A Review of Deepfake Techniques: Architecture, Detection and Datasets
Open accessThe DOI matches its Crossref record; 93 of the 93 references it cites were found in a registry or an index.
Abstract
Driven by continuous advancements in artificial intelligence, especially deep learning, the level of realism associated with deepfake technology continues to improve year after year, which poses unprecedented challenges to the field of deepfake detection. The boundary between what we as humans can detect as real or fake becomes evermore blurred as new generations of algorithms such as Dall-E 3 and Stable Diffusion are released. This paper provides a comprehensive study into the landscape of deepfake detection, exploring in-depth the key challenges, recognising recent successes, and suggesting promising avenues for future research. A meta-literature review is conducted to identify the current challenges and future directions, which form the foundation of this work. They are investigated by analysing state-of-the-art research with a focus on the key components that are crucial to the design of a deepfake detector, i.e., the architecture, detection methods and datasets. A major challenge identified by this study is the lack of dataset diversity leading to unfair attribute representation. This must be addressed by improving standardisation on dataset ethics and privacy. This is one of the main reasons for the insufficient generalisation capability of current deepfake detectors as demonstrated by their unsatisfactory performance when faced with unseen data or data in the wild. This literature review provides deepfake detection researchers and practitioners with the latest information that will serve as a vital resource for their continued and important activity, now and in the future.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
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?
In the citing paper
“Federated Learning enables decentralized model training, where sensitive data remain on local devices, and only model updates are shared with a central server.”
From FL-TENB4: A Federated-Learning-Enhanced Tiny EfficientNetB4-Lite Approach for Deepfake Detection in CCTV Environments (2025) · cites this paper as [29]
In the citing paper
“Traditional deep learning approaches often rely on centralized data processing, which poses significant risks to data privacy and security.”
From FL-TENB4: A Federated-Learning-Enhanced Tiny EfficientNetB4-Lite Approach for Deepfake Detection in CCTV Environments (2025) · cites this paper as [29]
In the citing paper
“This approach not only preserves user privacy but also reduces the risks associated with data breaches and unauthorized access, aligning with the principles of modern data protection regulations [29].”
From FL-TENB4: A Federated-Learning-Enhanced Tiny EfficientNetB4-Lite Approach for Deepfake Detection in CCTV Environments (2025) · cites this paper as [29]
In the citing paper
“Federated Learning enables decentralized model training, where sensitive data remain on local devices, and only model updates are shared with a central server.”
From FL-TENB4: A Federated-Learning-Enhanced Tiny EfficientNetB4-Lite Approach for Deepfake Detection in CCTV Environments (2025) · cites this paper as [29]
Citations
Cites
1 retracted
RETRACTED ARTICLE: Deepfake detection using rationale-augmented convolutional neural network
Ahmed, S. R., & Sonuç, E. (2021) · Applied Nanoscience · doi:10.1007/s13204-021-02072-3 · Cited by 125 · cited 3×
“Finally, although over seventy-percent of the papers apply evaluation to both inter and intra-dataset comparison, some studies limit their evaluation to inter-dataset only [54].”
BackgroundMethodInfluentialRetractedVaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404 · cited 1×
“The Transformer was proposed by Vaswani et al. [36] as an alternative to CNN and Recurrent Neural Network architectures in 2017 and is based on the use of self-attention.”
MethodGradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805 · cited 2×
“According to LeCun et al. [29], the use of back-propagation in the network can help reduce the number of trainable parameters since the weights in the feature extractor can be learned through a shared scheme.”
BackgroundMethodDensely Connected Convolutional Networks
Huang, G., Liu, Z., Van Der Maaten, L., & Weinberger, K. Q. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.243 · Cited by 48,566
Going deeper with convolutions
Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., & Rabinovich, A. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298594 · Cited by 48,562
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2020) · arXiv · doi:10.48550/arxiv.2010.11929 · Cited by 47,472 · cited 1×
“The Vision Transformer (ViT) was proposed by Doso-vitskiy et al. [37] in 2021 and builds on the success of the Transformer for Natural Language Processing.”
BackgroundRethinking the Inception Architecture for Computer Vision
Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., & Wojna, Z. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.308 · Cited by 32,138
A Unified Approach to Interpreting Model Predictions
Lundberg, S., & Lee, S.-I. (2017) · arXiv · doi:10.48550/arxiv.1705.07874 · Cited by 23,707 · cited 1×
“Local Interpretable Model-Agnostic Explanations and Shapley Additive Explanation [106] are two examples of existing techniques designed to provide model interpretability.”
