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 101–200 of 561
Performance Comparison and Visualization of AI-Generated-Image Detection Methods
Park, D., Na, H., & Choi, D. (2024) · IEEE Access · doi:10.1109/access.2024.3394250 · Cited by 53
“A study on the no-down method [39] tested its detection performance on images generated by five GAN models, four diffusion models, and three transformer models and compared it with those of other detection methods [37], [38], [40].”
MethodResultInteractive Two-Stream Network Across Modalities for Deepfake Detection
Wu, J., Zhang, B., Li, Z., Pang, G., Teng, Z., & Jianping Fan, B. (2023) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2023.3269841 · Cited by 37
“For example, Zhang et al. [34] extract frequency features from the residuals generated by the median filter on the image to detect forgery faces.”
BackgroundLeveraging Representations from Intermediate Encoder-Blocks for Synthetic Image Detection
Koutlis, C., & Papadopoulos, S. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73220-1_23 · Cited by 59
Watching the BiG artifacts: Exposing DeepFake videos via Bi-granularity artifacts
Chen, H., Li, Y., Lin, D., Li, B., & Wu, J. (2022) · Pattern Recognition · doi:10.1016/j.patcog.2022.109179 · Cited by 43
Detection of AI-Generated Synthetic Faces
Gragnaniello, D., Marra, F., & Verdoliva, L. (2022) · Advances in computer vision and pattern recognition · doi:10.1007/978-3-030-87664-7_9 · Cited by 40
“In this analysis, the following synthetic image detectors are considered: Xception [35], SRNet [5], Spec [61], M-Gb [56], Co-Net [42], Wang2020 [54], PatchForensics [8].”
BackgroundMethodInfluentialDetection of AI-Manipulated Fake Faces via Mining Generalized Features
Yu, Y., Ni, R., Li, W., & Zhao, Y. (2022) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3499026 · Cited by 54
“From the results, the performance of Reference [61] and Reference [15] drop significantly when the train/test manipulation techniques use different up-sampling methods.”
BackgroundMethodResultInfluentialRepMix: Representation Mixing for Robust Attribution of Synthesized Images
Bui, T., Yu, N., & Collomosse, J. P. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19781-9_9 · Cited by 41
Hierarchical Forgery Classifier on Multi-Modality Face Forgery Clues
Liu, D., Zheng, Z., Peng, C., Yukai, W. A. N. G., Wang, N., & Gao, X. (2023) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2023.3304913 · Cited by 40
“Zhang et al. [45] showed theoretically that checkboard artifacts are manifested in the DFT spectrum.”
BackgroundDetecting Images Generated by Deep Diffusion Models using their Local Intrinsic Dimensionality
Lorenz, P., Durall, R., & Keuper, J. (2023) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw60793.2023.00051 · Cited by 47
“For example, frequency-based approaches [99, 23, 68] have shown high detection scores when applied to images generated by generative adversarial networks (GANs), but they fail when DMs are employed.”
BackgroundBeyond Deepfake Images: Detecting AI-Generated Videos
Vahdati, D. S., Nguyen, T. D., Azizpour, A., & Stamm, M. C. (2024) · 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw63382.2024.00443 · Cited by 37
“For example, prior work by Zhang et al. has shown that the upsampling operation used in many generator architectures to grow an image from a small latent representation to a full sized image leaves behind traces similar to those left by resampling [94].”
BackgroundMethodInfluentialDeepFakes: Detecting Forged and Synthetic Media Content Using Machine Learning
Zobaed, S., Rabby, F., Hossain, M. I., Hossain, E., Hasan, S., Karim, A., & Hasib, K. M. (2021) · Advanced sciences and technologies for security applications · doi:10.1007/978-3-030-88040-8_7 · Cited by 39
Quality-Agnostic Deepfake Detection with Intra-model Collaborative Learning
Le, B. M., & Woo, S. S. (2023) · 2023 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.02045 · Cited by 36
“Such methods include analyzing the discrepancies of frequency spectrum [10, 31, 29], employing the checkerboard artifacts caused by the transposed convolutional operator [68, 13], or mining the statistical frequency features with dual deep learning models [45].”
