A Contemporary Survey on Deepfake Detection: Datasets, Algorithms, and Challenges
Open accessThe DOI matches its Crossref record; 37 of the 37 references it cites were found in a registry or an index.
Abstract
Deepfakes are notorious for their unethical and malicious applications to achieve economic, political, and social reputation goals. Recent years have seen widespread facial forgery, which does not require technical skills. Since the development of generative adversarial networks (GANs) and diffusion models (DMs), deepfake generation has been moving toward better quality. Therefore, it is necessary to find an effective method to detect fake media. This contemporary survey provides a comprehensive overview of several typical facial forgery detection methods proposed from 2019 to 2023. We also analyze and group them into four categories in terms of their feature extraction methods and network architectures: traditional convolutional neural network (CNN)-based detection, CNN backbone with semi-supervised detection, transformer-based detection, and biological signal detection. Furthermore, it summarizes several representative deepfake detection datasets with their advantages and disadvantages. Finally, we evaluate the performance of these detection models with respect to different datasets by comparing their evaluating metrics. Across all experimental results on these state-of-the-art detection models, we find that the accuracy is largely degraded if we utilize cross-dataset evaluation. These results will provide a reference for further research to develop more reliable detection algorithms.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Detecting manipulated content is essential to mitigate the risks of misinformation, identity fraud, and media integrity threats [9] while also serving as the foundation for forensic analysis, attribution, and authentication.”
From Deepfake Media Forensics: Status and Future Challenges (2025) · cites this paper as [9]
Experimental — accuracy not yet measuredThe check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“This approach eliminates the need for extensive infrastructure and centralized computation, making it highly adaptable to diverse and distributed CCTV systems [31].”
From FL-TENB4: A Federated-Learning-Enhanced Tiny EfficientNetB4-Lite Approach for Deepfake Detection in CCTV Environments (2025) · cites this paper as [31]
Experimental — accuracy not yet measuredThe check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“The Federated Learning framework not only supports decentralized training but also ensures scalability by allowing multiple devices to collaboratively train a global model.”
From FL-TENB4: A Federated-Learning-Enhanced Tiny EfficientNetB4-Lite Approach for Deepfake Detection in CCTV Environments (2025) · cites this paper as [31]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 1×
“The backbone selects ResNet-34 [35] initialized by pre-trained weights on ImageNet for extracting pre-processed video frames’ features.”
MethodVery Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 1×
“Moreover, Capsule Network [23] can use fewer parameters and data to perform similarly to CNN. Specifically, Nguyen et al. [23] proposed a VGG-19 [24] feature extraction backbone, which is from the first layer to the third layer as the backbone and pre-trained in ILSVRC datasets [25], able to reduce…”
MethodSemi-supervised classification with graph convolutional networks
Kipf, T. N., & Welling, M. (2016) · arXiv · doi:10.48550/arxiv.1609.02907 · Cited by 24,853 · cited 1×
“…also combine multi-modal information; for example, Stefanov et al. [50] proposed a method (See Figure 13) to extract physiological signals and utilized the graph convolutional network (GCN) [51] to fuse video and physiological signal and detect the dissimilarity between audio and video modality.”
MethodXception: Deep Learning with Depthwise Separable Convolutions
Chollet, F. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.195 · Cited by 20,380 · cited 2×
“For example, there is a video transformer with the incremental learning approach [42] published in 2021, which utilizes Xception [29] as the backbone and 12 transformer blocks to learn feature correlations.”
BackgroundMethodGenerative adversarial networks
Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2020) · Communications of the ACM · doi:10.1145/3422622 · Cited by 18,384
Unsupervised representation learning with deep convolutional generative adversarial networks
Radford, A., Metz, L., & Chintala, S. (2015) · arXiv · doi:10.48550/arxiv.1511.06434 · Cited by 12,235 · cited 1×
“…with 3D landmark face swap and auto-encoders [3] to generate fake media; however, the trend of deepfake generation nowadays involves more powerful generative models such as generative and adversarial networks (GANs) [4,5] and diffusion models (DMs) [6] for creating more realistic counterfeit media.”
