Towards Generalizable Deepfake Detection with Spatial-Frequency Collaborative Learning and Hierarchical Cross-Modal Fusion

Mengyu Qiao; Runze Tian; Yang Wang

REVIEW 5 major objections 5 minor 55 references

Towards Generalizable Deepfake Detection with Spatial-Frequency Collaborative Learning and Hierarchical Cross-Modal Fusion

T0 review · 5 major / 5 minor · reviewed 2026-08-16 · 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 deepfake detector which treats frequency evidence as a first-class modality, extracting block-wise DCT artifacts and global spectral statistics, reaches 99.58% AUC on FF++ (c23) and generalizes across datasets…

desk verdict A plausible but unverified deepfake detection architecture whose own table contradicts the headline generalizability claim, and with no code or error bars to back the numbers. read the letter →

arxiv 2504.17223 v1 pith:RM2K7CQQ submitted 2025-04-24 cs.CV

Mengyu Qiao , Runze Tian , Yang Wang This is my paper · ORCID

classification cs.CV

keywords deepfakedetectiondiscretecosinetransformfrequency-domainanalysiscross-modalfusionspatial-frequencylearningfaceforgerygeneralization

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 argues that deepfake detection should not choose between spatial and frequency evidence: a detector that extracts both and fuses them at multiple depths can recognize forgeries it has never seen. It proposes SFCL-HCMF, which runs an EfficientNet spatial stream alongside a frequency stream built on block-wise DCT, and reports 99.58% AUC on FF++ (c23), 94.21% on the heavily compressed FF++ (c40), and cross-dataset AUCs of 74.68% on Celeb-DF (v2) and 73.71% on DFDC after training only on FF++. The authors' point is that frequency-native artifacts, especially global distribution patterns, are the missing signal that keeps detectors from generalizing.

What carries the argument

The load-bearing object is the block-wise DCT representation: an image is converted to YCbCr, divided into 8x8 blocks, DCT-transformed, and each block's coefficients are flattened by zigzag scanning into a 4D tensor that preserves both spatial block positions and frequency bands. On this tensor, the local branch (SBCM plus a modified CNN) uses 3D convolutions along the frequency dimension to model intra-block coefficient correlations and inter-block transitions; the global branch (SIDA) applies horizontal, vertical, and intra-block differentials and compresses them into a 2304-dimensional vector of mean, std, skewness, and kurtosis; and the hierarchical cross-modal fusion modules (FAAE and HCMA) inject frequency context into shallow spatial features and dynamically gate deep fused features.

What would settle it

Train SFCL-HCMF on FF++ (c23) and evaluate it on forgeries produced by a generator family absent from FF++, Celeb-DF, and DFDC, with compression or post-processing not used in training; if the frame-level AUC falls to the level of the spatial-only backbone, the generalization claim is falsified. A cheaper check: compute the 2304-dimensional SIDA descriptor on real and fake images after downscaling to 128x128 and see whether the real and fake distributions still separate cleanly.

Watch

Extended reading notes

Core claim

The paper's central claim is that a deepfake detector can be made generalizable by treating frequency information as a first-class modality rather than a filter applied to spatial features. SFCL-HCMF does this in three moves: block-wise 8x8 DCT with zigzag reordering so that DCT coefficients keep spatial-frequency structure; a local branch that convolves along the spectral-band dimension to catch intra-block and inter-block anomalies; and a global branch that computes row, column, and intra-block differences of DCT blocks at original resolution and summarizes them with four statistics (mean, std, skewness, kurtosis) into a 2304-dimensional descriptor. A hierarchical fusion then uses frequency-localized attention on shallow spatial features and dynamic gating on deep fused features. The reported result is that this architecture beats prior detectors on FF++ at both compression levels and, after training only on FF++, transfers to Celeb-DF (v2) and DFDC better than or comparably to the strongest baselines.

Load-bearing premise

The load-bearing premise is that the hand-picked four statistics (mean, std, skewness, kurtosis) computed on block-wise DCT differences at the original resolution capture enough of a forgery's global spectral fingerprint; if those statistics stop separating real from fake under resizing, recompression, or a new generator, the cross-dataset gains claimed here lose their support.

