Pith. sign in

REVIEW 4 major objections 6 minor 80 references

Forensics Adapter: Unleashing CLIP for Generalizable Face Forgery Detection

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a 5.7M-parameter adapter, placed parallel to a frozen CLIP encoder and trained to predict blending boundaries, turns CLIP into the best cross-dataset face forgery detector among the compared methods.

desk verdict The genuinely new part is the ForAda++ text-prompt extension; the base adapter is the authors' CVPR 2025 paper, and the strong cross-dataset numbers come with a few evaluation caveats that a referee should ask them to fix. read the letter →

arxiv 2411.19715 v4 pith:2JP5MHXN submitted 2024-11-29 cs.CV cs.CRcs.LG

Xinjie Cui , Yuezun Li , Delong Zhu , Jiaran Zhou , Junyu Dong , Siwei Lyu This is my paper · ORCID
classification cs.CVcs.CRcs.LG
keywords faceforgerydetectiondeepfakeCLIPadaptationblendingboundarycontrastivelearningpromptcross-datasetgeneralizationforensicsadapter
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 CLIP can be turned into a generalizable face forgery detector by adding a small task-specific adapter, rather than by using CLIP only as a fixed feature extractor. The adapter runs alongside the frozen CLIP image encoder, is supervised to predict blending boundaries unique to forged faces, and pushes that trace knowledge back into CLIP through an attention bias. After training only on FaceForensics++ c23, the authors report frame-level AUC margins over previous state-of-the-art methods of 4.7%, 3.1%, 8.5%, 3.4%, 1.8%, and 0.2% on CDF-v1, CDF-v2, DFDC, DFDCP, DFD, and WDF, with just 5.7M trainable parameters. A text-augmented extension, Forensics Adapter++, adds a further 1.3% average gain through forgery-aware prompts. If the central claim is right, this provides a cheap, strong CLIP-based baseline that survives shifts between different manipulation pipelines.

What carries the argument

The central machinery is a parallel lightweight adapter built from the first eight layers of a tiny vision transformer, with only 5.7M trainable parameters, supervised by three task-specific objectives: masked blending-boundary regression, patch-wise contrastive learning, and sample-wise contrastive learning. The blending-boundary ground truth is generated from manipulation masks as Mbb' = 4M'(1-M'), the adapter and CLIP exchange knowledge through feature fusion in shallow layers and through an attention bias delta injected into CLIP's self-attention, and the refined [CLS*] token is used for authenticity prediction. This interaction lets the adapter guide the frozen CLIP toward forgery traces without destroying CLIP's general representation.

What would settle it

Train the same adapter on FaceForensics++ c23 but remove the blending-boundary supervision, or replace the boundary maps with randomly permuted versions, then measure frame-level AUC on CDF-v2 and DFDC; if performance stays near the reported 0.900 and 0.843, boundary supervision is not the load-bearing component. Alternatively, identify an unseen forgery generator whose outputs contain no measurable boundary artifact and show that detection performance collapses on it.

Watch

Extended reading notes

Core claim

The paper's central claim is that a lightweight adapter, designed around the characteristic blending boundary of face forgery, can unlock CLIP's pre-trained knowledge for generalizable face forgery detection. The adapter is a small vision transformer placed parallel to the frozen CLIP encoder, supervised by a masked blending-boundary loss, patch-wise contrastive learning on real versus fake local patches, and sample-wise contrastive learning on global CLIP tokens. Task-specific knowledge flows into CLIP by injecting an adapter-generated attention bias into CLIP's self-attention, using duplicated [CLS*] tokens, while CLIP's own parameters stay frozen. The extended Forensics Adapter++ adds a textual branch with learnable prompts that describe which facial regions are forged, uses text only during training, and shifts the trace supervision from blending boundaries to forged-region masks. The reported result is that this recipe outperforms existing detectors on six unseen datasets and that the text extension yields an additional 1.3% average improvement.

Load-bearing premise

The method assumes that blending boundaries, estimated from ground-truth manipulation masks, are the dominant generalizable trace of face forgery and that similar boundary artifacts appear in unseen forgeries produced by different pipelines.

Editorial extensions

If this is right

  • A detector trained only on FaceForensics++ c23 can outperform specialized detectors on six unseen datasets at frame level, so domain shift between forgery pipelines is not an unavoidable barrier for CLIP-based models.
  • A frozen CLIP encoder plus 5.7M trainable adapter parameters is sufficient for strong cross-dataset performance, which means large-scale fine-tuning of the vision-language backbone is not necessary for this task.
  • The 1.3% average gain from including text shows that CLIP's textual modality can contribute to forensics when prompts encode localized forgery regions rather than simple real/fake labels.
  • The method is more robust than comparison methods to six types of perturbations, which supports deployment under compression, blur, and noise.
  • Though designed for face forgery, the adapter also transfers partially to whole-face GAN and diffusion synthesis, beating the compared detectors on several such generators.

