Pith. sign in

REVIEW 3 major objections 4 minor 53 references

SELFI: Selective Fusion of Identity for Generalizable Deepfake Detection

T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read The paper claims that adaptively gating face identity features per image—rather than always suppressing or always using them—improves deepfake detection generalization across datasets and manipulation types.

desk verdict SELFI's adaptive gating claim is not isolated by the experiments; the identity-feature analysis is solid, but the paper needs a static-fusion baseline and error bars before the headline mechanism can be believed. read the letter →

arxiv 2506.17592 v1 pith:2QO7JJTQ submitted 2025-06-21 cs.CV

Younghun Kim , Minsuk Jang , Myung-Joon Kwon , Wonjun Lee , Changick Kim This is my paper · ORCID
classification cs.CV
keywords deepfakedetectionfaceidentityadaptivefusiongeneralizationcross-datasetevaluationforgery-awareadapteridentity-awaremodule
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 sets out to settle a dispute in deepfake detection: should a detector suppress face identity cues as a source of bias, or lean on them as forensic evidence? Its answer, supported by experiments, is neither: identity is genuinely informative, but its value depends on the manipulation method. SELFI therefore extracts identity embeddings from a frozen face recognizer, projects them into a forgery-relevant space, and uses a per-image relevance score to blend them with visual features. On four cross-dataset benchmarks the framework reports an average gain of 3.1% frame-level AUC over prior methods, with a 6% margin on the DFDC benchmark. A sympathetic reading is that the paper's core contribution is evidence that the identity-versus-visual trade-off should be learned per sample, not decided in advance.

What carries the argument

The load-bearing mechanism is the soft fusion rule in Eq. (5): $f_{\text{fused}} = \rho \cdot f_{\text{fi}} + (1-\rho) \cdot f_{\text{vis}}$, where $\rho$ is a scalar in $[0,1]$ produced by a two-layer relevance predictor that takes the concatenated identity and visual features. The identity feature comes from FAIA: a frozen face-recognition embedding (IResNet100, 512 dimensions) projected by a learnable matrix into the visual backbone's feature space, with a forgery-aware cross-entropy loss applied to that projection so the identity space is aligned with the fake-detection task. The gate $\rho$ is what lets the model amplify identity when it helps and suppress it when it would hurt.

What would settle it

Retrain SELFI with a face recognizer that provably never saw the benchmark identities—for example, trained only on synthetic faces or on a disjoint identity set—and compare cross-dataset AUC. If the reported 3.1% average gain and 6% DFDC gain shrink sharply or disappear, the gains depend on identity memorization rather than on a generalizable forgery cue.

Watch

Extended reading notes

Core claim

The paper's central claim is that selective identity use beats both extreme policies: discarding identity features entirely or trusting them unconditionally. The empirical discovery underneath it is that identity embeddings alone can separate real from fake faces, but they generalize unevenly across manipulation types—transferable for DeepFakes and FaceSwap, method-specific for Face2Face, and largely ineffective for NeuralTextures. SELFI operationalizes selectivity by predicting a relevance scalar $\rho$ for each input and computing the fused feature as $\rho$ times the projected identity feature plus $(1-\rho)$ times the visual feature, trained jointly with a forgery-aware guidance loss on the identity branch. If the claim holds, identity should be treated as a conditional forensic cue whose weight is learned from the input, not fixed by design choice.

Load-bearing premise

The central claim presumes that the frozen face recognizer's identity embeddings are a transferable and leakage-free representation of face identity for the test benchmarks, with no overlap between the recognizer's training identities and the benchmark faces.

Editorial extensions

If this is right

  • Cross-dataset generalization improves by an average of 3.1% frame-level AUC across Celeb-DF v2, DFD, DFDC, and DFDCP, and by 6% on DFDC.
  • Identity-aware features outperform other auxiliary sources—random, ImageNet, and deepfake-trained features—by 13.9 to 23.8% AUC, indicating the gain comes from identity semantics, not feature ensembling.
  • The fusion module transfers across backbones, adding 5.6% average AUC to CLIP, 2.6% to ResNet34, and 1.6% to EfficientNet-B4.
  • In cross-manipulation tests, training on one manipulation type and testing on others, SELFI beats CLIP and the implicit-identity baseline by up to 4.0% average AUC.