Editorial extensions

If this is right

Frequency-domain clues should be treated as a native input stream in forensic detectors, not merely as augmentation of spatial features.
Detectors trained on one dataset can transfer to unseen manipulation types and compression levels, making real-world deployment more plausible.
The compact global statistical descriptor offers a resolution-preserving representation that could be used alongside spatial features for efficient screening.
The hierarchical fusion pattern, shallow attention plus deep gating, provides a reusable design for other multimodal forgery detection tasks.

Reading between the lines

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

Beyond the paper: because the SIDA descriptor is computed from DCT blocks at the original resolution and does not depend on resizing, it may also serve as a lightweight, resolution-robust forensic fingerprint; a direct test would be to use it as a standalone detector on unseen generators.
Beyond the paper: the method's face cropping and fixed 380x380 input for the spatial and local-frequency streams leave open how much of the gain survives unaligned or very-low-resolution faces; that boundary is testable and not settled by the reported benchmarks.
Beyond the paper: the same block-wise DCT plus differential-statistics recipe is not face-specific beyond the crop, so it could plausibly be applied to synthetic-image detection for non-face content such as GAN-generated textures or documents; that is an extension, not a claim of the paper.

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.

Desk Editor's Note

read the letter

The paper is a real engineering effort: block-wise DCT with 3D spectral-band convolutions, a global differential-statistics branch (SIDA), and hierarchical cross-modal fusion. That combination is new, and the in-dataset numbers on FF++ (99.58 AUC on c23, 94.21 on c40) are strong. The ablations show each component contributes, which is good practice.

But the core generalizability claim does not survive contact with Table 2. SFDG gets 75.83 AUC on Celeb-DF(v2); your method gets 74.68. On DFDC the gap is 0.07. The abstract says "outperforms state-of-the-art in both accuracy and generalizability," while Section 4.5 quietly downgrades to "superior or comparable." Those are different claims, and the table supports only the weaker one.

The other major issue is verifiability. No code, no trained models, no error bars, no multiple seeds. The baselines in Table 2 are taken from their original papers, so training protocols, face crops, and frame sampling may differ. That makes the cross-dataset comparison unreliable even when the numbers look favorable. The SIDA ablation is only on FF++(c23), so the component explicitly designed for scale-invariant generalization is never tested cross-dataset. That is a gap.

Minor but real: the method name appears as "SPCL-HCMF" in Table 1, and there is a "Glocal" typo in Section 4.6. References [5] and [6] are identical. These are not load-bearing, but they lower confidence in the manuscript's preparation.

Who is this for? Someone working on deepfake detection who wants to see a plausible new frequency-domain architecture. This is a useful idea paper, not a verified result. I would send it to peer review, but with the expectation that the authors must release code and models, add error bars and multiple seeds, and either fix or explicitly temper the abstract claim. The claim problem is fixable; the missing artifacts are a demand, not a death sentence.

Referee Report

5 major / 5 minor

Summary. The paper proposes a deepfake detection framework, SFCL-HCMF, that combines a spatial EfficientNet branch with a frequency branch built on block-wise DCT. The frequency branch has a local stream that uses a 3D-convolution SBCM and a CNN-F to model intra- and inter-block DCT correlations, and a global stream (SIDA) that computes row, column, and intra-block differential statistics over DCT sub-band features. A hierarchical cross-modal fusion mechanism (FAAE in shallow layers and HCMA in deep layers) integrates the two modalities. The method is evaluated on FaceForensics++ for in-dataset performance and on Celeb-DF(v2) and DFDC for cross-dataset generalization, with reported AUCs of 99.58% on FF++(c23), 94.21% on FF++(c40), 74.68% on Celeb-DF(v2), and 73.71% on DFDC. The central claim is that this architecture outperforms state-of-the-art methods in both accuracy and generalizability.

