The DOI matches its Crossref record; 21 of the 21 references it cites were found in a registry or an index.
Detecting and Simulating Artifacts in GAN Fake Images
Open accessAbstract
To detect GAN generated images, conventional supervised machine learning algorithms require collecting a large number of real images as well as fake images generated by the targeted GAN model. However, the specific model used by the attacker is often unavailable. To address this, we propose a GAN simulator, AutoGAN, which can simulate the artifacts produced by the common pipeline shared by several popular GAN models. Additionally, we identify a unique artifact caused by the up-sampling component included in the common GAN pipelines. We show theoretically such artifacts are manifested as replications of spectra in the frequency domain and thus propose a classifier model based on the spectrum input, rather than the pixel input. By using the simulated images to train a spectrum based classifier, even without seeing the fake images produced by the targeted GAN model during training, our approach achieves state-of-the-art performances on detecting fake images generated by popular GAN models such as CycleGAN.
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
“Figure 7 represents the potentials of employing the frequency artifacts for detecting DeepFakes, where the GAN-based manipulation introduces invisible artifacts in the frequency domain.”
From Countering Malicious DeepFakes: Survey, Battleground, and Horizon (2022) · cites this paper as Zhang et al. 2019
In the citing paper
“Due to limited data, network structure,8 and optimization algorithms,6 the GANs can hardly fit the distribution of real data.”
From Deepfakes: A new threat to image fabrication in scientific publications? (2022) · cites this paper as [8]
In the citing paper
“The most well-known artifact is the checkerboard effect due to the image upsampling operation in GAN.8”
From Deepfakes: A new threat to image fabrication in scientific publications? (2022) · cites this paper as [8]
Citations
Cites
Unpaired image-to-image translation using cycle-consistent adversarial networks
Zhu, J.-Y., Park, T., Isola, P., & Efros, A. A. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.244 · Cited by 23,602
Image-to-Image Translation with Conditional Adversarial Networks
Isola, P., Zhu, J.-Y., Zhou, T., & Efros, A. A. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.632 · Cited by 22,930 · cited 1×
“image that is constrained by the layout of the sketch [11], [6], [10].”
BackgroundLabaca-Castro, R. (2023) · Springer eBooks · doi:10.1007/978-3-658-40442-0_9 · Cited by 21,119
GAN（Generative Adversarial Nets）
淳司, 柴. (2017) · Journal of Japan Society for Fuzzy Theory and Intelligent Informatics · doi:10.3156/jsoft.29.5_177_2 · Cited by 20,507
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×
“a deconvolution) [19], [6], [7] and nearest neighbor interpolation [9], [10].”
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 2×
“In image generation, GAN [1] can be applied in the following scenarios: 1) taking noise as input to produce an image [9], [2], [3]; 2) taking an image from one semantic category (such as horse) as input to produce an image of another semantic category (such as zebra) [6], [7], [10]; 3) taking a sketch or a pixel level…”
BackgroundMethodProgressive Growing of GANs for Improved Quality, Stability, and Variation
Karras, T., Aila, T., Laine, S., & Lehtinen, J. (2017) · arXiv · doi:10.48550/arxiv.1710.10196 · Cited by 4,727 · cited 4×
“) [6] and nearest neighbor interpolation (NN) [9].”
BackgroundMethodInfluentialStarGAN: Unified Generative Adversarial Networks for Multi-domain Image-to-Image Translation
Choi, Y., Choi, M., Kim, M., Ha, J.-W., Kim, S., & Choo, J. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00916 · Cited by 4,163
Large Scale GAN Training for High Fidelity Natural Image Synthesis
Brock, A., Donahue, J., & Simonyan, K. (2018) · arXiv · doi:10.48550/arxiv.1809.11096 · Cited by 3,717 · cited 2×
“In image generation, GAN [1] can be applied in the following scenarios: 1) taking noise as input to produce an image [9], [2], [3]; 2) taking an image from one semantic category (such as horse) as input to produce an image of another semantic category (such as zebra) [6], [7], [10]; 3) taking a sketch or a pixel level…”
BackgroundMethodSemantic Image Synthesis With Spatially-Adaptive Normalization
Park, T., Liu, M.-Y., Wang, T.-C., & Zhu, J.-Y. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00244 · Cited by 2,974 · cited 5×
“However, all classifiers fail on GauGAN images, since the generator structure of GauGAN (5 nearest neighbor up-samplers) is drastically different from the CycleGAN structure.”
BackgroundMethodInfluential
Cited by
Sorted by citations
Items 301–400 of 561
Learning Expressive And Generalizable Motion Features For Face Forgery Detection
Zhang, J., Zhang, P., Wang, J., Xie, D., & Pu, S. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10095913 · Cited by 6
“Despite the expressive backbones, [6, 7, 8, 9] design some texture and frequency aware architectures for better forgery clue detection.”