Reading between the lines

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

  • A natural check on the claim is whether the frozen face recognizer's training data overlaps with the benchmark identities; if overlap exists, part of the reported cross-dataset gain could be identity memorization rather than a transferable forgery cue. The paper does not measure this overlap.
  • A testable extension is to replace the single scalar gate with a per-channel or per-region gating vector, which may preserve identity cues where they are reliable and discard them where they are not.
  • The same per-sample relevance mechanism could be applied to other strong semantic features such as pose, expression, or lighting, turning SELFI from an identity-specific fix into a general template for bias-aware fusion.
  • Since the paper itself notes that identity embeddings degrade for partially visible or extreme-angle faces, conditioning the relevance gate on face quality or landmark confidence is a concrete next step.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper studies the role of face identity in deepfake detection and proposes SELFI, a framework that extracts identity embeddings from a frozen face recognition model (IResNet100), projects them into a forgery-relevant space via a trainable adapter (FAIA), and adaptively fuses them with visual features through a per-sample relevance gate (IAFM). The authors first report that identity embeddings alone can classify real vs. fake faces, and that their transferability varies across manipulation types. They then evaluate SELFI trained on FaceForensics++ on four cross-dataset benchmarks (CDFv2, DFD, DFDC, DFDCP), reporting average frame-level AUC improvements over prior methods, and provide ablations over modules, backbones, and auxiliary feature sources. The central claim is that selective, relevance-guided fusion of identity is preferable to both suppressing identity and using it unconditionally.

Significance. If established, the paper would help reconcile conflicting views on identity in deepfake detection by showing that the utility of identity cues is manipulation-dependent and that a learned per-sample gate can exploit this. The work is empirically grounded: cross-dataset test sets are held out from training, the identity extractor is frozen, and the auxiliary-source ablation in Table 5 partially controls for ensemble effects. The controlled comparisons across backbones in Table 4 (e.g., +5.6% average AUC for CLIP) are a strength, and the proposed modules are simple and potentially reusable. However, the distinguishing element of the method—adaptive per-sample fusion—is not yet isolated from static fusion, and the possibility of identity leakage through the pretrained face recognizer is not addressed. The paper is therefore of interest to the community, but the evidence for the 'selective' mechanism requires additional experiments.