Reading between the lines

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

  • The ablation singles out patch-wise contrastive learning as the largest contributor, which suggests that separating real and fake local tokens, not just regressing boundaries, may be the active ingredient; a direct test would be to remove the boundary loss while keeping both contrastive losses and measuring cross-dataset AUC.
  • The attention-bias interaction is a generic mechanism for injecting task-specific knowledge into frozen CLIP, so the same architectural recipe could be tested on other fine-grained forensic tasks such as document forgery or synthetic-image detection.
  • The strong performance on whole-face GAN and diffusion generators, despite training only on face forgery, hints that blending-boundary supervision captures a broader artifact class; evaluating on a held-out synthesis model family absent from the paper would test this.
  • Because the paper varies the adapter architecture and finds that larger adapters do not help under the same training budget, the result implies that the 5.7M-parameter scale is near a sweet spot for this interaction design rather than a limit forced by compute.
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

4 major / 6 minor

Summary. This manuscript proposes Forensics Adapter, a lightweight (5.7M trainable parameters) adapter network placed in parallel with a frozen CLIP image encoder, and an extension Forensics Adapter++ that adds a text-prompt learning branch. The adapter is trained on FF++ (c23) with a masked blending-boundary prediction loss (L1), patch-wise contrastive learning (L2), sample-wise contrastive learning (L3), and a cross-entropy loss (L0); Forensics Adapter++ additionally uses a forgery-aware learnable prompt and an updated region-mask loss (L4). The claimed contribution is that the adapter transforms CLIP into a state-of-the-art generalizable face forgery detector, with large frame-level AUC gains over prior methods on CDF-v1, CDF-v2, DFDC, DFDCP, DFD, and WDF, plus a 1.3% average improvement from the textual modality. The paper includes extensive ablations, robustness tests, complexity analysis, and an evaluation on whole-face GAN/diffusion synthesis.

Significance. If the reported results are robust, the paper would provide a strong and efficient CLIP-based baseline for generalizable face forgery detection, and the text-prompt extension is a useful contribution beyond the prior CVPR version. The paper is generally well structured, includes many ablations and analyses, and reports the method on six external datasets with several metrics. However, the significance is moderated by three concerns: the headline 7% improvement is not supported by the per-dataset margins in Table I; several key hyperparameters are selected using CDF-v2 and DFDC, which are also the test datasets, so the cross-dataset numbers are partially tuned on the targets; and the claimed mechanism of learning 'blending boundaries' is not verified as the causal source of generalization, since no control target with matched spatial statistics is tested. These issues, if addressed, would substantially strengthen the credibility of the central claim.

major comments (4)
  1. [Abstract and Section I] The abstract and Section I claim 'approximately 7%' average AUC improvement over state-of-the-art, but the margins over the best prior method per dataset in Table I are 4.7%, 3.1%, 8.5%, 3.4%, 1.8%, and 0.2%, which average to about 3.6% (or about 4.3% if WDF is excluded). The 7% figure appears to be computed only against LSDA on five datasets, not against the best previous state-of-the-art in each column. Since the paper's central claim is outperforming state-of-the-art, the correct baseline is the best prior result per dataset. The abstract and introduction should be revised to report the actual average margin, and the conclusion (Section VI) should be corrected accordingly.
  2. [Section V-A, Tables VII-XI] Several hyperparameters are selected using the evaluation datasets themselves. Table VII uses CDF-v2 and DFDC to choose the area threshold for fake patches, Table IX uses them to choose the region threshold ν, Table X uses them to choose the prompt suffix length K, and Table XI uses them to choose the loss weights λ0-λ4. CDF-v2 and DFDC are the same datasets on which the headline cross-dataset numbers are reported. This is test-set hyperparameter selection, which makes the reported gains on those datasets optimistic and undermines the claim that the method was only trained on FF++ and evaluated on unseen datasets. The authors should either use a held-out validation set (e.g., a split from FF++ or a separate dataset) for all hyperparameter choices, or explicitly disclose that the reported numbers include hyperparameters selected on the test sets.
  3. [Section III-A, Eq. (2); Table IV; Table XVI] The load-bearing assumption that blending-boundary prediction is the source of generalization is not tested. Table IV shows that removing L1 decreases AUC by roughly 2-4 points on several datasets, but this only establishes that some dense supervision helps; it does not show that the specific boundary target is what transfers. A control experiment with an alternative dense prediction target matched in spatial statistics (e.g., random masks, a generic segmentation mask, or a constant map) is needed to rule out the possibility that L1 acts mainly as a regularizer. Moreover, Table XVI shows that the same adapter detects whole-face GAN and diffusion synthesis (StarGAN, CramerGAN, MMDGAN) with high AUC; those images contain no blending boundary, so the model cannot be relying exclusively on this cue. The paper should either provide the control experiment or soften the mechanistic claim that the adapter is 'grounded in the core nature of face forgery' and that blending boundaries are the unique generalizable trace.
  4. [Section V-B, Table II] The text states that 'Our ForAda significantly outperforms others on all datasets' and then lists improvements on CDF-v2, DFDC, DFDCP, and DFD. In Table II, however, ForAda achieves 0.805 on WDF while CFM achieves 0.823, so ForAda is not the best on WDF. The sentence 'all datasets' is therefore inaccurate. Additionally, in Table I, Forensics Adapter++ achieves 0.922 on DFD, which is lower than Forensics Adapter's 0.933; the text reports ForAda++ gains only on the five datasets where it improves and omits DFD. These discrepancies should be acknowledged and the claims should be reworded to match the full tables.
minor comments (6)
  1. [Section V-B] The method name is misspelled as 'Forenscis Adapter' and 'Forenscis Adapter++' in the first paragraph of Section V-B; this should be corrected to 'Forensics Adapter'.
  2. [Section III-B, Eq. (5)] The description of the attention bias interaction is ambiguous: the text says 'for each layer' CLIP tokens are augmented and Eq. (5) includes a layer index ℓ, but the attention bias ∆ is described as generated 'by the end of the adapter' and has shape h×w×N. It is unclear whether the same ∆ is broadcast across all CLIP layers or whether per-layer biases are produced. Please clarify how ∆ is applied at each layer.
  3. [Section IV-B, Eq. (9)] The dimension alignment for the visual-textual contrastive loss is unclear. The text says 'Avg means average pooling to align the dimension of features from the textual and visual branch,' but average pooling does not change the feature dimension; a linear projection or a description of how the 1024-dimensional visual token and the text-encoder output are matched (and whether the text encoder output dimension is 768 for ViT-L/14) is needed.
  4. [Section V-A] The implementation details state 'The dimension N is set to 128,' but N is defined in Section III-A as the number of learnable query tokens, not a dimension. Please specify that the number of query tokens is 128.
  5. [Section IV-B, Eq. (8)] In Eq. (8), the notation ℓ2(I(p), I'(p)) is ambiguous; it should be clear whether this is the per-pixel Euclidean distance and how the threshold β is applied to it. Also, the indicator function is typically written with an explicit condition; please make the definition self-contained.
  6. [Section IV-A] The preliminary exploration with straightforward prompts is described as yielding no performance gains, but no quantitative result is shown for that baseline. Reporting the corresponding AUC would make the motivation for the forgery-aware prompt design more concrete.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: external-dataset AUC results do not reduce to training targets; self-citations are disclosed and not load-bearing.