MethodFace Manipulation Detection via Auxiliary Supervision
Wang, X., Yao, T., Ding, S., & Ma, L. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-63830-6_27 · Cited by 36
Zheng, C., Bian, F., Li, L., Xie, X., Liu, H., Liang, J., Chen, X., Wang, Z., Qiao, T., Yang, J., et al. (2021) · Translational Vision Science & Technology · doi:10.1167/tvst.10.4.34 · Cited by 36
DeepFake Creation and Detection:A Survey
Swathi, P. S., & Saritha, S. K. (2021) · International Conference on Inventive Research in Computing Applications (ICIRCA) · doi:10.1109/icirca51532.2021.9544522 · Cited by 37
Deepfakes: evolution and trends
Gil, R., Virgili‐Gomà, J., López, J. M., & Garcı́a, R. (2023) · Soft Computing · doi:10.1007/s00500-023-08605-y · Cited by 46
“…et al. (2020), Wang and Dantcheva (2020), Mi et al. (2020), Ranjan et al. (2020), Frank et al. (2020), Yang and Lim (2020), Hashmi et al. (2020), Ranjith Kumar et al. (2020), Xuan et al. (2019), Guan et al. (2019), Kharbat et al. (2019), Bose and Aarabi (2019), Zhang et al. (2019a), Ward (2019)”
A Representative Study on Human Detection of Artificially Generated Media Across Countries
Frank, J., Herbert, F., Ricker, J., Schönherr, L., Eisenhofer, T., Fischer, A., Dürmuth, M., & Holz, T. (2024) · IEEE Symposium on Security and Privacy (S&P) · doi:10.1109/sp54263.2024.00159 · Cited by 66
“These fakes have resulted in a flurry of techniques to automatically detect AI-generated media [9]–[22].”
BackgroundCSC-Net: Cross-Color Spatial Co-Occurrence Matrix Network for Detecting Synthesized Fake Images
Qiao, T., Chen, Y., Zhou, X., Shi, R., Shao, H., Shen, K., & Luo, X. (2023) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2023.3274450 · Cited by 33
“Alternatively, both [35] and [36] study the artifacts of GAN images in the frequential domain.”
BackgroundMultispectral Satellite Image Generation Using StyleGAN3
Alibani, M., Acito, N., & Corsini, G. U. (2024) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2024.3356618 · Cited by 35
“…generative models in the creation of “fake” images acquired by conventional RGB cameras have stimulated the analysis of their impact on real life [11] and the development of methods to evaluate the authenticity of a visual content and discriminate between false and real images [12], [13], [14].”
MethodDeepfake Image Detection Using Vision Transformer Models
Ghita, B., Kuzminykh, I., Usama, A., Bakhshi, T., & Marchang, J. (2024) · IEEE International Black Sea Conference on Communications and Networking · doi:10.1109/blackseacom61746.2024.10646310 · Cited by 38
“As content became more realistic and more effective at eliminating image artifacts, the detection methods also moved towards deep learning-based approach, such as AutoGAN [11], which is essentially a GAN that replicates the deepfake process.”
MethodLanguage-Guided Hierarchical Fine-Grained Image Forgery Detection and Localization
Guo, X. B., Liu, X., Masi, I., & Liu, X. (2024) · International Journal of Computer Vision · doi:10.1007/s11263-024-02255-9 · Cited by 38
“Similar visualization is adopted in [12, 68, 85, 100] also.”
InfluentialMFF-Net: Deepfake Detection Network Based on Multi-Feature Fusion
Zhao, L., Zhang, M., Ding, H., & Cui, X. (2021) · Entropy · doi:10.3390/e23121692 · Cited by 32
“To address this problem, researchers have begun to focus on general forged features [15,16], such as texture and frequency spectra.”
BackgroundTowards Universal GAN Image Detection
Cozzolino, D., Gragnaniello, D., Poggi, G., & Verdoliva, L. (2021) · IEEE International Conference on Visual Communications and Image Processing (VCIP) · doi:10.1109/vcip53242.2021.9675329 · Cited by 35
“Leveraging the opportunity to generate large datasets of GAN images, the most effective techniques work in a supervised setting and achieve excellent results both working in the spatial [11], [12] and in the frequency domain [13], [14].”