major comments (3)
  1. [§5.3, Tables 3 and 8; Eq. (5)] The claim that per-sample relevance gating (IAFM) is preferable to static identity usage is not supported by the presented evidence. On CLIP (Table 3), adding IAFM to FAIA-only changes average AUC from 0.845 to 0.849, and on DFDCP the score drops from 0.844 to 0.830; on EfficientNet-B4 (Table 8) the average drops from 0.759 to 0.745, while on ResNet34 it rises from 0.749 to 0.761. These differences are small and inconsistent, and no static-fusion baseline is reported—for example, a fixed scalar rho=0.5, a global learned scalar, or simple averaging of the two feature vectors. Because Eq. (5) is a convex combination that discards part of the identity stream when rho<1, any gain over concatenation could be due to reduced input dimensionality or implicit regularization rather than per-sample selectivity. The paper should compare IAFM against static fusion rules and report multiple seeds and error bars to establish that the adaptive mechanism itself is beneficial.
  2. [§1 and §3.1, Eq. (1)] The frozen IResNet100 model used for identity embeddings is pretrained on large-scale face recognition datasets (e.g., MS-Celeb-1M), which may contain identities that appear in the evaluation benchmarks CDFv2 and DFDC. If the recognizer has memorized those identities, the reported cross-dataset gains could be inflated by identity memorization rather than by a transferable forgery cue. The paper neither measures the overlap between the recognizer's training identities and the test identities nor ablates the effect, for example by evaluating on identities absent from the recognizer's training set. This is load-bearing for the headline cross-dataset results and should be addressed.
  3. [§5.3, Tables 3–5] The module and backbone ablations in Tables 3 and 4 omit the DFDC benchmark, while Table 5 and the headline results include it. DFDC is the benchmark on which the largest claimed gain (6%) is based, so its absence from the ablations makes it difficult to verify that the improvement on DFDC is attributable to the proposed modules. Please report the missing DFDC numbers for the ablation rows, or explicitly state why DFDC was excluded from these experiments.
minor comments (4)
  1. [§4.3 and Table 8] The notation for the auxiliary loss is inconsistent: Eq. (7) and the text use L_fag, while Table 8 and some ablation descriptions use L fs; the table header also writes 'EfficientNetb4' instead of 'EfficientNet-B4'. Please unify the notation throughout.
  2. [Table 5] There is a typo in the row label 'Random Intialization'—it should be 'Random Initialization'.
  3. [§5.2] The abstract states that SELFI outperforms prior state-of-the-art by an average of 3.1% AUC, while the text in §5.2 says it achieves an average improvement of 3.5% over LSDA. Please clarify which comparison is being reported and the exact sets of datasets, since these numbers differ.
  4. [§3.2 and Table 6] The text cites specific numbers such as '84.6% → 74.5%' in describing Fig. 2(b), but the corresponding Table 6 reports 0.846 and 0.745 in different row/column positions; the figure and table should be cross-referenced more clearly so the reader can map the patterns to the quantitative results.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: SELFI's adaptive-fusion claim is supported by held-out benchmark evaluations, and the only self-citation (FRIDAY) is background.

full rationale

SELFI does not derive a target quantity from itself. Its central evidence is empirical: the model is trained on FF++ and evaluated on CDFv2, DFD, DFDC, and DFDCP, which are not used for training or for optimizing the reported cross-dataset results; the AUC numbers are held-out measurements rather than predictions obtained from fitted constants. The identity embedding in Eq. (1) comes from a frozen, externally trained IResNet100, and the relevance gate rho in Eq. (4) is a learned function of the concatenated identity/visual features, not a quantity defined in terms of the test labels. The only self-citation is [20] (FRIDAY), mentioned in related work as a contrasting identity-suppression method; it is not used to justify SELFI's correctness or to forbid alternatives. The ablation result that IAFM adds only 0.4% average AUC on CLIP and reduces DFDCP performance weakens the adaptive-fusion claim, but that is an experimental-isolation concern, not a circularity: no equation or fitted value is equivalent to the claimed outcome by construction. The Limitations section likewise acknowledges only sensitivity to identity-embedding quality and other bias sources, not a circular derivation.

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

No new physical or conceptual entities are postulated. FAIA, IAFM, and the Relevance Predictor are compositions of standard neural network components (linear layers, sigmoid, weighted sum), not new postulated quantities with independent evidence requirements. The free parameters are standard hand-set hyperparameters; the load-bearing assumptions are about data leakage and evaluation protocol.

free parameters (3)
  • Loss weights alpha and beta = alpha=1.0, beta=1.0
    Global weights in Eq. (7). Set by hand to 1.0 in all experiments without tuning; the paper reports no search over them.
  • Training budget and resolution = 10 epochs, batch size 64, 224x224 input
    Chosen to follow DeepfakeBench defaults (Sec. 5.1); these choices affect all reported AUC numbers but are not justified.
  • Relevance predictor architectural size = not specified
    R(.) in Eq. (4) is described as two linear layers with ReLU in between, but hidden dimensions are omitted; the capacity of this gate influences the fusion behavior and is set by hand.
assumptions (4)
  • ad hoc to paper The frozen IResNet100 embeddings used in Eq. (1) represent face identity and do not leak test identities from pretraining.
    Load-bearing for both the Sec. 3 analysis and the SELFI gains; the paper never checks overlap between IResNet100's training identities and CDFv2/DFDC subjects. This is an unverified domain assumption specific to this paper's setup.
  • domain assumption FF++ c23 is a sufficient training distribution, and its validation AUC is a reliable model selection signal for cross-dataset performance.
    Sec. 5.1 trains on FF++ and selects the model by validation AUC; cross-dataset transfer is assumed to follow from in-distribution validation quality.
  • domain assumption The evaluation benchmarks (CDFv2, DFD, DFDC, DFDCP) are treated as disjoint from the training set in content and manipulation method.
    Required for the cross-dataset generalization claim; the paper follows DeepfakeBench splits but does not verify identity-level disjointness.
  • domain assumption Frame-level classification with 32 sampled frames approximates video-level detection.
    Used in Sec. A.2 to compare against video-level methods; no temporal modeling is employed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SELFI: Selective Fusion of Identity for Generalizable Deepfake Detection." pith.science (2026). https://pith.science/paper/2QO7JJTQ