BackgroundA Style-Based Generator Architecture for Generative Adversarial Networks
Karras, T., Laine, S., & Aila, T. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00453 · Cited by 11,714 · cited 1×
“There are other higher-quality deepfake datasets created by extensive application of the GAN-based method; for example, DFFD [17], which was published in 2020, created an entire synthesis of faces by StyleGAN [18].”
MethodDenoising Diffusion Probabilistic Models
Ho, J., Jain, A., & Abbeel, P. (2020) · arXiv · doi:10.48550/arxiv.2006.11239 · Cited by 8,712 · cited 1×
“…with 3D landmark face swap and auto-encoders [3] to generate fake media; however, the trend of deepfake generation nowadays involves more powerful generative models such as generative and adversarial networks (GANs) [4,5] and diffusion models (DMs) [6] for creating more realistic counterfeit media.”
BackgroundDynamic Routing Between Capsules
Sabour, S., Frosst, N., & Hinton, G. E. (2017) · arXiv · doi:10.48550/arxiv.1710.09829 · Cited by 4,167 · cited 1×
“Capsule Network [22] itself was not a new term in the field of deep learning, and it was first proposed to solve the problem that traditional convolutional neural networks (CNNs) have limited applicability in “inverse graphics”.”
BackgroundNoise-contrastive estimation: A new estimation principle for unnormalized statistical models
Gutmann, M. U., & Hyvärinen, A. (2010) · Edinburgh Research Explorer (University of Edinburgh) · Cited by 2,992
Self-Attention with Relative Position Representations
Shaw, P., Uszkoreit, J., & Vaswani, A. (2018) · Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers) · doi:10.18653/v1/n18-2074 · Cited by 2,505 · cited 1×
“A temporal attention block is inserted after each spatial attention block, and the relative position embedding [47] applied on each temporal attention block is used to distinguish the order of frames.”
MethodRich Models for Steganalysis of Digital Images
Fridrich, J., & Kodovsky, J. (2012) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2012.2190402 · Cited by 2,144 · cited 1×
“…views generation (DVG) module with four data enhancements to eliminate task-irrelevant contextual factors—random patch, high-frequency enhancement [37], frame shift, and corresponding mix-up, which are selected with certain probability—and randomly combine with traditional data augmentation to…”
BackgroundCeleb-DF: A Large-Scale Challenging Dataset for DeepFake Forensics
Li, Y., Yang, X., Sun, P., Qi, H., & Lyu, S. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00327 · Cited by 1,846 · cited 2×
“The Celeb-DF V2 [12] database followed the same operations, obtaining training, validating, and testing data values of 24827, 8275, and 8278, respectively.”
MethodFace X-Ray for More General Face Forgery Detection
Li, L., Bao, J., Zhang, T., Yang, H., Chen, D., Wen, F., & Guo, B. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00505 · Cited by 1,211
In Ictu Oculi: Exposing AI Created Fake Videos by Detecting Eye Blinking
Li, Y., Chang, M.-C., & Lyu, S. (2018) · 2018 IEEE International Workshop on Information Forensics and Security (WIFS) · doi:10.1109/wifs.2018.8630787 · Cited by 1,089
Determining Image Origin and Integrity Using Sensor Noise
Chen, M., Fridrich, J., Goljan, M., & Lukas, J. (2008) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2007.916285 · Cited by 893 · cited 1×
“More specifically, it regards the manipulated images to contain different source features, which come from the imaging pipeline (photo-response non-uniformity [33]) and encoding approaches (JPEG compression pattern [34]) at different locations.”
BackgroundExposing DeepFake Videos By Detecting Face Warping Artifacts
Li, Y., & Lyu, S. (2018) · arXiv · doi:10.48550/arxiv.1811.00656 · Cited by 785 · cited 1×
“Most defense methods focus on detecting suspicious artifacts of fake media such as eye blinking, blending boundaries [30], and face warping [31].”