BackgroundDeepfakes – Reality Under Threat?
Rana, M. S., Solaiman, M., Gudla, C., & Sohan, M. F. (2024) · IEEE Annual Computing and Communication Workshop and Conference (CCWC) · doi:10.1109/ccwc60891.2024.10427659 · Cited by 8
“Zhang [17] introduced a GAN simulator that replicates image artifacts generated by GANs and uses a classifier for Deepfake detection.”
MethodA Simple and Effective Way to Detect DeepFakes: Using 2D and 3D CNN
Suratkar, S. S., & Sharma, P. (2022) · Lecture notes in electrical engineering · doi:10.1007/978-981-16-9089-1_19 · Cited by 6
Wei, R., Liu, Y., Chen, X., Zhu, J., Yang, B., Men, K., & Dai, J. (2023) · Medical Physics · doi:10.1002/mp.16322 · Cited by 6
Data Redaction from Pre-trained GANs
Kong, Z., & Chaudhuri, K. (2023) · IEEE Conference on Secure and Trustworthy Machine Learning (SaTML) · doi:10.1109/satml54575.2023.00048 · Cited by 9
“…image samples [Kaneko and Harada, 2021], fairness issues [Tan et al., 2020, Karakas et al., 2022], and checkerboard artifacts [Odena et al., 2016, Zhang et al., 2019, Wang et al., 2020, Schwarz et al., 2021] in image generation, offensive text in language models [Abid et al., 2021, Perez et al.,…”
BackgroundZhang, J., Wang, Y., Tohidypour, H. R., & Nasiopoulos, P. (2023) · Proceedings - International Conference on Image Processing · doi:10.1109/icip49359.2023.10221905 · Cited by 7
“Note that the artifacts introduced by GANs are only visible when taking the summation/average of multiple frequency spectra of their synthesized images [13].”
BackgroundA Dynamic Ensemble Selection of Deepfake Detectors Specialized for Individual Face Parts
Kawabe, A., Haga, R., Tomioka, Y., Shin, J., & Okuyama, Y. (2023) · Electronics · doi:10.3390/electronics12183932 · Cited by 6
Generalized Fake Image Detection Method Based on Gated Hierarchical Multi-Task Learning
Zhou, Y., He, P., Li, W., Cao, Y., & Jiang, X. (2023) · IEEE Signal Processing Letters · doi:10.1109/lsp.2023.3336570 · Cited by 6
“As claimed in previous works, the construction of generators by stacking network modules with similar architectures can cause abnormal correlations between local patches, such as periodic artifact [12].”
BackgroundDT-TransUNet: A Dual-Task Model for Deepfake Detection and Segmentation
Zheng, J., Zhou, Y., Hu, X., & Tang, Z. (2023) · Lecture notes in computer science · doi:10.1007/978-981-99-8540-1_20 · Cited by 6
Detecting facial manipulated images via one-class domain generalization
Xu, P., Ma, Z., Mei, X., & Shen, J. (2024) · Multimedia Systems · doi:10.1007/s00530-023-01214-7 · Cited by 6
“Some recent works discovered that the images manipulated by GANs exist anomalous patterns in the frequency domain [7, 8, 37].”
Automatic Correction of Internal Units in Generative Neural Networks
Tousi, A., Jeong, H., Han, J., Choi, H., & Choi, J. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.00784 · Cited by 6
“Defective regions can be observed in synthesized images by deep neural networks [19, 24, 3, 22].”
Analyzing the Methods for Detecting Deepfakes
Khichi, M., & Yadav, R. K. (2021) · International Conference on Advances in Computing, Communication Control and Networking (ICAC3N) · doi:10.1109/icac3n53548.2021.9725773 · Cited by 6
“Durall et al. [30] proposed a framework for detecting Deepfakes based on frequency domain analysis as well.”