@misc{pith2026250617592,
  author       = {Pith},
  title        = {Pith review of: SELFI: Selective Fusion of Identity for Generalizable Deepfake Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2QO7JJTQ}},
  note         = {Machine review of arXiv:2506.17592}
}
read the original abstract

Face identity provides a powerful signal for deepfake detection. Prior studies show that even when not explicitly modeled, classifiers often learn identity features implicitly. This has led to conflicting views: some suppress identity cues to reduce bias, while others rely on them as forensic evidence. To reconcile these views, we analyze two hypotheses: (1) whether face identity alone is discriminative for detecting deepfakes, and (2) whether such identity features generalize poorly across manipulation methods. Our experiments confirm that identity is informative but context-dependent. While some manipulations preserve identity-consistent artifacts, others distort identity cues and harm generalization. We argue that identity features should neither be blindly suppressed nor relied upon, but instead be explicitly modeled and adaptively controlled based on per-sample relevance. We propose \textbf{SELFI} (\textbf{SEL}ective \textbf{F}usion of \textbf{I}dentity), a generalizable detection framework that dynamically modulates identity usage. SELFI consists of: (1) a Forgery-Aware Identity Adapter (FAIA) that extracts identity embeddings from a frozen face recognition model and projects them into a forgery-relevant space via auxiliary supervision; and (2) an Identity-Aware Fusion Module (IAFM) that selectively integrates identity and visual features using a relevance-guided fusion mechanism. Experiments on four benchmarks show that SELFI improves cross-manipulation generalization, outperforming prior methods by an average of 3.1\% AUC. On the challenging DFDC dataset, SELFI exceeds the previous best by 6\%. Code will be released upon paper acceptance.

Figures

Figures reproduced from arXiv: 2506.17592 by the authors.

Figure 1
Figure 1. Overview of the proposed SELFI framework. Traditional deepfake detectors implicitly rely on identity cues without control, risking overfitting and poor generalization across manipulation types. In contrast, SELFI explicitly incorporates face identity features through a Forgery-Aware Identity Adapter (FAIA) and adaptively fuses them with visual features using an Identity-Aware Fusion Module (IAFM). This enables the m… view at source ↗
Figure 2
Figure 2. Preliminary Analysis of Identity-Only Detection. (a) Training loss and accuracy curves for identity-only detection using a frozen face recognition model (IResNet100) on FF++, DFD, and CelebDFv2. The results show consistent decreases in loss and increases in accuracy, indicating that face identity alone provides strong discriminative power for real-vs-fake classification. (b) Cross-manipulation AUC heatmap, where row… view at source ↗
Figure 3
Figure 3. Architecture of the proposed SELFI framework. SELFI consists of two main modules: (1) the Forgery-Aware Identity Adapter (FAIA), which projects frozen identity embeddings into a forgery-relevant space using a trainable transformation and auxiliary supervision (Sec. 4.1); and (2) the Identity-Aware Fusion Module (IAFM), which adaptively fuses identity and visual features based on a predicted relevance score (Sec. 4.2… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Architecture of the identity-only detection setting. A lightweight classifier is trained on top of frozen face identity embeddings extracted from a face recognition model, without any access to visual content. This setup is used to investigate whether face identity alo…

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 31 canonical work pages

  1. [1]