BackgroundOn the Detection of Digital Face Manipulation
Dang, H., Liu, F., Stehouwer, J., Liu, X., & Jain, A. K. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00582 · Cited by 764
Deepfake Detection: A Systematic Literature Review
Rana, M. S., Nobi, M. N., Murali, B., & Sung, A. H. (2022) · IEEE Access · doi:10.1109/access.2022.3154404 · Cited by 573 · cited 2×
“In addition, Rana et al. [7] found that the top ten pornographic websites have released over 1790 deepfake videos to transfer celebrities’ faces to porn stars’ faces.”
BackgroundMethodDeep learning for deepfakes creation and detection: A survey
Nguyen, T. T., Nguyen, Q. V. H., Nguyen, D. T., Nguyen, D. T., Huynh-The, T., Nahavandi, S., Nguyen, T. T., Pham, Q.-V., & Nguyen, C. M. (2022) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2022.103525 · Cited by 464 · cited 1×
“Nguyen et al. [20] directly categorized based on data type, i.e., video detection and image detection.”
BackgroundLearning Self-Consistency for Deepfake Detection
Zhao, T., Xu, X., Xu, M., Ding, H., Xiong, Y., & Xia, W. (2021) · 2021 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.01475 · Cited by 383 · cited 1×
“Zhao et al. [32] proposed a method for finding out the inconsistency of content-independent and spatially local information within manipulated images by designing a multi-task learning architecture, as shown in Figure 8, including the consistency branch and classification branch.”
MethodSample and Computation Redistribution for Efficient Face DetectionPreprint
Guo, J., Deng, J., Lattas, A., & Zafeiriou, S. P. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.04714 · Cited by 245 · cited 1×
“The main facial manipulations are representative, which include DeepFakes, Face2Face, FaceSwap, FaceShifter, and Neural Textures methods, and data are of random compression levels and sizes [14].”
MethodAligned and non-aligned double JPEG detection using convolutional neural networks
Barni, M., Bondi, L., Bonettini, N., Bestagini, P., Costanzo, A., Maggini, M., Tondi, B., & Tubaro, S. (2017) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2017.09.003 · Cited by 221
Dual Contrastive Learning for General Face Forgery Detection
Sun, K., Yao, T., Chen, S., Ding, S., Li, J., & Ji, R. (2022) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v36i2.20130 · Cited by 209 · cited 2×
“Recently, Youtu Lab [36] proposed a dual-contrastive learning architecture that is aimed at distinguishing the authenticity of faces by different data augmentation.”
BackgroundMethodSpatiotemporal Inconsistency Learning for DeepFake Video Detection
Gu, Z., Chen, Y., Yao, T., Ding, S., Li, J., Huang, F., & Ma, L. (2021) · Proceedings of the 29th ACM International Conference on Multimedia · doi:10.1145/3474085.3475508 · Cited by 189 · cited 1×
“Thus, Gu et al. presented a detection work [40] based on learning fake videos’ spatial and temporal inconsistency.”
BackgroundISTVT: Interpretable Spatial-Temporal Video Transformer for Deepfake Detection
Zhao, C., Wang, C., Hu, G., Chen, H., Liu, C., & Tang, J. (2023) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2023.3239223 · Cited by 184 · cited 2×
“Following this inspiration, Zhao et al. [46] considered a video transformer that can jointly study spatial and temporal information in fake videos and has better performance and generalization in the deepfake detection field.”
MethodZhou, T., Wang, W., Liang, Z., & Shen, J. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.00572 · Cited by 156 · cited 2×
“This experiment only used the FaceForensics++ [13] database as the training dataset.”
BackgroundMethodDeepFake video detection using convolutional vision transformer
Wodajo, D., & Atnafu, S. (2021) · arXiv · doi:10.48550/arxiv.2102.11126 · Cited by 147 · cited 1×
“In addition, most studies are still using CNN as a feature extractor and trans-former block to classify deepfake images such as the convolutional vision transformer [43].”
