Section IV.A—another inconsistency.

The novelty is low: this is a reimplementation of known attention blocks on a relatively easy benchmark. That's fine for a workshop paper, but not for a strong venue.

My recommendation: don't send this to referees in its current form. The evaluation must be clarified, rerun with a single consistent protocol, and compared against baselines on the same dataset. If the authors do that, it could become a conference-workshop-tier paper. Right now the claims are not reproducible.

Referee Report

4 major / 6 minor

Summary. The paper proposes a lightweight CNN augmented with Squeeze-and-Excitation (SE) blocks for deepfake image detection, targeting a subset of the Diverse Fake Face Dataset (StyleGAN-generated fake images paired with FFHQ real images). The architecture combines five convolutional layers with SE blocks inserted after batch normalization, and the authors report 94.14% classification accuracy and an AUC-ROC of 0.985 on a held-out test set, alongside a comparison with a plain CNN baseline that achieves 91.13%. The stated contributions are a compact SE-CNN model and a demonstration that SE attention improves detection accuracy with minimal computational overhead. The paper also compares its AUC against prior work on different datasets, claiming competitive performance at a reduced model size.

Significance. If the empirical claims are reliable, the result would be a modest but useful data point: a relatively lightweight CNN with SE blocks can match heavier architectures on a deepfake detection benchmark at substantially lower model size. The paper provides a straightforward ablation (CNN vs. CNN+SE) on a public dataset, and the reported absolute numbers are plausible for the StyleGAN/FFHQ subset of DFFD. However, the current evidence is insufficient to support the central claim because the evaluation protocol is contradictory and the cross-dataset comparison in Table II is not a valid basis for a competitiveness claim. The work would be strengthened by a clear, reproducible evaluation protocol with variance estimates and by restricting comparisons to same-dataset evaluations.

major comments (4)

[Section IV.A, Dataset and Implementation] The evaluation protocol is internally contradictory: the text states both "we use train test split from sklearn to split the numpy arrays into 80% for training and 20% for testing" and "The model is trained for 100 epochs with 10-fold cross-validation." These are distinct protocols, and the reported 91.13% vs. 94.14% accuracy and the 0.985 AUC cannot be interpreted without knowing which protocol produced them, or whether the numbers are from a single split, a fold average, or a best fold. The central empirical claim is therefore not reproducible as written. Please specify a single protocol, report per-fold and per-seed results, and provide mean and standard deviation for all reported metrics.
[Table II, Comparison of Performance] Table II compares AUC values across different datasets (DFDC, 140k, StyleGAN/DFFD) and treats them as directly comparable. AUC is dataset-dependent, so the statement "our CNN with SE block demonstrates superior performance compared to existing methods" is not supported: the comparison conflates model quality with dataset difficulty and class balance. Likewise, the model-size comparison between the proposed model (56.18 MB) and WM (1722.85 MB) is only meaningful if both are measured on the same dataset and with the same counting methodology. Please either restrict comparisons to results on the same dataset or explicitly frame the table as indicative context rather than as evidence of superiority.
[Section IV.B, Results and Discussion and Abstract] The AUC reporting is inconsistent: the abstract and text report 0.985, while Section IV.B says "achieved an AUC-ROC score of 98.5" and Fig. 7 reports per-class AUCs of 0.99 (class 0) and 0.98 (class 1). If the per-class values are averaged, the result would be 0.985, but the text does not state this. More importantly, the AUC-ROC is a global scalar for a binary classifier, so per-class AUCs (which are equivalent to one-vs-rest AUCs) need to be tied to the decision threshold or probability outputs. Please clarify how the 0.985 AUC was computed and whether it corresponds to the model's probability output before thresholding.
[Table I and Section IV.A] The 3.01 percentage-point accuracy gap between CNN (91.13%) and CNN+SE (94.14%) is presented as the key benefit of the SE block, but no variance estimates are given. With 38,000 images and training for 100 epochs, the gap could be within training stochasticity (different initializations, data shuffling, and the unspecified split). Without repeated runs or significance testing, the improvement attributable to the SE block is not established. Please report results over multiple seeds (at least three) and state the standard deviation, and if 10-fold cross-validation was also run, report the fold-level accuracy distribution.

minor comments (6)