BackgroundBlack-Box Watermarking for Generative Adversarial Networks.Preprint
Skripniuk, V., Yu, N., Abdelnabi, S., & Fritz, M. (2020) · arXiv (Cornell University) · Cited by 6
Making GAN-Generated Images Difficult To Spot: A New Attack Against Synthetic Image Detectors.Preprint
Zhao, X., & Stamm, M. C. (2021) · arXiv (Cornell University) · Cited by 6
Beyond the Spectrum: Detecting Deepfakes via Re-SynthesisPreprint
He, Y., Yu, N., Keuper, M., & Fritz, M. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.14376 · Cited by 6
Vrettos, K., Vassalou, E. E., Vamvakerou, G., Karantanas, A. H., & Klontzas, M. E. (2025) · Academic Radiology · doi:10.1016/j.acra.2025.03.015 · Cited by 5
DC-BiNet: Towards interpretable generated image detection with dark channel prior
Tan, D., Niu, C., Yang, Y., Yang, D., & Tan, B. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.127508 · Cited by 6
Tran, V.-N., Choi, P., Le, H.-S., Lee, S., & Kwon, K. (2025) · IEEE Open Journal of the Computer Society · doi:10.1109/ojcs.2025.3575507 · Cited by 6
Towards Explainable Fake Image Detection with Multi-Modal Large Language Models
Ji, Y., Hong, Y., Zhan, J., Chen, H., Lan, J., Zhu, H., Wang, W., Zhang, L., & Zhang, J. (2025) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3746027.3755421 · Cited by 5
Hirofumi, S., Fukuchi, K., Akimoto, Y., & Sakuma, J. (2022) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn55064.2022.9892704 · Cited by 5
“[10] is the first learning-based GAN detection method.”
BackgroundMethodSynthetic Face Discrimination via Learned Image Compression
Iliopoulou, S., Tsinganos, P., Ampeliotis, D., & Skodras, A. N. (2024) · Algorithms · doi:10.3390/a17090375 · Cited by 5
“Zhang et al. [18] identified fake images by detecting the spectral peaks that show up because of the upsampling that takes place in many GAN architectures.”
BackgroundAin, Q. U., Javed, A., Malik, K. M., & Irtaza, A. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-20268-x · Cited by 7
Explainability-driven adversarial robustness assessment for generalized deepfake detectors
Cirillo, L., Gervasio, A., & Amerini, I. (2025) · EURASIP Journal on Information Security · doi:10.1186/s13635-025-00211-9 · Cited by 7
Pattern detection in the activation space for identifying synthesized content
Cintas, C., Speakman, S., Tadesse, G. A., Akinwande, V., McFowland, E., & Weldemariam, K. (2021) · Pattern Recognition Letters · doi:10.1016/j.patrec.2021.12.007 · Cited by 8
A GAN Based Malware Adversaries Detection Model
Umer, M., Saleem, Y., Saleem, M., & Aman, N. (2021) · International Conference on Open Source Systems and Technologies (ICOSST) · doi:10.1109/icosst53930.2021.9683863 · Cited by 6
GAN Fingerprints in Face Image Synthesis
Neves, J. C., Tolosana, R., Vera-Rodríguez, R., Lopes, V., Proença, H. P. M. C., & Fiérrez, J. (2022) · Advances in computer vision and pattern recognition · doi:10.1007/978-981-16-7621-5_8 · Cited by 5
"Is this my president speaking?" Tamper-proofing Speech in Live Recordings
Shahid, I., & Roy, N. (2023) · International Conference on Mobile Systems, Applications, and Services · doi:10.1145/3581791.3596862 · Cited by 5
“A class of techniques focus on the explicit technical inconsistencies generated during editing operations like warping [42], blending [39], and splicing [33] or signatures left by the commonly used learning models [65, 68, 71, 72].”
BackgroundThe Dual Role of Artificial Intelligence in Online Disinformation: A Critical Analysis
Montasari, R. (2024) · Advanced sciences and technologies for security applications · doi:10.1007/978-3-031-50454-9_11 · Cited by 5
Deepfake Videos Generation and Detection: A Comprehensive Survey
Gupta, A., & Pandey, D. (2024) · IEEE International Conference on Computing, Power and Communication Technologies (IC2PCT) · doi:10.1109/ic2pct60090.2024.10486232 · Cited by 5
“Use of AutoGAN for detecting presence of deepfakes was also given by researchers [33].”
MethodLeveraging Frequency Analysis for Image Denoising Network Pruning
Ren, D., Li, W., Huo, J., Wang, L., Pan, H., & Gao, Y. (2025) · IEEE Transactions on Image Processing · doi:10.1109/tip.2025.3544108 · Cited by 4
“Recently, several pieces of work [22], [23], [24] have emerged in the field of deep learning to utilize frequency analysis to explicitly model features in the frequency domain.”
GLFAFormer: DeepFake forgery detection with adaptive feature extract and align
Chen, L., Zhao, H., He, J., & Meng, W. (2025) · Digital Signal Processing · doi:10.1016/j.dsp.2025.105321 · Cited by 5
X-Edit: Detecting and Localizing Edits in Images Altered by Text-Guided Diffusion Models
Bazyleva, V., Bonettini, N., & Bharaj, G. (2025) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw67362.2025.00257 · Cited by 4
“Early detection efforts for AI-generated content focused on GANs, utilizing frequency domain analysis [34, 62] and deep learning classifiers [53, 54] for the detection.”