MethodCORE: Consistent Representation Learning for Face Forgery Detection
Ni, Y., Meng, D., Yu, C., Quan, C., Ren, D., & Zhao, Y. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw56347.2022.00011 · Cited by 136 · cited 1×
“COnsistency REpresentation Learning of Forgery Detection (CORE) [27] is an effective network that can capture different representations between two data augmentations and regularize feature similarity by the cosine distance to enhance consistency so that it can achieve relatively good results in…”
BackgroundUse of a Capsule Network to Detect Fake Images and Videos
Nguyen, H. H., Yamagishi, J., & Echizen, I. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.12467 · Cited by 121 · cited 1×
“Moreover, Capsule Network [23] can use fewer parameters and data to perform similarly to CNN. Specifically, Nguyen et al. [23] proposed a VGG-19 [24] feature extraction backbone, which is from the first layer to the third layer as the backbone and pre-trained in ILSVRC datasets [25], able to reduce…”
Method(2020) · doi:10.1007/978-3-030-68821-9_38 · Cited by 113
Deep learning for deepfakes creation and detection: a survey
Nguyen, T. T., Nguyen, Q. V. H., Nguyen, D. T., Nguyen, D. T., Huynh-The, T., Nahavandi, S., Nguyen, T. T., Pham, Q.-V., & Nguyen, C. M. (2022) · SSRN Electronic Journal · doi:10.2139/ssrn.4030341 · Cited by 112
DFDT: An End-to-End DeepFake Detection Framework Using Vision Transformer
Khormali, A., & Yuan, J.-S. (2022) · Applied Sciences · doi:10.3390/app12062953 · Cited by 88 · cited 1×
“DFDT [44] proposes an end-to-end deepfake detection framework (see Figure 11) using vision transformers that can basically solve the problem that traditional CNN cannot obtain the correlation between spatial patches and the information loss caused by the receptive field.”
MethodVideo Transformer for Deepfake Detection with Incremental Learning
Khan, S. A., & Dai, H. (2021) · Proceedings of the 29th ACM International Conference on Multimedia · doi:10.1145/3474085.3475332 · Cited by 79 · cited 1×
“For example, there is a video transformer with the incremental learning approach [42] published in 2021, which utilizes Xception [29] as the backbone and 12 transformer blocks to learn feature correlations.”
MethodA landscape view of deepfake techniques and detection methods
Abdulreda, A. S., & Obaid, A. J. (2022) · International journal of nonlinear analysis and applications · doi:10.22075/ijnaa.2022.5580 · Cited by 59 · cited 1×
“Although deepfakes were initially associated with entertainment such as movie visual effects, camera filters, and digital avatars [1], they are defined as “believable generated media by Deep Neural Network” and have evolved into a mainstream tool for facial forgery.”
BackgroundLocal attention and long-distance interaction of rPPG for deepfake detection
Wu, J., Zhu, Y. M., Jiang, X., Liu, Y., & Lin, J. (2023) · The Visual Computer · doi:10.1007/s00371-023-02833-x · Cited by 54 · cited 1×
“Wu et al. [49] realized that facial manipulation progress is inevitable with sudden facial color changes in some periods, and rPPG could become an indicator for deepfake detection.”
BackgroundE-Learning Modeling Technique and Convolution Neural Networks in Online Education
Alahmari, F. S., Naim, A., & Alqahtani, H. (2023) · River Publishers eBooks · doi:10.1201/9781003393030-10 · Cited by 30
Research works printed but not linked: 8
- Zhang, 2022, Overview of Facial Deepfake Video Detection Methods, J. Front. Comput. Sci. Technol., 17, 1
- Zhong, Z., Zheng, L., Kang, G., Li, S., and Yang, Y. (2020, January 7–12). Random Erasing data argumentation. Proceedings of the AAAI, New York, NY, USA.