    Mesonet: a compact facial video forgery detection network

    Darius Afchar, Vincent Nozick, Junichi Yamagishi, and Isao Echizen. Mesonet: a compact facial video forgery detection network. In2018 IEEE international workshop on information forensics and security (WIFS), pages 1–7. IEEE, 2018

    work page 2018

  2. [2]

    Contributing data to deepfake detection research, 2019

    Google AI. Contributing data to deepfake detection research, 2019. Accessed: 2025-04-08

    work page 2019

  3. [3]

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

    Junyi Cao, Chao Ma, Taiping Yao, Shen Chen, Shouhong Ding, and Xiaokang Yang. End- to-end reconstruction-classification learning for face forgery detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4113–4122, 2022

    work page 2022

  4. [4]

    Stacking Brick by Brick: Aligned Feature Isolation for Incremental Face Forgery Detection

    Jikang Cheng, Zhiyuan Yan, Ying Zhang, Li Hao, Jiaxin Ai, Qin Zou, Chen Li, and Zhongyuan Wang. Stacking brick by brick: Aligned feature isolation for incremental face forgery detection. arXiv preprint arXiv:2411.11396, 2024

    work page Pith review arXiv 2024

  5. [5]

    Can we leave deepfake data behind in training deepfake detector?arXiv preprint arXiv:2408.17052, 2024

    Jikang Cheng, Zhiyuan Yan, Ying Zhang, Yuhao Luo, Zhongyuan Wang, and Chen Li. Can we leave deepfake data behind in training deepfake detector?arXiv preprint arXiv:2408.17052, 2024

    arXiv 2024

  6. [6]

    Ex- ploiting style latent flows for generalizing deepfake video detection

    Jongwook Choi, Taehoon Kim, Yonghyun Jeong, Seungryul Baek, and Jongwon Choi. Ex- ploiting style latent flows for generalizing deepfake video detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 1133–1143, 2024

    work page 2024

  7. [7]

    Xception: Deep learning with depthwise separable convolutions

    François Chollet. Xception: Deep learning with depthwise separable convolutions. InProceed- ings of the IEEE conference on computer vision and pattern recognition, pages 1251–1258, 2017

    2017

  8. [8]

    On the detection of digital face manipulation

    Hao Dang, Feng Liu, Joel Stehouwer, Xiaoming Liu, and Anil K Jain. On the detection of digital face manipulation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern recognition, pages 5781–5790, 2020

    2020

Show all 53 references
  1. [9]

    Arcface: Additive angular margin loss for deep face recognition

    Jiankang Deng, Jia Guo, Niannan Xue, and Stefanos Zafeiriou. Arcface: Additive angular margin loss for deep face recognition. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4690–4699, 2019

    2019

  2. [10]

    Implicit identity leakage: The stumbling block to improving deepfake detection generalization

    Shichao Dong, Jin Wang, Renhe Ji, Jiajun Liang, Haoqiang Fan, and Zheng Ge. Implicit identity leakage: The stumbling block to improving deepfake detection generalization. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3994–4004, 2023

    2023

  3. [11]

    Protecting celebrities from deepfake with identity consistency transformer

    Xiaoyi Dong, Jianmin Bao, Dongdong Chen, Ting Zhang, Weiming Zhang, Nenghai Yu, Dong Chen, Fang Wen, and Baining Guo. Protecting celebrities from deepfake with identity consistency transformer. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition...

    2022

  4. [12]

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

    Ricard Durall, Margret Keuper, and Janis Keuper. Watch your up-convolution: Cnn based generative deep neural networks are failing to reproduce spectral distributions. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 7890–7899, 2020

    2020

  5. [13]

    Deepfakes: Faceswap

    FaceSwapDevs. Deepfakes: Faceswap. https://github.com/deepfakes/faceswap, 2019. Accessed: 2025-04-07

    2019

  6. [14]

    Exploiting fine- grained face forgery clues via progressive enhancement learning

    Qiqi Gu, Shen Chen, Taiping Yao, Yang Chen, Shouhong Ding, and Ran Yi. Exploiting fine- grained face forgery clues via progressive enhancement learning. InProceedings of the AAAI conference on artificial intelligence, volume 36, pages 735–743, 2022