Significance. The architecture is coherent and the paper gives detailed equations for each module; the ablation studies in Tables 3 and 4 separate the contributions of SBCM, SIDA, FAAE, and HCMA, which is a strength. An effective local/global frequency representation combined with cross-modal fusion would be a useful addition to deepfake detection. However, the reported evidence is not yet sufficient to support the central claim. The cross-dataset comparison in Table 2 is not controlled, one baseline (SFDG) beats the proposed method on Celeb-DF(v2), and no uncertainty estimates or code/data are provided, so the reported superiority over state-of-the-art is unverified. The contribution would be significant if the claims were supported by a unified evaluation with error bars.

major comments (5)

[Section 4.5, Table 2] The statement in the abstract and Section 4.5 that the method outperforms state-of-the-art deepfake detection methods in both accuracy and generalizability is not supported by Table 2: SFDG reports 75.83 AUC on Celeb-DF(v2) while SFCL-HCMF reports 74.68, and on DFDC the reported gap is only 0.07 AUC (73.64 vs 73.71). The authors need either a unified re-run of all baselines under the same protocol or a revised, more qualified claim.
[Section 4.5, Table 2] The cross-dataset numbers appear to be transcribed from the original baseline papers, with no evidence that training splits, face cropping, frame sampling, image-level versus video-level scoring, or post-processing were identical. Without a controlled protocol, the comparisons in Table 2 are not apples-to-apples and cannot support the generalizability claim. Please provide a protocol description and, ideally, re-run the baselines under the authors' evaluation setup.
[Section 4.6, Table 3] The ablation study validates SIDA only on FF++(c23); there is no cross-dataset ablation that measures how much SIDA contributes to the Celeb-DF/DFDC generalization results. Since SIDA is the component presented in Section 3.2.1 as the scale-invariant generalization mechanism, the claimed benefit to generalization is currently unsupported.
[Section 3.2.1, Equations (4)-(8)] The four hand-designed statistics (mean, std, skewness, kurtosis) are the entire global SIDA representation, but no justification or comparison to alternative aggregations is given, and there is no robustness test under resizing or re-compression. Because the cross-dataset gains are attributed to this branch, the choice of these particular statistics is load-bearing and needs empirical support.
[Section 4.1, Tables 1-2] There is no code, data, or model release, and all results are single-run image-level Acc/AUC without standard deviations or multiple seeds. Since several reported margins over baselines are below 0.1 AUC, the quantitative claims cannot be checked or assessed for statistical significance.

minor comments (5)

[Title and Tables 1-2] The method name is written SFCL-HCMF in the title and abstract but SPCL-HCMF in Tables 1 and 2; please make the notation consistent.
[Section 4.5] The sentence comparing Two-branch and SFDG and claiming improved robustness is contradicted by Table 2 on Celeb-DF(v2) and should be corrected or qualified.
[Section 3.2.2] The paragraph beginning with '2D CNNs exhibit inherent limitations' is repeated verbatim twice in Section 3.2.2; please remove the duplicate.
[Equations (4)-(7)] The normalization uses H times W, but the feature maps in these equations have dimensions such as C by 64 by (H/8) by ((W/8)-1); please clarify the dimensions and normalization over the actual number of elements.
[References [5] and [6]] References [5] and [6] appear to be the same paper; please consolidate and renumber. Also, the asterisks on MAD and M2TR in Tables 1 and 2 are not explained in the captions.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: SFCL-HCMF is supported by external benchmark evaluations and component ablations, not by a derivation that reduces to its own inputs.

full rationale

The paper's central claims are empirical: SFCL-HCMF reports accuracy/AUC on FaceForensics++ (Table 1) and cross-dataset AUC on Celeb-DF(v2) and DFDC (Table 2), all external public benchmarks, and compares with published baselines rather than with a quantity derived from its own equations. The SIDA descriptor, SBCM, and HCMF modules are learned or fixed feature-extraction components whose contributions are tested by ablation (Table 3 and Table 4), which is standard empirical validation, not circularity. The only methodological citation used in the construction, [44], supplies the YCbCr/DCT preprocessing step, not the paper's central result, and it is not authored by the present authors; no load-bearing self-citation or uniqueness theorem is invoked. The fact that Table 2 shows SFDG [42] at 75.83 AUC on Celeb-DF(v2) versus the proposed method's 74.68 AUC, while the abstract claims generalizability superiority, is a potential correctness or protocol-comparison concern, but it is not a case of a prediction being equivalent to its inputs by construction. Consequently, no circular step satisfying the evidentiary standard is present.