- Shi, X., Chen, Z., Wang, H., and Yeung, D. (2015, January 7–12). Convolutional LSTM Network: A Machine Learning Approach for Precipitation Network. Proceedings of the Advances in Neural Information Processing Systems, Montreal, QC, Canada.
- Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., and Gelly, S. (May, January 26). An Image is Worth 16 × 16 Words: Transformers for Image Recognition at Scale. Proceedings of the International Conference on Learning Representations, Virtual.
- Trockman, A., and Zico Kolter, J. Patches Are All You Need? In Proceedings of the Computer Vision and Pattern Recognition, New Orleans, LA, USA, 18–24 June 2022.
- Ciftci, U.A., Demir, İ., and Yin, L. (October, January 8). How Do the Hearts of Deep Fakes Beat? Deep Fake Source Detection via Interpreting Residuals with Biological Signals. Proceedings of the IEEE International Joint Conference on Biometrics (IJCB), Houston, TX, USA.
- Stefanov, K., Paliwal, B., and Dhall, A. (2022, January 18–24). Visual Representation of Physiological Signals for Fake Video Detection. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, USA.
- Liu, 2020, Multi-task temporal shift attention networks for on-device contactless vital measurements, Adv. Neural Inf. Process. Syst., 33, 19400
Cited by
Deepfake video detection: challenges and opportunities
Kaur, A., Noori Hoshyar, A., Saikrishna, V., Firmin, S., & Xia, F. (2024) · Artificial Intelligence Review · doi:10.1007/s10462-024-10810-6 · Cited by 214
“A recent survey by Gong and Li (2024) has classified deepfake detection methods as conventional CNN-based detection, CNN with semi-supervised detection, transformer-based detection, and biological signal detection.”
Deepfake Media Forensics: Status and Future Challenges
Amerini, I., Barni, M., Battiato, S., Bestagini, P., Boato, G., Bruni, V., Caldelli, R., De Natale, F., De Nicola, R., Guarnera, L., et al. (2025) · Journal of Imaging · doi:10.3390/jimaging11030073 · Cited by 113
Babaei, R., Cheng, S. S. M., Duan, R., & Zhao, S. (2025) · Journal of Sensor and Actuator Networks · doi:10.3390/jsan14010017 · Cited by 103
“A significant challenge in advancing detection methods is the lack of comprehensive datasets that include both AI-generated and human-generated content [150].”
Khan, A. A., Laghari, A. A., Inam, S. A., Ullah, S., Shahzad, M., & Syed, D. (2025) · Discover Computing · doi:10.1007/s10791-025-09550-0 · Cited by 98
“To assist assure the authenticity of media content, research is required to determine how blockchain can be included into deepfake detection algorithms to provide transparent and safe tracking of the content’s origin [94, 95].”
Deepfake video detection methods, approaches, and challenges
Alrashoud, M. (2025) · Alexandria Engineering Journal · doi:10.1016/j.aej.2025.04.007 · Cited by 49
Metadata
- Authors
- Gong, L. Y., & Li, X. J.
- Affiliations
- Auckland University of Technology (NZ)
- Corresponding author
- Gong, Liang Yu
- Li, Xue Jun
- Venue
- Electronics
- Publisher
- MDPI AG
- Published
- 2024-01-31
- Volume
- 13
- Issue
- 3
- Pages
- 585
- ISSN
- 2079-9292 (ISSN-L)
- OpenAlex
- W4391404116
- Cited by
- 75
- References
- 36
- Citations by year
- 2026: 28 · 2025: 39 · 2024: 8
- FWCI
- 10.47
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Electronics (publishedVersion, cc-by)
- Tuwhera (Auckland University of Technology) (submittedVersion, cc-by)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Digital Media Forensic Detection
- Social Sciences › Sociology and Political Science › Misinformation and Its Impacts
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Generative Adversarial Networks and Image Synthesis
- Keywords
- deepfake detection, generative adversarial networks, diffusion models, convolutional neural networks, transformer-based detection, biological signal detection, cross-dataset evaluation, detection accuracy
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.