    2022

  7. [15]

    Hierarchical con- trastive inconsistency learning for deepfake video detection

    Zhihao Gu, Taiping Yao, Yang Chen, Shouhong Ding, and Lizhuang Ma. Hierarchical con- trastive inconsistency learning for deepfake video detection. InEuropean conference on computer vision, pages 596–613. Springer, 2022

    2022

  8. [16]

    Leveraging real talking faces via self-supervision for robust forgery detection

    Alexandros Haliassos, Rodrigo Mira, Stavros Petridis, and Maja Pantic. Leveraging real talking faces via self-supervision for robust forgery detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14950–14962, 2022. 10

    2022

  9. [17]

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

    Alexandros Haliassos, Konstantinos V ougioukas, Stavros Petridis, and Maja Pantic. Lips don’t lie: A generalisable and robust approach to face forgery detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 5039–5049, 2021

    2021

  10. [18]

    Implicit identity driven deepfake face swapping detection

    Baojin Huang, Zhongyuan Wang, Jifan Yang, Jiaxin Ai, Qin Zou, Qian Wang, and Dengpan Ye. Implicit identity driven deepfake face swapping detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4490–4499, 2023

    2023

  11. [19]

    Deepfake detection challenge

    Kaggle. Deepfake detection challenge. https://www.kaggle.com/c/ deepfake-detection-challenge. Accessed: 2025-04-08

    2025

  12. [20]

    Friday: Mitigating unintentional facial identity in deepfake detectors guided by facial recognizers

    Younghun Kim, Myung-Joon Kwon, Wonjun Lee, and Changick Kim. Friday: Mitigating unintentional facial identity in deepfake detectors guided by facial recognizers. In2024 IEEE International Conference on Visual Communications and Image Processing (VCIP), pages 1–5. IEEE, 2024

    2024

  13. [21]

    Faceswap

    Marek Kowalski. Faceswap. https://github.com/MarekKowalski/FaceSwap, 2018. Ac- cessed: 2025-04-07

    2018

  14. [22]

    Freqblender: En- hancing deepfake detection by blending frequency knowledge.arXiv preprint arXiv:2404.13872, 2024

    Hanzhe Li, Jiaran Zhou, Yuezun Li, Baoyuan Wu, Bin Li, and Junyu Dong. Freqblender: En- hancing deepfake detection by blending frequency knowledge.arXiv preprint arXiv:2404.13872, 2024

    2024 arXiv

  15. [23]

    Faceshifter: Towards high fidelity and occlusion aware face swapping.arXiv preprint arXiv:1912.13457, 2019

    Lingzhi Li, Jianmin Bao, Hao Yang, Dong Chen, and Fang Wen. Faceshifter: Towards high fidelity and occlusion aware face swapping.arXiv preprint arXiv:1912.13457, 2019

    1912 arXiv

  16. [24]

    Face x-ray for more general face forgery detection

    Lingzhi Li, Jianmin Bao, Ting Zhang, Hao Yang, Dong Chen, Fang Wen, and Baining Guo. Face x-ray for more general face forgery detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 5001–5010, 2020

    2020

  17. [25]

    Exposing deepfake videos by detecting face warping artifacts.arXiv preprint arXiv:1811.00656, 2018

    Yuezun Li and Siwei Lyu. Exposing deepfake videos by detecting face warping artifacts.arXiv preprint arXiv:1811.00656, 2018

    2018 arXiv

  18. [26]

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

    Yuezun Li, Xin Yang, Pu Sun, Honggang Qi, and Siwei Lyu. Celeb-df: A large-scale challenging dataset for deepfake forensics. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 3207–3216, 2020

    2020

  19. [27]

    Preserving fairness generalization in deepfake detection

    Li Lin, Xinan He, Yan Ju, Xin Wang, Feng Ding, and Shu Hu. Preserving fairness generalization in deepfake detection. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16815–16825, 2024

    2024

  20. [28]

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

    Honggu Liu, Xiaodan Li, Wenbo Zhou, Yuefeng Chen, Yuan He, Hui Xue, Weiming Zhang, and Nenghai Yu. Spatial-phase shallow learning: rethinking face forgery detection in frequency domain. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages ...