Discovering Transferable Forensic Features for CNN-Generated Images Detection
Chandrasegaran, K., Tran, N.-T., Binder, A., & Cheung, N. M. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19784-0_39 · Cited by 32
Visual User-Generated Content Verification in Journalism: An Overview
Khan, S. A., Sheikhi, G., Opdahl, A. L., Rabbi, F., Stoppel, S., Trattner, C., & Dang‐Nguyen, D. (2023) · IEEE Access · doi:10.1109/access.2023.3236993 · Cited by 36
“For more general deepfake media detection, in [126] Zhang et al.”
MethodFCD-Net: Learning to Detect Multiple Types of Homologous Deepfake Face Images
Han, R., Wang, X., Bai, N., Wang, Q., Liu, Z., & Xue, J. (2023) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2023.3269152 · Cited by 34
“Zhang et al. [22] proposed a classifier model based on spectrum input, which can simulate artifacts generated from common channels shared by several popular GAN models.”
BackgroundGenerative Adversarial Network Applications in Industry 4.0: A Review
Akar, C. A., Massih, R. A., Yaghi, A., Khalil, J., Kamradt, M., & Makhoul, A. (2024) · International Journal of Computer Vision · doi:10.1007/s11263-023-01966-9 · Cited by 37
LaWa: Using Latent Space for In-Generation Image Watermarking
Rezaei, A., Akbari, M., Alvar, S. R., Fatemi, A., & Zhang, Y. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73024-5_8 · Cited by 44
NoiseScope: Detecting Deepfake Images in a Blind Setting
Pu, J., Mangaokar, N., Wang, B., Reddy, C. K. A., & Viswanath, B. (2020) · Annual Computer Security Applications Conference · doi:10.1145/3427228.3427285 · Cited by 31
“[83] uses real data to train “AutoGAN”, a component that aims to simulate a GAN generator.”
MethodMagDR: Mask-guided Detection and Reconstruction for Defending Deepfakes
Chen, Z., Xie, L., Pang, S., He, Y., & Zhang, B. (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/cvpr46437.2021.00890 · Cited by 33
“To avoid negative impacts to the public, researchers started to develop algorithms to detect the images and videos that have been contaminated by deepfakes [36, 56].”
BackgroundInfluentialDetection, Attribution and Localization of GAN Generated Images
Goebel, M. G., Nataraj, L., Nanjundaswamy, T., Mohammed, T. M., Chandrasekaran, S., & Manjunath, B. S. (2021) · Electronic Imaging · doi:10.2352/issn.2470-1173.2021.4.mwsf-276 · Cited by 41
Semi-Supervised Parametric Real-World Image Harmonization
Wang, K., Gharbi, M., Zhang, H., Xia, Z., & Shechtman, E. (2023) · 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/cvpr52729.2023.00574 · Cited by 31
“GAN training is known to be unstable or cause image artifacts [36], but because our parametric harmonization model adjusts color curves and adds low-resolution shadows, instead of predicting pixels directly, it has a strong regularizing effect, which prevents the GAN training to degenerate and cause spurious artifacts…”
BackgroundCombining Automatic Speaker Verification and Prosody Analysis for Synthetic Speech Detection
Attorresi, L., Salvi, D., Borrelli, C., Bestagini, P., & Tubaro, S. (2023) · Lecture Notes in Computer Science · doi:10.1007/978-3-031-37742-6_21 · Cited by 30
Dodging DeepFake Detection via Implicit Spatial-Domain Notch Filtering
Huang, Y., Juefei-Xu, F., Guo, Q., Liu, Y., & Pu, G. (2023) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2023.3325427 · Cited by 52
Casu, M., Guarnera, L., Caponnetto, P., & Battiato, S. (2024) · Forensic Science International: Digital Investigation · doi:10.1016/j.fsidi.2024.301795 · Cited by 30
“…researchers have demonstrated that generative engines leave traces on synthetic content that can be identified and detected in the frequency domain (Guarnera et al., 2020c; Zhang et al., 2019; Marra et al., 2019; Giudice et al., 2021; Dzanic et al., 2020; Durall et al., 2020) (Figure 12).”