[Section IV.A] The text says the images are "re-scaled and saved as numpy arrays" and then split using sklearn; please state the exact preprocessing (e.g., pixel scaling to [0,1] or standardization) and whether any data augmentation was applied, since Fig. 3 mentions augmentation but no augmentation details are given in the implementation section.
[Section III.A] The sentence "The 3 X 3 filter is used over 1 X 1 filter or 5 X 5 filter because of its efficiency in feature learning" is unclear and seems to be a fragment; also "He Normal" should be "He normal" throughout, and "prepocessed" in Section III.C is a typo.
[Section II.B] The last sentence of Section II.B ends with "A summary of the dee" which is an incomplete sentence; the section appears to be cut off mid-word.
[Section IV.B, Fig. 7] The AUC-ROC curve should show a single curve for the binary classifier; the figure text reports separate AUCs per class, which is nonstandard for binary classification and should be clarified or corrected.
[Table II] The table header "AUC" should specify that values are percentages (e.g., 90.9 vs. 98.5), and the proposed model's AUC should be reported consistently as 0.985 or 98.5, not both.
[Overall] The paper compares the proposed model with only a handful of prior methods and omits several standard baselines on DFFD (e.g., the original DFFD paper reports on the same dataset); citing and comparing with same-dataset baselines would substantially strengthen the evaluation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's SE-block equations are quoted from prior work, and the headline accuracy is an empirical held-out evaluation, not a quantity equivalent by construction to a fitted input.

full rationale

This paper is an empirical CNN training-and-evaluation study, not a derivation. The SE block equations (2)-(5) are explicitly quoted from Hu et al. [11] and are presented as components of the architecture, not as predictions derived from first principles. The central claims, 94.14% accuracy and AUC-ROC 0.985, are measured on a test portion of the StyleGAN/DFFD dataset after training, so they are not equivalent by construction to any fitted parameter or training loss. The comparison with the non-SE CNN baseline is also an experimental outcome rather than a definitional identity. No step in the paper defines its input in terms of its output, and no fitted parameter is renamed as a prediction. The only cited prior work with an overlapping author name, Roy et al. [17], is not load-bearing: the present paper does not rely on that citation to justify its architecture or results. The internally inconsistent evaluation protocol in Section IV.A (80/20 split versus 10-fold cross-validation) and the cross-dataset AUC comparison in Table II are genuine reproducibility and validity concerns, but they are not circularity. They do not make the reported result reduce to its own assumptions; they make it under-specified. Under the hard rules, a non-finding is appropriate because no quoted reduction or self-citation chain forces the central claim.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new theoretical entities are introduced; the paper applies an existing attention mechanism to a dataset. The free parameters are standard training hyperparameters, and the axioms are standard assumptions about the dataset and training procedure.

free parameters (5)

SE reduction ratio = not reported
The SE block has a bottleneck ratio in the excitation MLP; the paper never states it, yet it changes the model's capacity and is a hand-chosen hyperparameter.
Learning rate = 0.0001
Hand-selected with Adam; no sensitivity analysis given.
Batch size = 64
Hand-selected; no justification provided.
Number of epochs = 100
Hand-selected; combined with 10-fold cross-validation, which is ambiguous.
Filter counts = 32, 64, 128, 256, 512
Chosen without explanation; determines model capacity and the resulting 56.18 MB size.

assumptions (4)

standard math Backpropagation and standard CNN training
Assumed as standard background for all reported training.
standard math SE block equations from Hu et al. [11]
Equations (2)-(5) are taken directly from the cited paper and are not re-derived.
domain assumption DFFD StyleGAN subset and FFHQ labels are correct
The paper assumes StyleGAN images are fake and FFHQ images are real, and that this subset represents deepfake detection.
ad hoc to paper Evaluation protocol is coherent
Section IV.A describes two incompatible protocols (80/20 split and 10-fold CV); the reported numbers require one or the other, but the paper does not specify which.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Enhancing Deepfake Detection using SE Block Attention with CNN." pith.science (2026). https://pith.science/paper/VYEYYL5V

@misc{pith2026250610683,
  author       = {Pith},
  title        = {Pith review of: Enhancing Deepfake Detection using SE Block Attention with CNN},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VYEYYL5V}},
  note         = {Machine review of arXiv:2506.10683}
}

read the original abstract

In the digital age, Deepfake present a formidable challenge by using advanced artificial intelligence to create highly convincing manipulated content, undermining information authenticity and security. These sophisticated fabrications surpass traditional detection methods in complexity and realism. To address this issue, we aim to harness cutting-edge deep learning methodologies to engineer an innovative deepfake detection model. However, most of the models designed for deepfake detection are large, causing heavy storage and memory consumption. In this research, we propose a lightweight convolution neural network (CNN) with squeeze and excitation block attention (SE) for Deepfake detection. The SE block module is designed to perform dynamic channel-wise feature recalibration. The SE block allows the network to emphasize informative features and suppress less useful ones, which leads to a more efficient and effective learning module. This module is integrated with a simple sequential model to perform Deepfake detection. The model is smaller in size and it achieves competing accuracy with the existing models for deepfake detection tasks. The model achieved an overall classification accuracy of 94.14% and AUC-ROC score of 0.985 on the Style GAN dataset from the Diverse Fake Face Dataset. Our proposed approach presents a promising avenue for combating the Deepfake challenge with minimal computational resources, developing efficient and scalable solutions for digital content verification.