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

The method introduces no new physical or mathematical entity. The free parameters are architectural choices and experimental protocol choices; the most consequential is the 3D kernel progression and the choice of four statistics. The axioms are standard deep learning assumptions plus the central domain assumption that block-wise DCT artifacts are discriminative at original resolution. No code, weights, or analysis of alternative design choices is provided, so the ledger entries cannot be further constrained.

free parameters (5)

3D kernel size progression (7, 5, 3) in SBCM = 7, 5, 3
Chosen by hand and validated through the ablation study in Table 3 rather than derived.
Four differential statistics (mean, std, skewness, kurtosis) = N/A
The global branch representation is fixed to these four statistics with no comparison against alternative aggregations, so the choice is an implicit modeling decision.
DCT block size of 8x8 = 8
Adopted from JPEG standards and prior work rather than tuned for the target task.
Number of frames per video, face crop size, and augmentation count = 270 frames and 4-fold augmentation in-dataset; 50 frames cross-dataset; 380x380 input
These experimental protocol choices affect the reported numbers but are not systematically varied.
Training hyperparameters = Adam, lr 0.001, weight decay 1e-8, batch size 20, 20 epochs
Standard configuration reported in Section 4.1; no sensitivity analysis.

assumptions (4)

domain assumption Block-wise DCT of a face image encodes forgery-relevant artifacts that survive in the original resolution.
Motivates the entire frequency pipeline in Section 3.2, but the paper provides only illustrative heatmaps and no quantitative verification that these artifacts are discriminative across unseen generators.
domain assumption The four statistics on difference maps are sufficient to summarize global forgery patterns.
Equations 4 to 8 define the global descriptor without any theoretical justification or comparison to alternative descriptors.
domain assumption The spatial backbone (EfficientNet-B4) pretrained on ImageNet provides useful features for deepfake detection.
Standard transfer-learning assumption, stated in Section 3.1 without supporting analysis for this task.
domain assumption Cropping the face region and resizing to 380x380 does not destroy the spectral artifacts used by the local branch.
The paper explicitly argues resizing hurts frequency information for the global branch, yet the spatial and local branches still resize. The interaction between this resizing and frequency artifact preservation is not analyzed.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Towards Generalizable Deepfake Detection with Spatial-Frequency Collaborative Learning and Hierarchical Cross-Modal Fusion." pith.science (2026). https://pith.science/paper/RM2K7CQQ

@misc{pith2026250417223,
  author       = {Pith},
  title        = {Pith review of: Towards Generalizable Deepfake Detection with Spatial-Frequency Collaborative Learning and Hierarchical Cross-Modal Fusion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RM2K7CQQ}},
  note         = {Machine review of arXiv:2504.17223}
}

read the original abstract

The rapid evolution of deep generative models poses a critical challenge to deepfake detection, as detectors trained on forgery-specific artifacts often suffer significant performance degradation when encountering unseen forgeries. While existing methods predominantly rely on spatial domain analysis, frequency domain operations are primarily limited to feature-level augmentation, leaving frequency-native artifacts and spatial-frequency interactions insufficiently exploited. To address this limitation, we propose a novel detection framework that integrates multi-scale spatial-frequency analysis for universal deepfake detection. Our framework comprises three key components: (1) a local spectral feature extraction pipeline that combines block-wise discrete cosine transform with cascaded multi-scale convolutions to capture subtle spectral artifacts; (2) a global spectral feature extraction pipeline utilizing scale-invariant differential accumulation to identify holistic forgery distribution patterns; and (3) a multi-stage cross-modal fusion mechanism that incorporates shallow-layer attention enhancement and deep-layer dynamic modulation to model spatial-frequency interactions. Extensive evaluations on widely adopted benchmarks demonstrate that our method outperforms state-of-the-art deepfake detection methods in both accuracy and generalizability.

Figures

Figures reproduced from arXiv: 2504.17223 by the authors.