full rationale

The derivation chain is not circular. The adapter is trained on FF++ c23 with the four losses of Eq. (7), whose targets are ground-truth manipulation masks and real/fake labels from FF++ only, and the claimed predictions are frame-level and video-level AUC on CDF-v1, CDF-v2, DFDC, DFDCP, DFD, and WDF, which are external datasets not used to define any loss or architecture. The blending-boundary supervision M'_bb = 4M'(1-M') in Section III-A is an assumption about a transferable forgery trace, and the ablations in Tables IV-VIII show that removing components hurts performance, but that is a mechanism claim, not a logical reduction: the final detector is not defined by the test-set results, so the prediction does not equal the input by construction. The only self-references are disclosed: baseline numbers for many comparators are taken from the authors' prior LSDA paper, and the method extends the authors' CVPR 2025 paper; these are comparison and prior-work citations, not load-bearing premises for the adapter's design. Hyperparameters such as the 10% fake-patch threshold, the 8-layer adapter, and the 8-token prompt suffix were selected using CDF-v2 and DFDC ablations, which mildly favors those two datasets, but this is a selection or overfitting concern rather than circularity because the model still must correctly classify held-out frames and generalize to the other four datasets. Table XVI's strong whole-face GAN results weaken the claimed boundary-only mechanism, but that is an incomplete-explanation concern, not a reduction of the prediction to its inputs. No step in the paper reduces to its own input, so circularity is low; the score of 1.0 reflects the disclosed self-citation in baseline sourcing and the mild test-set-informed hyperparameter choices.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

No free-floating physical entities are introduced; the learnable architectural parameters are listed as free parameters. The main burden is the domain assumption that boundary artifacts generalize across unseen forgery pipelines.

free parameters (7)
  • Loss weights lambda0..lambda4 = lambda0=10, lambda1=200, lambda2=20 (adapter)/10 (CLIP), lambda3=10, lambda4=1.5
    Chosen by authors; Table XI shows sensitivity to lambda0, lambda1, and lambda4, and values were selected using cross-dataset test performance.
  • Region manipulation threshold nu = 0.15
    Selected from 0.10/0.15/0.20/0.40 based on CDF-v2 and DFDC test AUC in Table IX, which are part of the reported evaluation benchmarks.
  • Learnable prompt suffix length K = 8
    Selected from 0/4/8/16 based on CDF-v2 and DFDC test AUC in Table X; 16 sharply degrades performance.
  • Fake-patch area threshold = 10%
    Selected from 10/50/90/100% in Table VII based on CDF-v2 and DFDC test AUC.
  • Token enhancement factor alpha = 0.05
    Hand-set in Section V-A; no ablation is reported for this value.
  • Pixel-difference threshold beta = 20
    Fixed in Equation 8 for generating forgery-aware region prompts and labels.
  • Contrastive temperature tau = not reported
    Appears in Equations 3 and 4 but no value is given in the implementation details, leaving a missing hyperparameter that affects both contrastive losses.
assumptions (5)
  • domain assumption Blending boundaries are the dominant generalizable trace of face forgery.
    Motivates the entire adapter design in Section III-A; if false, task-specific losses teach dataset-specific artifacts.
  • domain assumption Frozen CLIP visual tokens contain reusable forgery-related priors that can be steered by a parallel adapter.
    Section I and III assume CLIP's versatility contains forgery knowledge; Table XII's BLIP result suggests this is not true for all vision-language models.
  • domain assumption Face region manipulation labels computed by L2 color difference between paired real and fake images are accurate enough to supervise text prompts.
    Used in Forensics Adapter++ Section IV-B; errors in region labels would propagate to losses L1 and L4.
  • domain assumption Training on FF++ c23 and testing on other datasets is a valid protocol for measuring generalization.
    All experiments adopt this protocol; the protocol is partly violated by ablations that use CDF-v2 and DFDC test performance to select hyperparameters.
  • ad hoc to paper Adding an attention bias computed from adapter outputs does not harm CLIP's general representations.
    Equation 5 design choice, not proven; robustness experiments partially test the effect but provide no theoretical guarantee.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Forensics Adapter: Unleashing CLIP for Generalizable Face Forgery Detection." pith.science (2026). https://pith.science/paper/2JP5MHXN