BackgroundGong, D. (2020) · International Journal of Advanced Trends in Computer Science and Engineering · doi:10.30534/ijatcse/2020/58932020 · Cited by 31
“It is a new classifier design method based on spectrum input[59].”
MethodGuarnera, L., Giudice, O., & Battiato, S. (2024) · Lecture notes in networks and systems · doi:10.1007/978-3-031-66431-1_43 · Cited by 32
Spectral Distribution Aware Image Generation
Jung, S., & Keuper, M. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i2.16267 · Cited by 7
A review of techniques to detect the GAN-generated fake images
Arora, T., & Soni, R. (2021) · Elsevier eBooks · doi:10.1016/b978-0-12-823519-5.00004-x · Cited by 32
Zero-Shot Detection of AI-Generated Images
Cozzolino, D., Poggi, G., Nießner, M., & Verdoliva, L. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72649-1_4 · Cited by 37
“…pre-training models on large general-purpose datasets [46], preserving fine-grain details of images [7,27], exploiting high-frequency artifacts in the spatial [43, 68, 72] or Fourier domain [18, 24, 78], leveraging inter-pixel correlation discrepancies [71,79], adopting inversion techniques [1,75].”
BackgroundMethodA Bias-Free Training Paradigm for More General AI-generated Image Detection
Guillaro, F., Zingarini, G., Usman, B., Sud, A., Cozzolino, D., & Verdoliva, L. (2025) · 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/cvpr52734.2025.01741 · Cited by 53
“For GAN-generated images, producing realistic simulated fakes requires training the generation architecture specifically for this task [20, 52].”
MD-CSDNetwork: Multi-Domain Cross Stitched Network for Deepfake Detection
Agarwal, A., Agarwal, A., Sinha, S., Mayank Vatsa, A., & Singh, R. (2021) · International Conference on Automatic Face and Gesture Recognition · doi:10.1109/fg52635.2021.9666937 · Cited by 29
Improving GAN-Generated Image Detection Generalization Using Unsupervised Domain Adaptation
Zhang, M., Wang, H., He, P., Malik, A., & Liu, H. (2022) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme52920.2022.9859763 · Cited by 28
“[12] designed AutoGAN that can simulate the artificial artifacts of GAN images.”
BackgroundDeep Image Fingerprint: Towards Low Budget Synthetic Image Detection and Model Lineage Analysis
Sinitsa, S., & Fried, O. (2024) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00402 · Cited by 46
“During the inference of the model, this fingerprint was subtracted from the output according to:”
MethodThe Potential Dangers of Artificial Intelligence for Radiology and Radiologists
Chu, L. C. H., Anandkumar, A., Shin, H. C., & Fishman, E. K. (2020) · Journal of the American College of Radiology · doi:10.1016/j.jacr.2020.04.010 · Cited by 28
Face manipulated deepfake generation and recognition approaches: a survey
Rehaan, M., Kaur, N., & Kingra, S. (2023) · Smart Science · doi:10.1080/23080477.2023.2268380 · Cited by 34
“In addition, one of the GAN-based techniques [101] also performed well for attribute manipulation detection with an accuracy of 100%.”
MethodArtifact feature purification for cross-domain detection of AI-generated images
Meng, Z., Peng, B., Dong, J., Tan, T., & Cheng, H. (2024) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2024.104078 · Cited by 31
Deep learning technology for face forgery detection: A survey
Ma, L., Yang, P., Xu, Y., Yang, Z., Li, P., & Huang, H. (2024) · Neurocomputing · doi:10.1016/j.neucom.2024.129055 · Cited by 31
GazeForensics: DeepFake detection via gaze-guided spatial inconsistency learning
He, Q., Peng, C., Liu, D., Wang, N., & Gao, X. (2024) · Neural Networks · doi:10.1016/j.neunet.2024.106636 · Cited by 34
A spatio-frequency cross fusion model for deepfake detection and segmentation
Zheng, J., Zhou, Y., Zhang, N., Hu, X., Xu, K., Gao, D., & Tang, Z. (2025) · Neurocomputing · doi:10.1016/j.neucom.2025.129683 · Cited by 29
Detecting Tampered Scene Text in the Wild
Wang, Y., Xie, H., Xing, M., Wang, J., Zhu, S., & Zhang, Y. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19815-1_13 · Cited by 26
“Secondly, image manipulations are proved to leave high-frequency traces [41,17].”