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 1×
“In 2017, Chollet [33] presented the Xception network as a parameter-efficient adaptation to the Inception-V3 [34] architecture, where stackable depthwise separable convolu-tions are used as a replacement for the Inception module while achieving notably improved performance.”
MethodBackpropagation Applied to Handwritten Zip Code Recognition
LeCun, Y., Boser, B., Denker, J. S., Henderson, D., Howard, R. E., Hubbard, W., & Jackel, L. D. (1989) · Neural Computation · doi:10.1162/neco.1989.1.4.541 · Cited by 11,891 · cited 1×
“Figure 2 One of the most prominent contributions to research on Convolutional Neural Networks (CNNs) was conducted by LeCun et al. [28] in 1989 to overcome the challenge of performing numerical character recognition and classification through ML.”
Background
Research works printed but not linked: 16
- Sample, 2020, What Are Deepfakes and How Can You Spot Them
- 2022, Facelab
- 2022, FaceApp: Face Editor
- 2023, Publications—Dimensions
- Hong, 2022, EVA3D: Compositional 3D human generation from 2D image collections, arXiv:2210.04888 · arXiv
- Betker, 2023, Improving image generation with better captions, Comput. Sci., 2, 8
- 2024, Stable Diffusion 2.0 Release
- Goodfellow, Generative adversarial nets, 1
- 2024, Fake Biden Robocall Tells Voters to Skip New Hampshire Primary Election
- 2024, Taylor Swift Deepfakes Spark Calls in Congress for New Legislation
- 2024, Baltimore High School Teacher Arrested Over Deepfake Racist Audio of Principal
- Peng, 2021, Mitigating dataset harms requires stewardship: Lessons from 1000 papers, arXiv:2108.02922 · arXiv
- Schuhmann, 2022, LAION-5B: An open large-scale dataset for training next generation image-text models, arXiv:2210.08402 · arXiv
- Growcoot, 2023, Shocked Artist Finds Private Medical Photos in AI Training Data Set
- Le, 2023, Why do deepfake detectors fail?, arXiv:2302.13156 · arXiv
- Gangan, 2023, Exploring fairness in pre-trained visual transformer based natural and GAN generated image detection systems and understanding the impact of image compression in fairness, arXiv:2310.12076 · arXiv
Cited by
Zhang, Y., Pang, Z., Huang, S., Wang, C., & Zhou, X. (2025) · Journal of King Saud University - Computer and Information Sciences · doi:10.1007/s44443-025-00154-8 · Cited by 47
“Since the taking off of deep learning before 2012, deep learning architectures have rapidly evolved, driving significant advancements in deepfake research (Edwards et al. 2024).”
Explainable deepfake detection across different modalities: An overview of methods and challenges
Momin, M. S., Sufian, A., Barman, D., Leo, M., Distante, C., & Damer, N. (2025) · Image and Vision Computing · doi:10.1016/j.imavis.2025.105738 · Cited by 37
A Comparative Analysis of Compression and Transfer Learning Techniques in DeepFake Detection Models
Karathanasis, A., Violos, J., & Kompatsiaris, I. Y. (2025) · Mathematics · doi:10.3390/math13050887 · Cited by 19
Ha, J., El Azzaoui, A., & Park, J. H. (2025) · Sensors · doi:10.3390/s25030788 · Cited by 13
“This approach not only preserves user privacy but also reduces the risks associated with data breaches and unauthorized access, aligning with the principles of modern data protection regulations [29].”
Bharati, N., Wong, P., Mostéfaoui, S. K., Ismail, D. K. B., & Collie, J. (2025) · Machine Learning with Applications · doi:10.1016/j.mlwa.2025.100819 · Cited by 15
Metadata
- Authors
- Edwards, P., Nebel, J.-C., Greenhill, D., & Liang, X.
- Affiliations
- Kingston University London (GB)
- Venue
- IEEE Access
- Published
- 2024
- Volume
- 12
- Pages
- 154718-154742
- ISSN
- 2169-3536 (ISSN-L)
- OpenAlex
- W4403277964
- Cited by
- 50
- References
- 93
- Citations by year
- 2026: 24 · 2025: 25 · 2024: 1
- FWCI
- 7.00
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- IEEE Access (publishedVersion, cc-by)
- Research Repository (Kingston University London) (acceptedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- 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, deepfake generation, data diversity, DALL-E 3, Stable Diffusion, generative AI
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.