@misc{pith2026241119715,
  author       = {Pith},
  title        = {Pith review of: Forensics Adapter: Unleashing CLIP for Generalizable Face Forgery Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2JP5MHXN}},
  note         = {Machine review of arXiv:2411.19715}
}
abstract

We describe Forensics Adapter, an adapter network designed to transform CLIP into an effective and generalizable face forgery detector. Although CLIP is highly versatile, adapting it for face forgery detection is non-trivial as forgery-related knowledge is entangled with a wide range of unrelated knowledge. Existing methods treat CLIP merely as a feature extractor, lacking task-specific adaptation, which limits their effectiveness. To address this, we introduce an adapter to learn face forgery traces -- the blending boundaries unique to forged faces, guided by task-specific objectives. Then we enhance the CLIP visual tokens with a dedicated interaction strategy that communicates knowledge across CLIP and the adapter. Since the adapter is alongside CLIP, its versatility is highly retained, naturally ensuring strong generalizability in face forgery detection. {With only $\textbf{5.7M}$ trainable parameters, our method achieves superior performance across six standard datasets.} Additionally, we describe Forensics Adapter++, an extended method that incorporates textual modality via a newly proposed forgery-aware prompt learning strategy. This extension leads to a further $\textbf{1.3\%}$ performance boost over the original Forensics Adapter. We believe the proposed methods can serve as a baseline for future CLIP-based face forgery detection methods. The code has been released at https://github.com/OUC-VAS/ForensicsAdapter.

Figures

Figures reproduced from arXiv: 2411.19715 by the authors.

Figure 1
Figure 1. (a) Exisiting CLIP-based methods. (b) The proposed Forensics [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Pipeline of the proposed Forensics Adapter in training. The top stream denotes CLIP and the bottom stream corresponds adapter. See text for details. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Pipeline of Forensics Adapter++. The upper part illustrates the generation of forgery-aware prompts and masks. The lower part shows the training [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Robustness Analysis. Our method is compared with CLIP [ [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: T-SNE Visualizations. The left figure compares the feature distributions of the ForAda method and CLIP, while the right figure presents the [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

80 extracted references · 73 canonical work pages

  1. [1]

    Generative adversarial networks,

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y . Bengio, “Generative adversarial networks,” Communications of the ACM , pp. 139–144, 2020. 1, 2

    work page 2020

  2. [2]

    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,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2019. 1

    work page 2019

  3. [3]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” Advances in neural information processing systems , 2020. 1, 2

    work page 2020

  4. [4]

    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 Recogni- tion, 2022. 1, 2

    work page 2022

  5. [5]

    Deep fakes: A looming challenge for privacy, democracy, and national security,

    B. Chesney and D. Citron, “Deep fakes: A looming challenge for privacy, democracy, and national security,” Calif. L. Rev., 2019. 1

    work page 2019

  6. [6]

    Deepfakes, phrenology, surveillance, and more! a taxonomy of ai privacy risks,

    H.-P. Lee, Y .-J. Yang, T. S. V on Davier, J. Forlizzi, and S. Das, “Deepfakes, phrenology, surveillance, and more! a taxonomy of ai privacy risks,” in Proceedings of the CHI Conference on Human Factors in Computing Systems , 2024. 1

    work page 2024

  7. [7]

    Deepfake: Creation, purpose, risks,

    A. Busacca and M. A. Monaca, “Deepfake: Creation, purpose, risks,” in Innovations and Economic and Social Changes due to Artificial Intelligence: The State of the Art . Springer, 2023, pp. 55–68. 1

    work page 2023

  8. [8]

    Deepfakes and image manipulation: criminalisation and control,

    T. Kirchengast, “Deepfakes and image manipulation: criminalisation and control,” Information & Communications Technology Law , 2020. 1

    work page 2020

Show all 80 references
  1. [9]

    Celeb-df: A large- scale challenging dataset for deepfake forensics,

    Y . Li, X. Yang, P. Sun, H. Qi, and S. Lyu, “Celeb-df: A large- scale challenging dataset for deepfake forensics,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ,