BackgroundMethodZhang, 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 36
A Novel Neural Model based Framework for Detection of GAN Generated Fake Images
Agarwal, S., Girdhar, N., & Raghav, H. (2021) · International Conference on Cloud Computing, Data Science & Engineering (Confluence) · doi:10.1109/confluence51648.2021.9377150 · Cited by 25
Fake Face Detection using Local Binary Pattern and Ensemble Modeling
Wang, Y., Zarghami, V., & Cui, S. (2021) · Proceedings - International Conference on Image Processing · doi:10.1109/icip42928.2021.9506460 · Cited by 25
“The authors of [15] demonstrated that fake faces tend to have anomalies in frequency domain compared to real faces, and the authors of [6] trained their deepfake image detector based on image spectrum and showed promising results.”
BackgroundPatch Diffusion: A General Module for Face Manipulation Detection
Zhang, B., Li, S., Feng, G., Qian, Z., & Zhang, X. (2022) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v36i3.20233 · Cited by 25
“Some studies (Chai et al. 2020; Mayer and Stamm 2019) have shown that, the subtle traces located inside small local patches such as manipulation contours or anomaly in texture (Zhang, Karaman, and Chang 2019), would be helpful for differentiating the genuine and fake area.”
BackgroundSelf-citationFingerprintNet: Synthesized Fingerprints for Generated Image Detection
Jeong, Y., Kim, D., Ro, Y., Kim, P., & Choi, J. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19781-9_5 · Cited by 25
“[18] analyzed the GAN-based artifacts using Discrete Cosine Transform [1], and [56] suggested studying the artifacts induced by the up-sampler of GANs.”
MethodXiao, S., Lan, G., Yang, J., Li, Y., & Wen, J. (2022) · IEEE Transactions on Computational Social Systems · doi:10.1109/tcss.2022.3213832 · Cited by 27
“The system in [35] extracts and uses mesoscopic features, showing 96.8% Acc on F2F original quality.”
MethodInfluentialFake visual content detection using two-stream convolutional neural networks
Yousaf, B., Usama, M., Sultani, W., Mahmood, A., & Qadir, J. (2022) · Neural Computing and Applications · doi:10.1007/s00521-022-06902-5 · Cited by 23
“Zhang et al. [42] proposed a generalizable architecture named AutoGAN and evaluated its generalization ability on two types of generative networks.”
BackgroundExposing unseen GAN-generated image using unsupervised domain adaptation
Zhang, M., Wang, H., He, P., Malik, A., & Liu, H. (2022) · Knowledge-Based Systems · doi:10.1016/j.knosys.2022.109905 · Cited by 23
FeatureTransfer: Unsupervised Domain Adaptation for Cross-Domain Deepfake Detection
Chen, B., & Tan, S. (2021) · Security and Communication Networks · doi:10.1155/2021/9942754 · Cited by 23
“[14] found that the upsample or transposed Hindawi”
BackgroundLiu, D., Chen, T., Peng, C., Wang, N., Hu, R., & Gao, X. (2024) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2024.3516561 · Cited by 22
“The literature [5] reveals that the forged images generated by GAN models have a special distortion feature and distinguish them in the frequency domain spectrogram.”
BackgroundFake or JPEG? Revealing Common Biases in Generated Image Detection Datasets
Grommelt, P., Weiss, L., Pfreundt, F., & Keuper, J. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-92089-9_6 · Cited by 48
Misleading Deep-Fake Detection with GAN Fingerprints
Wesselkamp, V., Rieck, K., Arp, D., & Quiring, E. (2022) · IEEE Security and Privacy Workshops (SPW) · doi:10.1109/spw54247.2022.9833860 · Cited by 26
On the Exploitation of Deepfake Model Recognition
Guarnera, L., Giudice, O., Nießner, M., & Battiato, S. (2022) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw56347.2022.00016 · Cited by 26
“On this topic, several overviews on Media forensics with a particular focus on Deepfakes have been recently published [11, 26, 33, 35, 44].”