Robust AI-Synthesized Image Detection via Multi-feature Frequency-Aware Learning
Cai, H., Liu, C., Shen, S., Qu, Y., & Gui, P. (2025) · Lecture notes in computer science · doi:10.1007/978-981-95-3001-4_12 · Cited by 5
Frequency Bias Matters: Diving Into Robust and Generalized Deep Image Forgery Detection
Liu, C., Zhu, T., Zhou, W., & Zhao, W. (2025) · IEEE Transactions on Dependable and Secure Computing · doi:10.1109/tdsc.2025.3639022 · Cited by 6
“Current forensic studies generally focus on designing sophisticated detector networks or feature engineering methods to enhance these two properties [6], [7], [8], [9], [10], [11], [12], [13].”
LOTA: Bit-Planes Guided AI-Generated Image Detection
Wang, H. M., Cheng, R., Zhang, Y., Han, C., & Gui, J. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.01602 · Cited by 4
“The seminal work of [40] identified GAN-specific frequency artifacts.”
Robust watermarking for diffusion models based on STDM and latent space fine-tuning
Li, L., Zhang, X., Feng, G., Wang, Z., Wu, D., & Wu, H. (2025) · Journal of Information Security and Applications · doi:10.1016/j.jisa.2025.104167 · Cited by 4
Self-citationHolistic Image Manipulation Detection using Pixel Cooccurrence Matrices
Nataraj, L., Goebel, M. G., Mohammed, T. M., Chandrasekaran, S., & Manjunath, B. S. (2021) · Electronic Imaging · doi:10.2352/issn.2470-1173.2021.4.mwsf-277 · Cited by 4
Survey on Vision based Fake News Detection and its Impact Analysis
Raval, M. S., Roy, M., & Kuribayashi, M. (2022) · Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) · doi:10.23919/apsipaasc55919.2022.9980089 · Cited by 5
“Zhang et al. introduced a GAN simulator to create a copy of GAN-image artifacts [23], [24].”
BackgroundFake Satellite Image Detection via Parallel Subspace Learning (PSL)
Chen, H.-S., Zhang, K., Hu, S., You, S. O., & Kuo, C. ‐. J. (2022) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas48785.2022.9937539 · Cited by 4
“Zhang et al. [22] trained a convolutional neural network (CNN) with simulated fake images.”
BackgroundGAN-Based One-Class Classification SVM for Real time Medical Image Intrusion Detection
Hussain, F., Tbarki, K., & Ksantini, R. (2023) · International Journal of Computing and Digital Systems · doi:10.12785/ijcds/130150 · Cited by 4
A novel blind tamper detection and localization scheme for multiple faces in digital images
Thabit, R. (2023) · IET Image Processing · doi:10.1049/ipr2.12909 · Cited by 4
Beyond Human Forgeries: An Investigation into Detecting Diffusion-Generated Handwriting
Carrière, G., Nikolaidou, K., Kordon, F., Mayr, M., Seuret, M., & Christlein, V. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-41498-5_1 · Cited by 5
Lee, S.-J., Jung, S., & Seo, H. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i4.28074 · Cited by 9
“First, an upsampling operation in the most generative network can have aliasing on the spectrum ( i . e ., checkerboard artifact in image domain) and insufficient high-frequency components (Zhang, Karaman, and Chang 2019; Frank et al. 2020).”
BackgroundMethodLu, J., Zhou, J., Dong, J., Li, B., Lyu, S., & Li, Y. (2024) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2024.3395013 · Cited by 4
“For better 100.0 100.0 98.4 99.9 89.9 96.1 64-8 100.0 100.0 98.8 99.9 89.3 95.7 128-8 100.0 100.0 98.6 99.9 87.9 96.0 understanding, for example, we use “ a − b ” to represent adding two FC layers and these layers have a, b parameters respectively.”
BackgroundInfluentialAzizpour, A., Nguyen, T. D., Shrestha, M., Xu, K., Kim, E., & Stamm, M. C. (2024) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw63382.2024.00437 · Cited by 5
“To address the rapid release of image generators, a broad spectrum of new synthetic image detectors were developed [13, 25, 39, 56, 58, 64, 65, 68].”
BackgroundRAW Data: A Key Component for Effective Deepfake Detection
Husseinil, S., & Dugelay, J. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10887800 · Cited by 4
Wang, S., Zhou, F., Zhang, W., Wang, X., & Zhang, L. (2025) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2025.3556819 · Cited by 4
“However, Wang et al. [36], Zhang et al. [37], and Huang et al. [38] have demonstrated that neural networks exhibit Spectral Bias in reconstruction and synthesis tasks: they tend to focus on learning low-frequency components, which are easier to synthesize.”