**Figure 2.** The architecture of SFCL-HCMF. along the intra-block dimension to maintain dimensional consistency. 𝑋˜ 𝑖𝑛𝑡𝑟𝑎 = 𝑋˜ :,𝑙+1,:,: − 𝑋˜ :,𝑙,:,: ,𝑙 ∈ [1, 63] (3) [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** The block-wise DCT transform. Subsequently, four statistical metrics—mean, std, skewness, and kurtosis, are computed on the absolute values of the three derivative feature maps, as shown in equation 4-7: 𝑀𝑒𝑎𝑛𝑘 = 1 𝐻 ×𝑊 ∑︁ 𝐻 𝑖=1 ∑︁ 𝑊 𝑗=1 [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (4 more)

**Figure 5.** Figure 5: The structure of Frequency-Aware Attention En [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]

**Figure 4.** Figure 4: The pipeline of Local Branch (inter/intra-block [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 6.** Figure 6: The structure of Hybrid Cross-Modal Attention [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

**Figure 7.** Figure 7: The line charts of the differential statistical metrics [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 30 canonical work pages

[6]

Zhongjie Ba, Qingyu Liu, Zhenguang Liu, Shuang Wu, Feng Lin, Li Lu, and Kui Ren. 2024. Exposing the deception: Uncovering more forgery clues for deepfake detection. In Proceedings of the AAAI Conference on Artificial Intelligence , Vol. 38. 719–728

work page 2024
[1]

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

2018
[2]

Aayushi Agarwal, Akshay Agarwal, Sayan Sinha, Mayank Vatsa, and Richa Singh

work page
[3]

Shruti Agarwal, Hany Farid, Ohad Fried, and Maneesh Agrawala. 2020. De- tecting deep-fake videos from phoneme-viseme mismatches. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops . 660–661

2020
[4]

Martin Arjovsky and Soumith Chintala. 2017. Bottou. wasserstein gan. arXiv preprint arXiv:1701.07875 7 (2017)

arXiv 2017
[7]

Yunjey Choi, Minje Choi, Munyoung Kim, Jung-Woo Ha, Sunghun Kim, and Jaegul Choo. 2018. Stargan: Unified generative adversarial networks for multi- domain image-to-image translation. In Proceedings of the IEEE conference on computer vision and pattern recognition . 8789–8797

2018
[8]

François Chollet. 2017. Xception: Deep learning with depthwise separable con- volutions. In Proceedings of the IEEE conference on computer vision and pattern recognition. 1251–1258

2017
[9]

Umur Aybars Ciftci, Ilke Demir, and Lijun Yin. 2020. Fakecatcher: Detection of synthetic portrait videos using biological signals. IEEE transactions on pattern analysis and machine intelligence (2020)

work page 2020

Show all 55 references

[10]

Brian Dolhansky, Joanna Bitton, Ben Pflaum, Jikuo Lu, Russ Howes, Menglin Wang, and Cristian Canton Ferrer. 2020. The deepfake detection challenge (dfdc) dataset. arXiv preprint arXiv:2006.07397 (2020)

2020 arXiv
[11]

Shichao Dong, Jin Wang, Jiajun Liang, Haoqiang Fan, and Renhe Ji. 2022. Explain- ing deepfake detection by analysing image matching. In European conference on computer vision. Springer, 18–35

2022
[12]

Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xi- aohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. 2020. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv prepri...

2020 arXiv
[13]

Ricard Durall, Margret Keuper, Franz-Josef Pfreundt, and Janis Keuper. 2019. Unmasking deepfakes with simple features. arXiv preprint arXiv:1911.00686 (2019)

2019 arXiv
[14]

Shuaijv Fang, Zhiyong Zhang, and Bin Song. 2025. Deepfake Detection Model Combining Texture Differences and Frequency Domain Information.ACM Trans- actions on Privacy and Security 28, 2 (2025), 1–16

2025
[15]

Steven Fernandes, Sunny Raj, Eddy Ortiz, Iustina Vintila, Margaret Salter, Gor- dana Urosevic, and Sumit Jha. 2019. Predicting heart rate variations of deepfake videos using neural ode. In Proceedings of the IEEE/CVF international conference on computer vision workshops . 0–0

2019
[16]

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

2022
[17]

Zhenliang He, Wangmeng Zuo, Meina Kan, Shiguang Shan, and Xilin Chen
[18]

Javier Hernandez-Ortega, Ruben Tolosana, Julian Fierrez, and Aythami Morales
[19]