BackgroundMaLP: Manipulation Localization Using a Proactive Scheme
Asnani, V., Yin, X., Hassner, T., & Liu, X. (2023) · 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/cvpr52729.2023.01188 · Cited by 24
“Binary Detection We compare with prior proactive and passive approaches for binary manipulation detection [1, 40, 60, 68].”
BackgroundMethodImproving Synthetically Generated Image Detection in Cross-Concept Settings
Dogoulis, P., Kordopatis-Zilos, G., Kompatsiaris, I. Y., & Papadopoulos, S. (2023) · ACM International Workshop on Multimedia AI against Disinformation (MAD) · doi:10.1145/3592572.3592846 · Cited by 27
“Zhang et al. [41] propose detecting artifacts in the spectrum space and use them as input to a ResNet34 network.”
MethodNarrowing Domain Gaps With Bridging Samples for Generalized Face Forgery Detection
Yu, Y., Ni, R., Yang, S., Zhao, Y., & Kot, A. C. (2023) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2023.3310341 · Cited by 22
“Typically, a number of works focus on mining intrinsic features to prevent distribution gaps, e.g., frequency-aware clues [27], [28], [29], [30], [31], [32], up-sampling traces [48], and multimodal information [49], [50].”
BackgroundMulti-Modality Image Manipulation Detection
Yang, C., Wang, Z., Shen, H., Li, H., & Jiang, B. (2021) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme51207.2021.9428232 · Cited by 20
“We show the spectra of each high-pass filtered image, similar to [9].”
Deep Learning-Based Gender Classification by Training With Fake Data
Oulad-Kaddour, M., Haddadou, H., Conde, C., Alonso, D. P., Benatchba, K., & Cabello, E. (2023) · IEEE Access · doi:10.1109/access.2023.3328210 · Cited by 20
“In recent years, learning-based approaches that allow the creation of fake identities for living beings have emerged [21], [22].”
BackgroundFighting deepfakes using body language analysis
Yasrab, R., Jiang, W., & Riaz, A. (2021) · Forecasting · doi:10.3390/forecast3020020 · Cited by 20
MaskSim: Detection of synthetic images by masked spectrum similarity analysis
Li, Y., Bammey, Q., Gardella, M., Nikoukhah, T., Morel, J., Colom, M., & von Gioi, R. G. (2024) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw63382.2024.00390 · Cited by 22
“AutoGAN [77] employs a classifier in the spectral domain to identify synthetic images based on their frequency artifacts.”
BackgroundDao, V., Arellano, C. J., Nguyen, P., Almutlaq, F., Hsu, K., & Sorooshian, S. (2025) · Journal of Geophysical Research Atmospheres · doi:10.1029/2024jd042181 · Cited by 24
“These issues likely arise from the adversarial training process, where the emphasis on visual realism can sometimes result in patterns that lack physical consistency (Kinoshita & Kiya, 2020; Zhang et al., 2019).”
D 3 : Scaling Up Deepfake Detection by Learning from Discrepancy
Yang, Y., Qian, Z., Zhu, Y., Russakovsky, O., & Wu, Y. (2025) · 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/cvpr52734.2025.02221 · Cited by 39
Attributing and Detecting Fake Images Generated by Known GANs
Joslin, M., & Hao, S. (2020) · IEEE Security and Privacy Workshops (SPW) · doi:10.1109/spw50608.2020.00019 · Cited by 19
“While we do observe degraded performance for most of the evasions, our method proves fairly robust against these evasions, as compared to prior work which experienced dramatically decreased accuracy without proper retraining [10, 19].”
ResultWG 2 AN: Synthetic wound image generation using generative adversarial network
Sarp, S., Kuzlu, M., Wilson, E., & Güler, Ö. (2021) · The Journal of Engineering · doi:10.1049/tje2.12033 · Cited by 19
“The up-sampling layer in the generation pipeline of a GAN model, which produces high-resolution images from low-resolution ones, causes checkerboard artefacts [31].”