  2. [10]

    The deepfake detection challenge (dfdc) dataset,

    B. Dolhansky, J. Bitton, B. Pflaum, J. Lu, R. Howes, M. Wang, and C. C. Ferrer, “The deepfake detection challenge (dfdc) dataset,” arXiv preprint arXiv:2006.07397, 2020. 1, 2, 6

    2006 arXiv

  3. [11]

    The deepfake detection challenge (dfdc) preview dataset

    B. Dolhansky, R. Howes, B. Pflaum, N. Baram, and C. Ferrer, “The deepfake detection challenge (dfdc) preview dataset.” arXiv preprint arXiv:1910.08854, 2019. 1, 2, 6

    1910 arXiv

  4. [12]

    Deepfakedetection, https://ai.googleblog.com/2019/09/ contributing-data-to-deepfake-detection.html/ Accessed: 2021-11-

    2019

  5. [13]

    Fakecatcher: Detection of synthetic portrait videos using biological signals,

    U. A. Ciftci, I. Demir, and L. Yin, “Fakecatcher: Detection of synthetic portrait videos using biological signals,” IEEE Transactions on Pattern Analysis and Machine Intelligence , pp. 1–1, 2020. 1

    2020

  6. [14]

    Exposing deep fakes using inconsistent head poses,

    X. Yang, Y . Li, and S. Lyu, “Exposing deep fakes using inconsistent head poses,” in ICASSP IEEE International Conference on Acoustics, Speech and Signal Processing , 2019. 1, 2

    2019

  7. [15]

    Deeprhythm: Exposing deepfakes with attentional visual heartbeat rhythms,

    H. Qi, Q. Guo, F. Juefei-Xu, X. Xie, L. Ma, W. Feng, Y . Liu, and J. Zhao, “Deeprhythm: Exposing deepfakes with attentional visual heartbeat rhythms,” in Proceedings of the 28th ACM international conference on multimedia, 2020. 1, 2

    2020

  8. [16]

    Deepfake detection based on discrepancies between faces and their context,

    Y . Nirkin, L. Wolf, Y . Keller, and T. Hassner, “Deepfake detection based on discrepancies between faces and their context,” IEEE Transactions on Pattern Analysis and Machine Intelligence , pp. 6111–6121, 2022. 1

    2022

  9. [17]

    Exposing deepfake videos by detecting face warping artifacts,

    Y . Li and S. Lyu, “Exposing deepfake videos by detecting face warping artifacts,” arXiv preprint arXiv:1811.00656 , 2019. 1

    2019 arXiv

  10. [18]

    Exploiting visual artifacts to expose deepfakes and face manipulations,

    F. Matern, C. Riess, and M. Stamminger, “Exploiting visual artifacts to expose deepfakes and face manipulations,” in IEEE Winter Applications of Computer Vision Workshops , 2019. 1

    2019

  11. [19]

    Thinking in frequency: Face forgery detection by mining frequency-aware clues,

    Y . Qian, G. Yin, L. Sheng, Z. Chen, and J. Shao, “Thinking in frequency: Face forgery detection by mining frequency-aware clues,” in European conference on computer vision , 2020. 1, 2, 7

    2020

  12. [20]

    Generalizing face forgery de- tection with high-frequency features,

    Y . Luo, Y . Zhang, J. Yan, and W. Liu, “Generalizing face forgery de- tection with high-frequency features,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2021. 1, 2, 7

    2021

  13. [21]

    Spatial-phase shallow learning: Rethinking face forgery detection in frequency domain,

    H. Liu, X. Li, W. Zhou, Y . Chen, Y . He, H. Xue, W. Zhang, and N. Yu, “Spatial-phase shallow learning: Rethinking face forgery detection in frequency domain,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2021. 1, 2, 7

    2021

  14. [22]

    Seeable: Soft discrepancies and bounded contrastive learning for exposing deep- fakes,

    N. Larue, N.-S. Vu, V . Struc, P. Peer, and V . Christophides, “Seeable: Soft discrepancies and bounded contrastive learning for exposing deep- fakes,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023. 1, 3, 7, 8

    2023

  15. [23]

    Detecting deepfakes with self-blended images,

    K. Shiohara and T. Yamasaki, “Detecting deepfakes with self-blended images,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022. 1, 2, 3, 7, 8

    2022

  16. [24]

    Fully unsupervised deepfake video detection via enhanced contrastive learning,

    T. Qiao, S. Xie, Y . Chen, F. Retraint, and X. Luo, “Fully unsupervised deepfake video detection via enhanced contrastive learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence , pp. 4654– 4668, 2024. 1, 2

    2024

  17. [25]

    Preserving fairness generalization in deepfake detection,

    L. Lin, X. He, Y . Ju, X. Wang, F. Ding, and S. Hu, “Preserving fairness generalization in deepfake detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024. 1

    2024

  18. [26]

    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 , 2021. 2, 3, 5, 6, 7, 9, 10, 11

    2021

  19. [27]

    Standing on the shoulders of giants: Reprogramming visual-language model for general deepfake detection,