Blockwise Spectral Analysis for Deepfake Detection in High-fidelity Videos
Huang, H., Sun, N., & Lin, X. (2022) · IEEE International Conference on Data Science and Advanced Analytics (DSAA) · doi:10.1109/dsaa54385.2022.10032370 · Cited by 5
Multi-Channel Cross Modal Detection of Synthetic Face Images
Ibsen, M., Rathgeb, C., Marcel, S., & Busch, C. (2024) · International Workshop on Biometrics and Forensics (IWBF) · doi:10.1109/iwbf62628.2024.10593873 · Cited by 4
“Frequency Domain Transformation To transform an RGB image into the frequency spectrum we follow the approach in [26] and normalize the input pixels to the range [0, 1] whereafter we perform DFT for each of the 3 colour channels and obtain corresponding channels in the frequency spectrum.”
BackgroundMethodAdversarial Attacks on Co-Occurrence Features for GAN DetectionPreprint
Goebel, M. G., & Manjunath, B. S. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2009.07456 · Cited by 5
“these gray-box methods was proposed in [30].”
BackgroundMethodTanaka, M., Shiota, S., & Kiya, H. (2021) · IEEE Global Conference on Consumer Electronics (GCCE) · doi:10.1109/gcce53005.2021.9621809 · Cited by 3
Robust manipulated media localization and detection based on high frequency and texture features
Jiang, Q., Liu, S., Miao, S., Yi, H., Jin, X., Kou, Y., & Duan, H. (2025) · Discover Computing · doi:10.1007/s10791-025-09500-w · Cited by 4
“Furthermore, techniques [26, 27] employ Fast Fourier Transform (FFT) or Discrete Fourier Transform (DFT) to derive the frequency spectrum, subsequently utilizing these frequency-based features for classification, thereby enhancing detection capabilities.”
Evaluating Features and Variations in Deepfake Videos Using the CoAtNet Model
Alattas, E., Clark, J., Al-Aama, A. Y., & Jarraya, S. K. (2025) · Journal of Imaging · doi:10.3390/jimaging11060194 · Cited by 3
“Moreover, it poses an issue for security and ethics, as visual media can no longer be considered trustworthy content [25].”
RADAR: Reasoning AI-Generated Image Detection for Semantic Fakes
Wang, H., Liu, X., Lu, Z., Yan, C., Jiang, X., Wang, R., & Gavves, E. (2025) · Technologies · doi:10.3390/technologies13070280 · Cited by 3
DeepFeatureX-SN: Generalization of deepfake detection via contrastive learning
Pontorno, O., Guarnera, L., & Battiato, S. (2025) · Multimedia Tools and Applications · doi:10.1007/s11042-025-21060-1 · Cited by 5
“Current deepfake detection methods focus largely on distinguishing real from AI-generated content [12, 13], with some success in identifying specific GAN architectures [14, 15].”
Boosting Geometric Invariants for Discriminative Forensics of Large-Scale Generated Visual Content
Qi, S., Wang, C., Huang, Z., Zhang, Y., Chen, X., Zhang, Y., Zeng, T., & Fan, F. (2025) · IEEE Transactions on Image Processing · doi:10.1109/tip.2025.3633580 · Cited by 3
“Regarding the competitors at forensic level, we cover a series of major developments in forensics for large-scale generated images – CNNSpot [48], Spec [49], F3Net [50], GramNet [51], DIRE [52], Universal [53], and GenDet [54].”
ForgeLens: Data-Efficient Forgery Focus for Generalizable Forgery Image Detection
Chen, Y., Zhang, L., & Niu, Y. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.01510 · Cited by 13
A Universal Detection Method for Adversarial Examples and Fake Images
Lai, J., Huo, Y., Hou, R., & Wang, X. (2022) · Sensors · doi:10.3390/s22093445 · Cited by 4
“Limited by the two time-scale update rule (TTUR), we only use AutoGAN [41] and TransGAN [42] generators on the CIFAR10 dataset.”
Cozzolino, D., Nagano, K., Thomaz, L. A., Majumdar, A., & Verdoliva, L. (2023) · IEEE Signal Processing Magazine · doi:10.1109/msp.2023.3294720 · Cited by 7
IPD-Net: Detecting AI-Generated Images via Inter-Patch Dependencies
Chen, J., Lo, M., Liao, H., & Huang, T. S. (2024) · International Journal of Advanced Computer Science and Applications · doi:10.14569/ijacsa.2024.01507133 · Cited by 4
“However, with the continuous evolution of the field of the generative model, some early methods fail to generalize well to new models [6], [8], [9], [10].”