MethodAnti-Forensic Attacks Using Generative Adversarial Networks
Stamm, M. C., & Zhao, X. (2022) · Advances in computer vision and pattern recognition · doi:10.1007/978-981-16-7621-5_17 · Cited by 20
AI-Generated Face Image Identification with Different Color Space Channel Combinations
Mo, S., Lü, P., & Liu, X. (2022) · Sensors · doi:10.3390/s22218228 · Cited by 19
“[9] theoretically analyzed the existence of the spectral replication of artifacts in the frequency domain and trained a classifier based on spectral rather than pixel inputs to achieve the effective detection of generated artifacts.”
BackgroundPerceptual Artifacts Localization for Image Synthesis Tasks
Zhang, L., Xu, Z., Barnes, C., Zhou, Y., Liu, Q., Zhang, H., Amirghodsi, S., Lin, Z., Shechtman, E., & Shi, J. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00697 · Cited by 27
“Given that new state-of-the-art generative models emerge frequently, an essential question arises: Can we train a model that generalizes to entirely unseen generative models? This query is explored in several works [11, 54, 67, 61].”
BackgroundDetection of GAN-Synthesized Image Based on Discrete Wavelet Transform
Tang, G., Sun, L., Mao, X., Guo, S., Zhang, H., & Wang, X. (2021) · Security and Communication Networks · doi:10.1155/2021/5511435 · Cited by 19
“[6] explored the fingerprint of GAN [22] and proposed a classifier model named AutoGAN based on the input of frequency spectrum.”
BackgroundMethodInfluentialA-PixelHop: A Green, Robust and Explainable Fake-Image Detector
Zhu, Y. C., Wang, X., Chen, H.-S., Salloum, R., & Kuo, C. ‐. J. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9747901 · Cited by 19
Deepfakes Detection Methods: A Literature Survey
Weerawardana, M., & Fernando, T. (2021) · 2021 10th International Conference on Information and Automation for Sustainability (ICIAfS) · doi:10.1109/iciafs52090.2021.9606067 · Cited by 23
A comparative study of deepfake facial manipulation technique using generative adversarial networks
Alkishri, W., Yousif, J. H., Bahri, M. A., Zakarya, M., Khan, N., Al-Maskari, S., & Gürhanlı, A. (2025) · Discover Artificial Intelligence · doi:10.1007/s44163-025-00337-2 · Cited by 21
Accurate MRI Reconstruction via Multi-Domain Recurrent Networks
Wei, J., Wang, Z., Wang, K., Guo, L., Fu, X., Liu, J., & Chen, X. (2023) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2023/169 · Cited by 18
Borghi, G., Domenico, N. D., Franco, A., Ferrara, M., & Maltoni, D. (2023) · IEEE Access · doi:10.1109/access.2023.3328227 · Cited by 22
“In morphed faces generated with the StyleGAN-based approach visual artifacts are less visible, but common GAN-related textures are still present and visible [47].”
MethodCoexistence of Deepfake Defenses: Addressing the Poisoning Challenge
Park, J., Park, L. H., Ahn, H. E., & Kwon, T. (2024) · IEEE Access · doi:10.1109/access.2024.3353785 · Cited by 24
“Techniques for deepfake detection include spatial-based detection [11], which seeks visual artifacts; frequency-based detection [12], which identifies features in the frequency domain; and biological signal-based detection [13], which leverages biological signals.”
BackgroundDeepfake-Image Anti-Forensics with Adversarial Examples Attacks
Fan, L., Li, W., & Cui, X. (2021) · Future Internet · doi:10.3390/fi13110288 · Cited by 17
GAN-generated fake face detection via two-stream CNN with PRNU in the wild
Wang, J., Zeng, K., Ma, B., Luo, X., Yin, Q., Liu, G., & Jha, S. K. (2022) · Multimedia Tools and Applications · doi:10.1007/s11042-021-11592-7 · Cited by 17
“Zhang et al. [35] designed a classifier model named AutoGAN based on spectral input to detect fake faces by exploring GAN fingerprint features.”