    K. Lin, Y . Lin, W. Li, T. Yao, and B. Li, “Standing on the shoulders of giants: Reprogramming visual-language model for general deepfake detection,” in Proceedings of the AAAI Conference on Artificial Intelli- gence, 2025. 2, 3, 7

    2025

  20. [28]

    Ffaa: Multimodal large language model based explainable open-world face forgery analysis assistant,

    Z. Huang, B. Xia, Z. Lin, Z. Mou, and W. Yang, “Ffaa: Multimodal large language model based explainable open-world face forgery analysis assistant,” arXiv preprint arXiv:2408.10072 , 2024. 2, 3, 7, 8

    2024 arXiv

  21. [29]

    Clipping the deception: Adapting vision-language models for universal deepfake detection,

    S. A. Khan and D.-T. Dang-Nguyen, “Clipping the deception: Adapting vision-language models for universal deepfake detection,” in Proceed- ings of the International Conference on Multimedia Retrieval , 2024. 2, 3, 7, 8

    2024

  22. [30]

    Laa-net: Localized artifact attention network for quality-agnostic and generalizable deepfake detection,

    D. Nguyen, N. Mejri, I. P. Singh, P. Kuleshova, M. Astrid, A. Kacem, E. Ghorbel, and D. Aouada, “Laa-net: Localized artifact attention network for quality-agnostic and generalizable deepfake detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern R...

    2024

  23. [31]

    Forensics adapter: Adapting clip for generalizable face forgery detection,

    X. Cui, Y . Li, A. Luo, J. Zhou, and J. Dong, “Forensics adapter: Adapting clip for generalizable face forgery detection,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,

  24. [32]

    Faceforensics++: Learning to detect manipulated facial images,

    A. Rossler, D. Cozzolino, L. Verdoliva, C. Riess, J. Thies, and M. Niess- ner, “Faceforensics++: Learning to detect manipulated facial images,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019. 2, 6, 7

    2019

  25. [33]

    Learning disen- tangled representation for one-shot progressive face swapping,

    Q. Li, W. Wang, C. Xu, Z. Sun, and M.-H. Yang, “Learning disen- tangled representation for one-shot progressive face swapping,” IEEE Transactions on Pattern Analysis and Machine Intelligence , pp. 8348– 8364, 2024. 2

    2024

  26. [34]

    In ictu oculi: Exposing ai created fake videos by detecting eye blinking,

    Y . Li, M.-C. Chang, and S. Lyu, “In ictu oculi: Exposing ai created fake videos by detecting eye blinking,” in IEEE International workshop on information forensics and security , 2018. 2

    2018

  27. [35]

    Aunet: Learning relations between action units for face forgery detection,

    W. Bai, Y . Liu, Z. Zhang, B. Li, and W. Hu, “Aunet: Learning relations between action units for face forgery detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ,

  28. [36]

    Detecting deep-fake videos from appearance and behavior,

    S. Agarwal, H. Farid, T. El-Gaaly, and S.-N. Lim, “Detecting deep-fake videos from appearance and behavior,” in IEEE international workshop on information forensics and security , 2020. 2

    2020

  29. [37]

    Dual contrastive learning for general face forgery detection,

    K. Sun, T. Yao, S. Chen, S. Ding, J. Li, and R. Ji, “Dual contrastive learning for general face forgery detection,” in Proceedings of the AAAI Conference on Artificial Intelligence , 2022. 2, 11 JOURNAL OF LATEX CLASS FILES, VOL. 18, NO. 9, SEPTEMBER 2020 13

    2022

  30. [38]

    End-to- end reconstruction-classification learning for face forgery detection,

    J. Cao, C. Ma, T. Yao, S. Chen, S. Ding, and X. Yang, “End-to- end reconstruction-classification learning for face forgery detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022. 2, 7

    2022

  31. [39]

    Face x-ray for more general face forgery detection,

    L. Li, J. Bao, T. Zhang, H. Yang, D. Chen, F. Wen, and B. Guo, “Face x-ray for more general face forgery detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ,

  32. [40]

    Gm-df: Generalized multi-scenario deepfake detection,

    Y . Lai, Z. Yu, J. Yang, B. Li, X. Kang, and L. Shen, “Gm-df: Generalized multi-scenario deepfake detection,” arXiv preprint arXiv:2406.20078 ,

    arXiv

  33. [41]

    Exploring unbiased deepfake detection via token-level shuffling and mixing,

    X. Fu, Z. Yan, T. Yao, S. Chen, and X. Li, “Exploring unbiased deepfake detection via token-level shuffling and mixing,” in Proceedings of the AAAI Conference on Artificial Intelligence , 2025. 3, 7

    2025

  34. [42]

    Towards more general video-based deepfake detection through facial feature guided adaptation for foundation model,

    Y .-H. Han, T.-M. Huang, S.-T. Lo, P.-H. Huang, K.-L. Hua, and J.-C. Chen, “Towards more general video-based deepfake detection through facial feature guided adaptation for foundation model,” in Proceedings of the Conference on Computer Vision and Pattern Recognition , 2025. 3, 7

    2025

  35. [43]