BackgroundMethodA Comprehensive Exploration on Detecting Fake Images Generated by Stable Diffusion
Chen, J., Wang, X., He, Z., & Peng, X. (2024) · Lecture notes in computer science · doi:10.1007/978-981-97-8487-5_32 · Cited by 5
Review of Recent Trends in Face Image Authentication (FIA) Techniques and Their Limitations
Jawad, A. H., Thabit, R., & Zidan, K. A. (2024) · Al-Iraqia Journal of Scientific Engineering Research · doi:10.58564/ijser.3.3.2024.236 · Cited by 2
Explainable Artifacts for Synthetic Western Blot Source Attribution
Cardenuto, J. P., Mandelli, S., Moreira, D., Bestagini, P., Delp, E. J., & Rocha, A. D. R. (2024) · IEEE International Workshop on Information Forensics and Security · doi:10.1109/wifs61860.2024.10810680 · Cited by 4
“Therefore, this operation alone cannot eliminate the periodic artifacts, which could still be detected in the Fourier spectrum [17].”
GANFR: GAN fingerprint removal network for image anti-forensics
Lu, Y., Liu, J., & Zhang, R. (2025) · Knowledge-Based Systems · doi:10.1016/j.knosys.2025.114134 · Cited by 5
Forensic Challenges in Face Manipulated Videos
Sharma, N., Mallampalli, K., & Iyengar, R. S. (2025) · International Conference on Multimedia Analysis and Pattern Recognition (MAPR) · doi:10.1109/mapr67746.2025.11134019 · Cited by 3
“A GAN simulator was introduced in [24] to replicate such artifacts for classifier training.”
Fahn, C., & Wu, T. (2022) · International Conference on Machine Learning and Cybernetics · doi:10.1109/icmlc56445.2022.9941295 · Cited by 3
“Therefore, when restoring the image of the original resolution, there will exist checkerboard artifacts [13].”
BackgroundMethodTotems: Physical Objects for Verifying Visual Integrity
Ma, J., Chai, L., Huh, M., Wang, T., Lim, S.-N., Isola, P., & Torralba, A. B. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19781-9_10 · Cited by 3
SE-ResNet56: Robust Network Model for Deepfake Detection
Wang, X., Zhao, Z., Zhang, C., Bai, N., & Hu, X. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-25115-3_3 · Cited by 3
An Unsupervised Framework for Joint MRI Super Resolution and Gibbs Artifact Removal
Liu, Y., Chen, E. Z., Chen, X., Chen, T., & Sun, S. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-34048-2_31 · Cited by 3
Spatial-Temporal Variables for Deepfake Video Identification
Ralhen, Y., & Sharma, S. C. (2023) · International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE) · doi:10.1109/icacite57410.2023.10182819 · Cited by 3
Forensics Forest: Multi-scale Hierarchical Cascade Forest for Detecting GAN-generated Faces
Lu, J., Li, Y., Zhou, J., Li, B., & Lyu, S. (2023) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme55011.2023.00394 · Cited by 3
“, upsampling artifacts [19], [20]), frequency signals (e.”
BackgroundAI-Generated Image Detection With Wasserstein Distance Compression and Dynamic Aggregation
Lyu, Z., Xiao, J., Zhang, C., & Lam, K. (2024) · Proceedings - International Conference on Image Processing · doi:10.1109/icip51287.2024.10648186 · Cited by 4
DeepFake Detection Using Fine-Tuned CNN Architectures
Çetintaş, D., & Yücel, Z. (2025) · Celal Bayar Üniversitesi Fen Bilimleri Dergisi · doi:10.18466/cbayarfbe.1530209 · Cited by 3
Towards Generalized Deepfake Detection With Continual Learning On Limited New Data
Huang, H., Sun, N., Lin, X., & Moustafa, N. (2022) · International Conference on Digital Image Computing: Techniques and Applications (DICTA) · doi:10.1109/dicta56598.2022.10034569 · Cited by 5
“In this work, we use the GAN simulator to generate fake horse images from real horse images to train the model, and test the model with other categories generated by CycleGAN that has the same generative network architecture with GAN simulator in this dataset [3].”
BackgroundMethodInfluential[RE] CNN-generated images are surprisingly easy to spot...for nowPreprint
Frank, J., & Holz, T. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.02984 · Cited by 3
DeepFake Detection with Inconsistent Head Poses: Reproducibility and AnalysisPreprint
Lutz, K., & Bassett, R. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2108.12715 · Cited by 13
“Along similar lines, [36] uses a discrete Fourier transform to detect artifacts of the upsampling procedure used by many GANbased image generators.”