Tero Karras, Timo Aila, Samuli Laine, and Jaakko Lehtinen. 2017. Progressive growing of gans for improved quality, stability, and variation. arXiv preprint arXiv:1710.10196 (2017)

2017 arXiv
[20]

Tero Karras, Samuli Laine, and Timo Aila. 2019. A style-based generator ar- chitecture for generative adversarial networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 4401–4410

2019
[21]

Jiaming Li, Hongtao Xie, Jiahong Li, Zhongyuan Wang, and Yongdong Zhang
[22]

Jiaming Li, Hongtao Xie, Lingyun Yu, and Yongdong Zhang. 2022. Wavelet- enhanced weakly supervised local feature learning for face forgery detection. In Proceedings of the 30th ACM international conference on multimedia . 1299–1308

2022
[23]

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

2019 arXiv
[24]

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

2020
[25]

In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

Frequency-aware discriminative feature learning supervised by single- center loss for face forgery detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 6458–6467
[26]

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

2024
[27]

Decheng Liu, Tao Chen, Chunlei Peng, Nannan Wang, Ruimin Hu, and Xinbo Gao. 2024. Attention Consistency Refined Masked Frequency Forgery Represen- tation for Generalizing Face Forgery Detection. IEEE Transactions on Information Forensics and Security (2024)

2024
[28]

Honggu Liu, Xiaodan Li, Wenbo Zhou, Yuefeng Chen, Yuan He, Hui Xue, Weiming Zhang, and Nenghai Yu. 2021. 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
[29]

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

2020
[30]

Iacopo Masi, Aditya Killekar, Royston Marian Mascarenhas, Shenoy Pratik Guru- datt, and Wael AbdAlmageed. 2020. Two-branch recurrent network for isolating deepfakes in videos. In Computer vision–ECCV 2020: 16th European conference, glasgow, UK, August 23–28, 2020, proceedings,...

2020
[31]

Changtao Miao, Zichang Tan, Qi Chu, Nenghai Yu, and Guodong Guo. 2022. Hier- archical frequency-assisted interactive networks for face manipulation detection. IEEE Transactions on Information Forensics and Security 17 (2022), 3008–3021

2022
[32]

Dat Nguyen, Nesryne Mejri, Inder Pal Singh, Polina Kuleshova, Marcella Astrid, Anis Kacem, Enjie Ghorbel, and Djamila Aouada. 2024. Laa-net: Localized artifact attention network for quality-agnostic and generalizable deepfake detection. In Proceedings of the IEEE/CVF Conferenc...

2024
[33]

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

2021
[34]

Hua Qi, Qing Guo, Felix Juefei-Xu, Xiaofei Xie, Lei Ma, Wei Feng, Yang Liu, and Jianjun Zhao. 2020. Deeprhythm: Exposing deepfakes with attentional visual heartbeat rhythms. In Proceedings of the 28th ACM international conference on multimedia. 4318–4327

2020
[35]

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

2020
[36]

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

2019
[37]

Yuval Nirkin, Lior Wolf, Yosi Keller, and Tal Hassner. 2021. Deepfake detection based on discrepancies between faces and their context. IEEE transactions on pattern analysis and machine intelligence 44, 10 (2021), 6111–6121

2021
[38]

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

2019
[39]

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

2016
[40]

Junke Wang, Zuxuan Wu, Wenhao Ouyang, Xintong Han, Jingjing Chen, Yu- Gang Jiang, and Ser-Nam Li. 2022. M2tr: Multi-modal multi-scale transformers for deepfake detection. In Proceedings of the 2022 international conference on multimedia retrieval. 615–623

2022
[41]

Chuangchuang Tan, Yao Zhao, Shikui Wei, Guanghua Gu, Ping Liu, and Yunchao Wei. 2024. Frequency-aware deepfake detection: Improving generalizability through frequency space domain learning. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 38. 5052–5060

2024
[42]

Yuan Wang, Kun Yu, Chen Chen, Xiyuan Hu, and Silong Peng. 2023. 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. 7278–7287

2023
[43]