    Towards general visual-linguistic face forgery detection,

    K. Sun, S. Chen, T. Yao, H. Yang, X. Sun, S. Ding, and R. Ji, “Towards general visual-linguistic face forgery detection,” in Proceedings of the Conference on Computer Vision and Pattern Recognition , 2025. 3, 7, 8

    2025

  36. [44]

    Tinyvit: Fast pretraining distillation for small vision transformers,

    K. Wu, J. Zhang, H. Peng, M. Liu, B. Xiao, J. Fu, and L. Yuan, “Tinyvit: Fast pretraining distillation for small vision transformers,” in European conference on computer vision , 2022. 3

    2022

  37. [45]

    Forgery- aware adaptive transformer for generalizable synthetic image detection,

    H. Liu, Z. Tan, C. Tan, Y . Wei, J. Wang, and Y . Zhao, “Forgery- aware adaptive transformer for generalizable synthetic image detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024. 3

    2024

  38. [46]

    Side adapter network for open-vocabulary semantic segmentation,

    M. Xu, Z. Zhang, F. Wei, H. Hu, and X. Bai, “Side adapter network for open-vocabulary semantic segmentation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ,

  39. [47]

    End-to-end object detection with transformers,

    N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov, and S. Zagoruyko, “End-to-end object detection with transformers,” in European conference on computer vision , 2020. 3

    2020

  40. [48]

    Per-pixel classification is not all you need for semantic segmentation,

    B. Cheng, A. Schwing, and A. Kirillov, “Per-pixel classification is not all you need for semantic segmentation,” Advances in neural information processing systems, 2021. 3

    2021

  41. [49]

    Masked-attention mask transformer for universal image segmentation,

    B. Cheng, I. Misra, A. G. Schwing, A. Kirillov, and R. Girdhar, “Masked-attention mask transformer for universal image segmentation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022. 3

    2022

  42. [50]

    Learning to prompt for vision-language models,

    K. Zhou, J. Yang, C. C. Loy, and Z. Liu, “Learning to prompt for vision-language models,” International Journal of Computer Vision , pp. 2337–2348, 2022. 5, 6

    2022

  43. [51]

    Deepfakebench: A comprehensive benchmark of deepfake detection,

    Z. Yan, Y . Zhang, X. Yuan, S. Lyu, and B. Wu, “Deepfakebench: A comprehensive benchmark of deepfake detection,” Advances in Neural Information Processing Systems , 2023. 6

    2023

  44. [52]

    Beyond the prior forgery knowledge: Mining critical clues for general face forgery detection,

    A. Luo, C. Kong, J. Huang, Y . Hu, X. Kang, and A. C. Kot, “Beyond the prior forgery knowledge: Mining critical clues for general face forgery detection,” IEEE Transactions on Information Forensics and Security , pp. 1168–1182, 2024. 6, 7, 11

    2024

  45. [53]

    Diffusionfake: Enhancing generalization in deepfake detection via guided stable dif- fusion,

    S. Chen, T. Yao, H. Liu, X. Sun, S. Ding, R. Ji et al., “Diffusionfake: Enhancing generalization in deepfake detection via guided stable dif- fusion,” Advances in Neural Information Processing Systems , 2024. 6, 7

    2024

  46. [54]

    Wilddeepfake: A challenging real-world dataset for deepfake detection,

    B. Zi, M. Chang, J. Chen, X. Ma, and Y .-G. Jiang, “Wilddeepfake: A challenging real-world dataset for deepfake detection,” in Proceedings of the 28th ACM international conference on multimedia , 2020. 6

    2020

  47. [55]

    Efficientnet: Rethinking model scaling for con- volutional neural networks,

    M. Tan and Q. Le, “Efficientnet: Rethinking model scaling for con- volutional neural networks,” in International conference on machine learning, 2019. 7

    2019

  48. [56]

    On the detection of digital face manipulation,

    H. Dang, F. Liu, J. Stehouwer, X. Liu, and A. K. Jain, “On the detection of digital face manipulation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2020. 7

    2020

  49. [57]

    Ucf: Uncovering common features for generalizable deepfake detection,

    Z. Yan, Y . Zhang, Y . Fan, and B. Wu, “Ucf: Uncovering common features for generalizable deepfake detection,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023. 7

    2023

  50. [58]

    Exposing the deception: Uncovering more forgery clues for deepfake detection,

    Z. Ba, Q. Liu, Z. Liu, S. Wu, F. Lin, L. Lu, and K. Ren, “Exposing the deception: Uncovering more forgery clues for deepfake detection,” in Proceedings of the AAAI Conference on Artificial Intelligence , 2024. 7

    2024

  51. [59]

    Learning to discover forgery cues for face forgery detection,

    J. Tian, P. Chen, C. Yu, X. Fu, X. Wang, J. Dai, and J. Han, “Learning to discover forgery cues for face forgery detection,” IEEE Transactions on Information Forensics and Security , pp. 3814–3828, 2024. 7

    2024

  52. [60]

    Transcending forgery specificity with latent space augmentation for generalizable deepfake detection,