Li, Z., Sun, J., & Wang, S. (2025) · Symmetry · doi:10.3390/sym17040536 · Cited by 2
He, Y., Cao, Y., Yang, B., & Zhang, Z. (2025) · ACM Workshop on Information Hiding and Multimedia Security (IH&MMSec) · doi:10.1145/3733102.3733138 · Cited by 4
“Frequency Feature Learning utilizes frequency-domain analysis to detect artifacts from generative models [7–10, 12, 15, 32, 45], employing features like spectrum magnitude and 2D-FFT [10, 12, 32, 45].”
BOSC: A Backdoor-Based Framework for Open Set Synthetic Image Attribution
Wang, J., Tondi, B., & Barni, M. (2025) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2025.3592531 · Cited by 6
“Several methods have been proposed which rely on artifacts or signatures (fingerprints) left by the models in the images they generate [11]–[13].”
Wavelet-based GAN Fingerprint Detection using ResNet50
Erukude, S. T., Veluru, S. R., & Marella, V. C. (2025) · International Conference on Innovative Mechanisms for Industry Applications · doi:10.1109/icimia67127.2025.11200674 · Cited by 2
“Studies suggest that frequency analysis can amplify subtle artifacts that are less visible in the pixel domain, thereby aiding detection [2], [5]–[7].”
Cui, J., Du, J., Li, Y., Gao, L., Jiang, H., & Bao, C. (2025) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3746027.3755137 · Cited by 2
“Specifically, the HAMLET module operates as an external plugin that accesses intermediate representations from CLIP’s vision trans-former and generates vision-conditioned text prompt embeddings for image-adaptive authenticity assessment in the shared CLIP embedding space.”
Kumar, V., & Purohit, A. (2026) · Journal of Future Artificial Intelligence and Technologies · doi:10.62411/faith.3048-3719-291 · Cited by 2
Optimal transport-guided framework for adversarial robust deepfake detection
Chatterjee, S. (2026) · Discover Computing · doi:10.1007/s10791-026-09904-2 · Cited by 2
“Studies have shown that models trained on a specific dataset, such as FaceForensics++, tend to perform poorly when evaluated on others like Celeb-DF or Deep-fakeTIMIT due to differences in synthesis techniques and content diversity [18, 19].”
Shi, Y., Weng, S., Yu, L., & Ni, R. (2026) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2026.3656696 · Cited by 2
“Prior works [34], [35], [36] have reported the wide availability of upsampling operations (e.g., nearest neighbor or bilinear upsampling) in common GAN architectures.”
Yang, G. (2025) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw69036.2025.00685 · Cited by 2
“Frequency domain analysis has proven particularly effective for AI-generated image detection due to the fundamental differences in how generative models and natural imaging processes distribute energy across frequency bands [6, 17].”
Ultra‐Wide‐Field Noninvasive Imaging Through Scattering Media Via Physics‐Guided Deep Learning
Peng, L., He, M., Zhu, J., Sahoo, S. K., Bian, L., & Dang, C. (2026) · Advanced Science · doi:10.1002/advs.75390 · Cited by 2
Flat and Shallow: Understanding Fake Image Detection Models by Architecture Profiling
Xu, J., Zhang, W., Bai, Y., Sun, Q., & Mei, T. (2021) · ACM International Conference on Multimedia in Asia (MMAsia) · doi:10.1145/3469877.3490566 · Cited by 2
“Unlike prior works that seek patterns on generated fake contents [14, 23], we take a different perspective by directly analyzing the model structures.”
BackgroundIn-group and Out-group Performance Bias in Facial Retouching Detection
Bharati, A., Connors, E., Mayank Vatsa, A., Singh, R., & Bowyer, K. W. (2022) · IEEE International Joint Conference on Biometrics (IJCB) · doi:10.1109/ijcb54206.2022.10007942 · Cited by 3
“As retouched examples are often a slightly modified version of the original, they may not have the same artifacts or cues that drive detection of fully synthesized generated images, such as asymmetry in eye color, shape and artifacts around strong boundaries [65, 68].”
BackgroundDeep Fake Image Detection Based on Decoupled Dynamic Convolution
Li, J., & Li, B. (2022) · IEEE International Conference on Frontiers Technology of Information and Computer (ICFTIC) · doi:10.1109/icftic57696.2022.10075251 · Cited by 2
“In order to cope with some unknown attacks, Zhang et al. [8] used the spectral input of images to train the classifier.”