Xi Wu, Zhen Xie, YuTao Gao, and Yu Xiao. 2020. Sstnet: Detecting manipulated faces through spatial, steganalysis and temporal features. In ICASSP 2020-2020 IEEE international conference on acoustics, speech and signal processing (ICASSP) . IEEE, 2952–2956

2020
[44]

Kai Xu, Minghai Qin, Fei Sun, Yuhao Wang, Yen-Kuang Chen, and Fengbo Ren
[45]

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

2023
[46]

Qilin Yin, Wei Lu, Bin Li, and Jiwu Huang. 2023. Dynamic difference learning with spatio–temporal correlation for deepfake video detection. IEEE Transactions on Information Forensics and Security 18 (2023), 4046–4058

2023
[47]

Dengyong Zhang, Jiahao Chen, Xin Liao, Feng Li, Jiaxin Chen, and Gaobo Yang
[48]

Daichi Zhang, Fanzhao Lin, Yingying Hua, Pengju Wang, Dan Zeng, and Shiming Ge. 2022. Deepfake video detection with spatiotemporal dropout transformer. In Proceedings of the 30th ACM international conference on multimedia . 5833–5841

2022
[49]

InProceedings of the IEEE/CVF conference on computer vision and pattern recognition

Learning in the frequency domain. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition . 1740–1749
[50]

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

2019
[55]

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

2021
[56]

Peng Zhou, Xintong Han, Vlad I Morariu, and Larry S Davis. 2017. Two-stream neural networks for tampered face detection. In 2017 IEEE conference on computer vision and pattern recognition workshops (CVPRW) . IEEE, 1831–1839

2017
[2019]

IEEE transactions on image processing 28, 11 (2019), 5464–5478

Attgan: Facial attribute editing by only changing what you want. IEEE transactions on image processing 28, 11 (2019), 5464–5478

2019
[2020]

arXiv preprint arXiv:2010.00400 (2020)

Deepfakeson-phys: Deepfakes detection based on heart rate estimation. arXiv preprint arXiv:2010.00400 (2020)

2020 arXiv
[2021]

In 2021 16th IEEE international conference on automatic face and gesture recognition (FG 2021)

MD-CSDNetwork: Multi-domain cross stitched network for deepfake detection. In 2021 16th IEEE international conference on automatic face and gesture recognition (FG 2021). IEEE, 1–8

2021
[2024]

IEEE Transactions on Circuits and Systems for Video Technology (2024)

Face forgery detection via multi-feature fusion and local enhancement. IEEE Transactions on Circuits and Systems for Video Technology (2024)

2024

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [6]

Zhongjie Ba, Qingyu Liu, Zhenguang Liu, Shuang Wu, Feng Lin, Li Lu, and Kui Ren. 2024. Exposing the deception: Uncovering more forgery clues for deepfake detection. In Proceedings of the AAAI Conference on Artificial Intelligence , Vol. 38. 719–728

work page 2024

[2] [1]

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

2018

[3] [2]

Aayushi Agarwal, Akshay Agarwal, Sayan Sinha, Mayank Vatsa, and Richa Singh

work page

[4] [3]

Shruti Agarwal, Hany Farid, Ohad Fried, and Maneesh Agrawala. 2020. De- tecting deep-fake videos from phoneme-viseme mismatches. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops . 660–661

2020

[5] [4]

Martin Arjovsky and Soumith Chintala. 2017. Bottou. wasserstein gan. arXiv preprint arXiv:1701.07875 7 (2017)

arXiv 2017

[6] [7]

Yunjey Choi, Minje Choi, Munyoung Kim, Jung-Woo Ha, Sunghun Kim, and Jaegul Choo. 2018. Stargan: Unified generative adversarial networks for multi- domain image-to-image translation. In Proceedings of the IEEE conference on computer vision and pattern recognition . 8789–8797

2018

[7] [8]

François Chollet. 2017. Xception: Deep learning with depthwise separable con- volutions. In Proceedings of the IEEE conference on computer vision and pattern recognition. 1251–1258

2017

[8] [9]

Umur Aybars Ciftci, Ilke Demir, and Lijun Yin. 2020. Fakecatcher: Detection of synthetic portrait videos using biological signals. IEEE transactions on pattern analysis and machine intelligence (2020)

work page 2020