    Z. Yan, Y . Luo, S. Lyu, Q. Liu, and B. Wu, “Transcending forgery specificity with latent space augmentation for generalizable deepfake detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024. 7, 11

    2024

  53. [61]

    Tall: Thumbnail layout for deepfake video detection,

    Y . Xu, J. Liang, G. Jia, Z. Yang, Y . Zhang, and R. He, “Tall: Thumbnail layout for deepfake video detection,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023. 7

    2023

  54. [62]

    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,” in Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023. 7, 11

    2023

  55. [63]

    Learning spatiotemporal inconsistency via thumbnail layout for face deepfake detection,

    Y . Xu, J. Liang, L. Sheng, and X.-Y . Zhang, “Learning spatiotemporal inconsistency via thumbnail layout for face deepfake detection,” Inter- national Journal of Computer Vision , pp. 5663–5680, 2024. 7

    2024

  56. [64]

    Exploiting style latent flows for generalizing deepfake video detection,

    J. Choi, T. Kim, Y . Jeong, S. Baek, and J. Choi, “Exploiting style latent flows for generalizing deepfake video detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ,

  57. [65]

    Implicit identity leakage: The stumbling block to improving deepfake detection gener- alization,

    S. Dong, J. Wang, R. Ji, J. Liang, H. Fan, and Z. Ge, “Implicit identity leakage: The stumbling block to improving deepfake detection gener- alization,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023. 7

    2023

  58. [66]

    Dynamic graph learning with content-guided spatial-frequency relation reasoning for deepfake detection,

    Y . Wang, K. Yu, C. Chen, X. Hu, and S. Peng, “Dynamic graph learning with content-guided spatial-frequency relation reasoning for deepfake detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023. 7

    2023

  59. [67]

    Freqblender: Enhanc- ing deepfake detection by blending frequency knowledge,

    H. Li, J. Zhou, Y . Li, B. Wu, B. Li, and J. Dong, “Freqblender: Enhanc- ing deepfake detection by blending frequency knowledge,” Advances in Neural Information Processing Systems , 2024. 7

    2024

  60. [68]

    Pytorch: An imperative style, high-performance deep learning library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga et al. , “Pytorch: An imperative style, high-performance deep learning library,” Advances in neural information processing systems , 2019. 6

    2019

  61. [69]

    Adam: A method for stochastic optimization,

    D. P. Kingma, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014. 6

    2014 arXiv

  62. [70]

    Retinaface: Single-shot multi-level face localisation in the wild,

    J. Deng, J. Guo, E. Ververas, I. Kotsia, and S. Zafeiriou, “Retinaface: Single-shot multi-level face localisation in the wild,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,

  63. [71]

    Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation,

    J. Li, D. Li, C. Xiong, and S. Hoi, “Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation,” in International conference on machine learning , 2022. 9, 10

    2022

  64. [72]

    Stargan: Unified generative adversarial networks for multi-domain image-to- image translation,

    Y . Choi, M. Choi, M. Kim, J.-W. Ha, S. Kim, and J. Choo, “Stargan: Unified generative adversarial networks for multi-domain image-to- image translation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2018. 11

    2018

  65. [73]

    The cramer distance as a solution to biased wasserstein gradients,

    M. G. Bellemare, I. Danihelka, W. Dabney, S. Mohamed, B. Lakshmi- narayanan, S. Hoyer, and R. Munos, “The cramer distance as a solution to biased wasserstein gradients,” arXiv preprint arXiv:1705.10743, 2017. 11

    2017 arXiv

  66. [74]

    Mmd gan: Towards deeper understanding of moment matching network,

    C.-L. Li, W.-C. Chang, Y . Cheng, Y . Yang, and B. P ´oczos, “Mmd gan: Towards deeper understanding of moment matching network,” Advances in neural information processing systems , 2017. 11

    2017

  67. [75]

    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 , 2021. 11

    2021

  68. [76]

    Pseudo numerical methods for diffusion models on manifolds,

    L. Liu, Y . Ren, Z. Lin, and Z. Zhao, “Pseudo numerical methods for diffusion models on manifolds,” in International Conference on Learning Representations, 2022. 11

    2022

  69. [77]

    Deeperforensics- 1.0: A large-scale dataset for real-world face forgery detection,

    L. Jiang, R. Li, W. Wu, C. Qian, and C. C. Loy, “Deeperforensics- 1.0: A large-scale dataset for real-world face forgery detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020. 10

    2020

  70. [78]

    Lips don’t lie: A generalisable and robust approach to face forgery detection,

    A. Haliassos, K. V ougioukas, S. Petridis, and M. Pantic, “Lips don’t lie: A generalisable and robust approach to face forgery detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021. 11

    2021

  71. [79]

    Exploring temporal coherence for more general video face forgery detection,

    Y . Zheng, J. Bao, D. Chen, M. Zeng, and F. Wen, “Exploring temporal coherence for more general video face forgery detection,” inProceedings of the IEEE/CVF International Conference on Computer Vision , 2021. 11

    2021

  72. [80]

    Visualizing data using t-sne

    L. Van der Maaten and G. Hinton, “Visualizing data using t-sne.” Journal of machine learning research , 2008. 11

    2008

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.