    2021

  21. [29]

    Generalizing face forgery detection with high-frequency features

    Yuchen Luo, Yong Zhang, Junchi Yan, and Wei Liu. Generalizing face forgery detection with high-frequency features. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16317–16326, 2021

    2021

  22. [30]

    Capsule-forensics: Using capsule networks to detect forged images and videos

    Huy H Nguyen, Junichi Yamagishi, and Isao Echizen. Capsule-forensics: Using capsule networks to detect forged images and videos. InICASSP 2019-2019 IEEE international conference on acoustics, speech and signal processing (ICASSP), pages 2307–2311. IEEE, 2019

    2019

  23. [31]

    Core: Consistent representation learning for face forgery detection

    Yunsheng Ni, Depu Meng, Changqian Yu, Chengbin Quan, Dongchun Ren, and Youjian Zhao. Core: Consistent representation learning for face forgery detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 12–21, 2022

    2022

  24. [32]

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

    Yuyang Qian, Guojun Yin, Lu Sheng, Zixuan Chen, and Jing Shao. Thinking in frequency: Face forgery detection by mining frequency-aware clues. InEuropean conference on computer vision, pages 86–103. Springer, 2020. 11

    2020

  25. [33]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervision. InInternational conference on machine learning, pag...

    2021

  26. [34]

    Faceforensics++: Learning to detect manipulated facial images

    Andreas Rossler, Davide Cozzolino, Luisa Verdoliva, Christian Riess, Justus Thies, and Matthias Nießner. Faceforensics++: Learning to detect manipulated facial images. InProceedings of the IEEE/CVF international conference on computer vision, pages 1–11, 2019

    2019

  27. [35]

    Detecting deepfakes with self-blended images

    Kaede Shiohara and Toshihiko Yamasaki. Detecting deepfakes with self-blended images. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 18720–18729, 2022

    2022

  28. [36]

    Dual contrastive learning for general face forgery detection

    Ke Sun, Taiping Yao, Shen Chen, Shouhong Ding, Jilin Li, and Rongrong Ji. Dual contrastive learning for general face forgery detection. InProceedings of the AAAI conference on artificial intelligence, volume 36, pages 2316–2324, 2022

    2022

  29. [37]

    Efficientnet: Rethinking model scaling for convolutional neural networks

    Mingxing Tan and Quoc Le. Efficientnet: Rethinking model scaling for convolutional neural networks. InInternational conference on machine learning, pages 6105–6114. PMLR, 2019

    2019

  30. [38]

    Deferred neural rendering: Image synthesis using neural textures.Acm Transactions on Graphics (TOG), 38(4):1–12, 2019

    Justus Thies, Michael Zollhöfer, and Matthias Nießner. Deferred neural rendering: Image synthesis using neural textures.Acm Transactions on Graphics (TOG), 38(4):1–12, 2019

    2019

  31. [39]

    Face2face: Real-time face capture and reenactment of rgb videos

    Justus Thies, Michael Zollhofer, Marc Stamminger, Christian Theobalt, and Matthias Nießner. Face2face: Real-time face capture and reenactment of rgb videos. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 2387–2395, 2016

    2016

  32. [40]

    Cnn- generated images are surprisingly easy to spot

    Sheng-Yu Wang, Oliver Wang, Richard Zhang, Andrew Owens, and Alexei A Efros. Cnn- generated images are surprisingly easy to spot... for now. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8695–8704, 2020

    2020

  33. [41]

    Noise based deepfake detection via multi-head relative- interaction

    Tianyi Wang and Kam Pui Chow. Noise based deepfake detection via multi-head relative- interaction. InProceedings of the AAAI Conference on Artificial Intelligence, volume 37, pages 14548–14556, 2023

    2023

  34. [42]