BackgroundMethodShufan, 彭. P., Manchun, 蔡. C., Rui, 马. M., & Xiaowen, 刘. L. (2023) · Laser & Optoelectronics Progress · doi:10.3788/lop213318 · Cited by 2
WPCA: Wavelet Packets with Channel Attention for Detecting Face Manipulation
Cao, J., Deng, J., Yin, X., Yan, S., & Li, Z. (2023) · International Conference on Machine Learning and Computing (ICMLC) · doi:10.1145/3587716.3587763 · Cited by 2
An Image Forgery Detection Technology Based on Deep Learning
Zou, Y., Liu, J., Li, Y., Xie, Y., Liu, B., Yi, T., & Li, X. (2024) · International Conference on Algorithms, Software Engineering and Network Security (ASENS) · doi:10.1145/3677182.3677245 · Cited by 2
A Face Forgery Video Detection Model Based on Knowledge Distillation
Liang, H., Leng, Y., Luo, J., Chen, J., & Guo, X. (2024) · ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing (SNPD) · doi:10.1109/snpd61259.2024.10673906 · Cited by 3
“Early deepfake detection methods could often rely on visible artifacts or inconsistencies in the videos, such as unnatural blinking patterns or poor lip-sync [11-15] .”
BackgroundDeepfake Detection Using Deep Learning
Prakash Raj, S., Pravin, D. V., Sabareeswaran, G., Sanjith, R. K., & Gomathi, B. (2024) · International Conference on Advanced Computing and Communication Systems (ICACCS) · doi:10.1109/icaccs60874.2024.10717155 · Cited by 3
“Our exploration entails how these deep learning models have exhibited promising capabilities across multiple domains, including image and text classification, underscoring their potential applicability in the domain of deep fake detection [19].”
Detecting computer-generated images by using only real images
Li, J., & Wang, K. (2025) · International Conference on Machine Vision · doi:10.1117/12.3055092 · Cited by 2
“Zhang et al. [14] proposed a GAN simulator to replicate artifacts generated by the common pipeline shared among different GANs.”
Murugesu, M. P., Krishnan, V., & Kovscek, A. R. (2024) · Applied Computing and Geosciences · doi:10.1016/j.acags.2024.100208 · Cited by 2
Diffusion Models as a Representation Learner for Deepfake Image Detection
Ganguly, R., Bah, M. D., & Dahmane, M. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-78305-0_15 · Cited by 2
Addressing Diffusion Model Based Counter-Forensic Image Manipulation for Synthetic Image Detection
Herur, A. N., Santhosh, V., Shetty, N., & Seelamantula, C. S. (2024) · Indian Conference on Computer Vision, Graphics and Image Processing · doi:10.1145/3702250.3702296 · Cited by 3
“Synthetic image detection algorithms that exploit the presence of frequency artifacts were first introduced to detect synthetic images generated by Generative Adversarial Networks (GANs) as discussed in papers like [35] and [5].”
Face image authentication scheme based on MTCNN and SLT
Thabit, R., Al‐Askari, M. A., Mohammed, D. Z., Anaam, E. A., Mahmood, Z. H., Jabbar, D. J., & Salih, Z. A. (2025) · Multimedia Tools and Applications · doi:10.1007/s11042-025-20684-7 · Cited by 3
“…v a il - a b l e f o r fr ee a nd i n m o s t ca s e s , it r e qu i r e s p e r m i ss i on s fr o m t h e o w n e r s o f t h e i m a g e s [ 41 –48 ] N o t r a i n i ng i s r e qu i r e d t h e r e f o r e a ny i m a g e ca n b e u s e d i n t h e p r opo s e d s c h e m e 7 T h e a l go r it…”
InfluentialAI-Generated Image Detection Using Semantic Feature
Chen, H.-C., Shen, Y.-P., Yeh, Y.-H., & Chen, J.-L. (2025) · International Conference on Advanced Communications Technology (ICACT) · doi:10.23919/icact63878.2025.10936792 · Cited by 2
Metadata
- Authors
- Zhang, X., Karaman, S., & Chang, S.-F.
- Affiliations
- Columbia University (US)
- Venue
- 2019 IEEE International Workshop on Information Forensics and Security (WIFS)
- Publisher
- IEEE
- Published
- 2019-12
- Pages
- 1-6
- ISSN
- 2157-4774 (ISSN-L)
- OpenAlex
- W3012472557
- Cited by
- 567
- References
- 21
- Citations by year
- 2026: 60 · 2025: 113 · 2024: 89 · 2023: 100 · 2022: 84 · 2021: 81 · 2020: 39
- FWCI
- 16.52
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE International Workshop on Information Forensics and Security (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Digital Media Forensic Detection
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Generative Adversarial Networks and Image Synthesis
- Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
- Keywords
- GAN-generated image detection, CycleGAN, adversarial image generation
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