Figures

Figures reproduced from arXiv: 2506.10683 by the authors.

**Figure 5.** The Confusion Matrix (Sequential) [PITH_FULL_IMAGE:figures/full_fig_p004_5.png] view at source ↗

**Figure 6.** The Confusion Matrix (Sequential + SE) Clip value of 1.0. Categorical cross entropy loss function is used with a batch size of 64. The model is trained for 100 epochs with 10-fold cross-validation. B. Results and Discussion The Deepfake detection task was conducted using the sequential SE block model with facial images taken from the Style GAN dataset. In this study, we analysed the performance of the model based o… view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: Samples of the images from Style GAN dataset [8], [18] [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]

**Figure 7.** Figure 7: AUC-ROC curve of the (Sequential + SE) model. AUC-ROC score [PITH_FULL_IMAGE:figures/full_fig_p005_7.png]

**Figure 8.** Figure 8: Correct Predictions of our model on test data [PITH_FULL_IMAGE:figures/full_fig_p005_8.png]

**Figure 9.** Figure 9: Incorrect Predictions of our model on test data [PITH_FULL_IMAGE:figures/full_fig_p005_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 13 canonical work pages

[1]

The creation and detection of deepfakes: A survey,

Y . Mirsky and W. Lee, “The creation and detection of deepfakes: A survey,”ACM Computing Surveys (CSUR), vol. 54, no. 1, pp. 1–41, 2021

work page 2021
[2]

A survey on deepfake video detection,

P. Yu, Z. Xia, J. Fei, and Y . Lu, “A survey on deepfake video detection,” Iet Biometrics, vol. 10, no. 6, pp. 607–624, 2021

work page 2021
[3]

A comprehensive overview of deepfake: Generation, detection, datasets, and opportuni- ties,

J. W. Seow, M. K. Lim, R. C. Phan, and J. K. Liu, “A comprehensive overview of deepfake: Generation, detection, datasets, and opportuni- ties,”Neurocomputing, vol. 513, pp. 351–371, 2022

2022
[4]

Digital and physical face attacks: Reviewing and one step further,

C. Kong, S. Wang, H. Liet al., “Digital and physical face attacks: Reviewing and one step further,”APSIPA Transactions on Signal and Information Processing, vol. 12, no. 1, 2022

work page 2022
[5]

Implicit identity driven deepfake face swapping detection,

B. Huang, Z. Wang, J. Yang, J. Ai, Q. Zou, Q. Wang, and D. Ye, “Implicit identity driven deepfake face swapping detection,” inPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 4490–4499

2023
[6]

Deepfake attacks: Generation, detection, datasets, challenges, and research direc- tions,

A. Naitali, M. Ridouani, F. Salahdine, and N. Kaabouch, “Deepfake attacks: Generation, detection, datasets, challenges, and research direc- tions,”Computers, vol. 12, no. 10, p. 216, 2023

work page 2023
[7]

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,” inProceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 1–11

2019
[8]

On the de- tection of digital face manipulation,

H. Dang, F. Liu, J. Stehouwer, X. Liu, and A. K. Jain, “On the de- tection of digital face manipulation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern recognition, 2020, pp. 5781–5790

work page 2020

Show all 21 references

[9]

An attention module for convolutional neural networks,

B. Zhu, P. Hofstee, J. Lee, and Z. Al-Ars, “An attention module for convolutional neural networks,” inArtificial Neural Networks and Machine Learning–ICANN 2021: 30th International Conference on Artificial Neural Networks, Bratislava, Slovakia, September 14–17, 2021, Proceedin...