BackgroundOverview of Image Inpainting and Forensic Technology
Liu, K., Li, J., & Bukhari, S. S. H. (2022) · Security and Communication Networks · doi:10.1155/2022/9291971 · Cited by 19
Matatov, H., Naaman, M., & Amir, O. (2022) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3555599 · Cited by 17
“, political misinformation campaigns on Twitter (or other similar social media platforms), should consider investing their efforts in other attributes of the shared post such as date and account characteristics, perhaps alongside the efforts of looking for manipulated content suggested in several recent works [6, 24,…”
BackgroundRobust Watermarking for Video Forgery Detection with Improved Imperceptibility and Robustness
Zhou, Y., Ying, Q., Wang, Y., Zhang, X., Qian, Z., & Zhang, X. (2022) · IEEE International Workshop on Multimedia Signal Processing (MMSP) · doi:10.1109/mmsp55362.2022.9949466 · Cited by 17
“[6] mine some traces in the frequency spectrum and detected the images generated by GAN in the frequency domain.”
BackgroundSelf-citationExploring the Effect of High-frequency Components in GANs Training
Li, Z., Xia, P., Xue, R., & Li, B. (2022) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3578585 · Cited by 19
“The same phenomenon has also been found in other works [13], [12], [9], [11], [10], [8].”
BackgroundResultSpatial-Intensity Transforms for Medical Image-to-Image Translation
Wang, C. J., Rost, N. S., & Golland, P. (2023) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2023.3283948 · Cited by 18
“3) Artifacts in Generative Models: Even GANs that produce perceptually convincing outputs can introduce subtle artifacts into their images [34], [35].”
BackgroundOn the Exploitation of DCT-Traces in the Generative-AI Domain
Pontorno, O., Guarnera, L., & Battiato, S. (2024) · Proceedings - International Conference on Image Processing · doi:10.1109/icip51287.2024.10648013 · Cited by 22
“The analysis in the Fourier domain [16] is the base of most methods in literature [5, 6] in order to discover and study those discriminative traces on synthetic data with respect to real content.”
MethodDiscrete Fourier Transform in Unmasking Deepfake Images: A Comparative Study of StyleGAN Creations
Convertini, V. N., Impedovo, D., Lopez, U., Pirlo, G., & Sterlicchio, G. (2024) · Information · doi:10.3390/info15110711 · Cited by 25
High Frequency Detail Accentuation in CNN Image Restoration
Ayyoubzadeh, S. M., & Wu, X. (2021) · IEEE Transactions on Image Processing · doi:10.1109/tip.2021.3120678 · Cited by 16
Deepfake Detection Method Based on Cross-Domain Fusion
Sun, F., Zhang, N., Xu, P., & Song, Z. (2021) · Security and Communication Networks · doi:10.1155/2021/2482942 · Cited by 21
Haubold, J., Jost, G., Theysohn, J. M., Ludwig, J. M., Li, Y., Kleesiek, J., Schaarschmidt, B. M., Forsting, M., Nensa, F., Pietsch, H., et al. (2022) · Investigative Radiology · doi:10.1097/rli.0000000000000875 · Cited by 16
Two-Stream Xception Structure Based on Feature Fusion for DeepFake Detection
Wang, B., Huang, L., Huang, T., & Ye, F. (2023) · International Journal of Computational Intelligence Systems · doi:10.1007/s44196-023-00312-8 · Cited by 17
Pedestrian wind flow prediction using spatial-frequency generative adversarial network
Wang, P., Guo, M., Cao, Y., Hao, S., Zhou, X., & Zhao, L. (2023) · Building Simulation · doi:10.1007/s12273-023-1071-8 · Cited by 16
“Moreover, recent studies have revealed that the frequency domain gap between truths and GAN-generated images may be a common issue, albeit slight variations (Zhang et al. 2019; Frank et al. 2020).”
MCW: A Generalizable Deepfake Detection Method for Few-Shot Learning
Guan, L., Liu, F., Zhang, R., Liu, J., & Tang, Y. (2023) · Sensors · doi:10.3390/s23218763 · Cited by 17
“Xu Zhang et al. proposed a classifier based on frequency domain features [12].”
Background
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