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

    Yuan Wang, Kun Yu, Chen Chen, Xiyuan Hu, and Silong Peng. Dynamic graph learning with content-guided spatial-frequency relation reasoning for deepfake detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 7278–7287, 2023

    2023

  35. [43]

    Altfreezing for more general video face forgery detection

    Zhendong Wang, Jianmin Bao, Wengang Zhou, Weilun Wang, and Houqiang Li. Altfreezing for more general video face forgery detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4129–4138, 2023

    2023

  36. [44]

    Tall: Thumbnail layout for deepfake video detection

    Yuting Xu, Jian Liang, Gengyun Jia, Ziming Yang, Yanhao Zhang, and Ran He. Tall: Thumbnail layout for deepfake video detection. InProceedings of the IEEE/CVF international conference on computer vision, pages 22658–22668, 2023

    2023

  37. [45]

    Transcending forgery specificity with latent space augmentation for generalizable deepfake detection

    Zhiyuan Yan, Yuhao Luo, Siwei Lyu, Qingshan Liu, and Baoyuan Wu. Transcending forgery specificity with latent space augmentation for generalizable deepfake detection. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8984–8994, 2024

    2024

  38. [46]

    Ucf: Uncovering common features for generalizable deepfake detection

    Zhiyuan Yan, Yong Zhang, Yanbo Fan, and Baoyuan Wu. Ucf: Uncovering common features for generalizable deepfake detection. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 22412–22423, 2023

    2023

  39. [47]

    Deepfakebench: A comprehensive benchmark of deepfake detection.arXiv preprint arXiv:2307.01426, 2023

    Zhiyuan Yan, Yong Zhang, Xinhang Yuan, Siwei Lyu, and Baoyuan Wu. Deepfakebench: A comprehensive benchmark of deepfake detection.arXiv preprint arXiv:2307.01426, 2023

    2023 arXiv

  40. [48]

    Crossdf: Improving cross-domain deepfake detection with deep information decomposition

    Shanmin Yang, Hui Guo, Shu Hu, Bin Zhu, Ying Fu, Siwei Lyu, Xi Wu, and Xin Wang. Crossdf: Improving cross-domain deepfake detection with deep information decomposition. arXiv preprint arXiv:2310.00359, 2023. 12

    2023 arXiv

  41. [49]

    Exposing deep fakes using inconsistent head poses

    Xin Yang, Yuezun Li, and Siwei Lyu. Exposing deep fakes using inconsistent head poses. In ICASSP 2019-2019 IEEE international conference on acoustics, speech and signal processing (ICASSP), pages 8261–8265. IEEE, 2019

    2019

  42. [50]

    Learning natural consistency representation for face forgery video detection

    Daichi Zhang, Zihao Xiao, Shikun Li, Fanzhao Lin, Jianmin Li, and Shiming Ge. Learning natural consistency representation for face forgery video detection. InEuropean Conference on Computer Vision, pages 407–424. Springer, 2024

    2024

  43. [51]

    Istvt: interpretable spatial-temporal video transformer for deepfake detection.IEEE Transactions on Information Forensics and Security, 18:1335–1348, 2023

    Cairong Zhao, Chutian Wang, Guosheng Hu, Haonan Chen, Chun Liu, and Jinhui Tang. Istvt: interpretable spatial-temporal video transformer for deepfake detection.IEEE Transactions on Information Forensics and Security, 18:1335–1348, 2023

    2023

  44. [52]

    Multi-attentional deepfake detection

    Hanqing Zhao, Wenbo Zhou, Dongdong Chen, Tianyi Wei, Weiming Zhang, and Nenghai Yu. Multi-attentional deepfake detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2185–2194, 2021

    2021

  45. [53]

    Exploring temporal coherence for more general video face forgery detection

    Yinglin Zheng, Jianmin Bao, Dong Chen, Ming Zeng, and Fang Wen. Exploring temporal coherence for more general video face forgery detection. InProceedings of the IEEE/CVF international conference on computer vision, pages 15044–15054, 2021. 13 Supplementary Material A Appendix ...

    2021

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.