2021
[10]

Deepfake video detection using convolu- tional vision transformer,

D. Wodajo and S. Atnafu, “Deepfake video detection using convolu- tional vision transformer,”arXiv preprint arXiv:2102.11126, 2021

2021 arXiv
[11]

Squeeze-and-excitation networks,

J. Hu, L. Shen, and G. Sun, “Squeeze-and-excitation networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 7132–7141

2018
[12]

Deepfake detection: A systematic literature review,

M. S. Rana, M. N. Nobi, B. Murali, and A. H. Sung, “Deepfake detection: A systematic literature review,”IEEE access, vol. 10, pp. 25 494–25 513, 2022

2022
[13]

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. Gellyet al., “An image is worth 16x16 words: Transformers for image recognition at scale,”arXiv preprint arXiv:2010.11929, 2020

2010 arXiv
[14]

Combining efficientnet and vision transformers for video deepfake detection,

D. A. Coccomini, N. Messina, C. Gennaro, and F. Falchi, “Combining efficientnet and vision transformers for video deepfake detection,” in International conference on image analysis and processing. Springer, 2022, pp. 219–229

2022
[15]

Cvt: Introducing convolutions to vision transformers,

H. Wu, B. Xiao, N. Codella, M. Liu, X. Dai, L. Yuan, and L. Zhang, “Cvt: Introducing convolutions to vision transformers,” inProceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 22–31

2021
[16]

Very deep convolutional networks for large-scale image recognition,

K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,”arXiv preprint arXiv:1409.1556, 2014

2014 arXiv
[17]

3d cnn architectures and attention mechanisms for deepfake detection,

R. Roy, I. Joshi, A. Das, and A. Dantcheva, “3d cnn architectures and attention mechanisms for deepfake detection,” inHandbook of Digital Face Manipulation and Detection: From DeepFakes to Morphing Attacks. Springer International Publishing Cham, 2022, pp. 213–234

2022
[18]

A style-based generator architecture for generative adversarial networks,

T. Karras, S. Laine, and T. Aila, “A style-based generator architecture for generative adversarial networks,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 4401–4410

2019
[19]

Fake- buster: A lightweight solution for deepfake detection,

N. Hubens, M. Mancas, B. Gosselin, M. Preda, and T. Zaharia, “Fake- buster: A lightweight solution for deepfake detection,” inApplications of Digital Image Processing XLIV, vol. 11842. SPIE, 2021, pp. 146– 154

2021
[20]

Deepfake detection an- alyzing hybrid dataset utilizing cnn and svm,

L. Pryor, R. Dave, M. Vanamalaet al., “Deepfake detection an- alyzing hybrid dataset utilizing cnn and svm,”arXiv preprint arXiv:2302.10280, 2023

2023 arXiv
[21]

kaggle-dfdc: Deepfake Detection Challenge (DFDC) solu- tion,

H. Cui, “kaggle-dfdc: Deepfake Detection Challenge (DFDC) solu- tion,” https://github.com/cuihaoleo/kaggle-dfdc, 2022, accessed: April 2024

2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

The creation and detection of deepfakes: A survey,

Y . Mirsky and W. Lee, “The creation and detection of deepfakes: A survey,”ACM Computing Surveys (CSUR), vol. 54, no. 1, pp. 1–41, 2021

work page 2021

[2] [2]

A survey on deepfake video detection,

P. Yu, Z. Xia, J. Fei, and Y . Lu, “A survey on deepfake video detection,” Iet Biometrics, vol. 10, no. 6, pp. 607–624, 2021

work page 2021

[3] [3]

A comprehensive overview of deepfake: Generation, detection, datasets, and opportuni- ties,

J. W. Seow, M. K. Lim, R. C. Phan, and J. K. Liu, “A comprehensive overview of deepfake: Generation, detection, datasets, and opportuni- ties,”Neurocomputing, vol. 513, pp. 351–371, 2022

2022

[4] [4]

Digital and physical face attacks: Reviewing and one step further,

C. Kong, S. Wang, H. Liet al., “Digital and physical face attacks: Reviewing and one step further,”APSIPA Transactions on Signal and Information Processing, vol. 12, no. 1, 2022

work page 2022

[5] [5]

Implicit identity driven deepfake face swapping detection,

B. Huang, Z. Wang, J. Yang, J. Ai, Q. Zou, Q. Wang, and D. Ye, “Implicit identity driven deepfake face swapping detection,” inPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 4490–4499

2023

[6] [6]

Deepfake attacks: Generation, detection, datasets, challenges, and research direc- tions,

A. Naitali, M. Ridouani, F. Salahdine, and N. Kaabouch, “Deepfake attacks: Generation, detection, datasets, challenges, and research direc- tions,”Computers, vol. 12, no. 10, p. 216, 2023

work page 2023

[7] [7]

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,” inProceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 1–11

2019

[8] [8]

On the de- tection of digital face manipulation,

H. Dang, F. Liu, J. Stehouwer, X. Liu, and A. K. Jain, “On the de- tection of digital face manipulation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern recognition, 2020, pp. 5781–5790

work page 2020