The DOI matches its Crossref record; 15 of the 15 references it cites were found in a registry or an index.
Deepfake Video Detection through Optical Flow Based CNN
Open accessAbstract
Recent advances in visual media technology have led to new tools for processing and, above all, generating multimedia contents. In particular, modern AI-based technologies have provided easy-to-use tools to create extremely realistic manipulated videos. Such synthetic videos, named Deep Fakes, may constitute a serious threat to attack the reputation of public subjects or to address the general opinion on a certain event. According to this, being able to individuate this kind of fake information becomes fundamental. In this work, a new forensic technique able to discern between fake and original video sequences is given; unlike other state-of-the-art methods which resorts at single video frames, we propose the adoption of optical flow fields to exploit possible inter-frame dissimilarities. Such a clue is then used as feature to be learned by CNN classifiers. Preliminary results obtained on FaceForensics++ dataset highlight very promising performances.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
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
“the re- cent common underlying mechanism for deepfake cre- ation is deep learning models such as autoencoders and generative adversarial networks (GANs)”
From Deep learning for deepfakes creation and detection: A survey (2022) · cites this paper as [7]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Amerini etal. (2019) have used CNN with optical flow to differentiate between fake and real videos.”
From Deepfake video detection: challenges and opportunities (2024) · cites this paper as Amerini et al. 2019
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
Very Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148
StarGAN: 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
FaceForensics++: Learning to Detect Manipulated Facial Images
Rossler, A., Cozzolino, D., Verdoliva, L., Riess, C., Thies, J., & Niessner, M. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00009 · Cited by 3,314
PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume
Sun, D., Yang, X., Liu, M.-Y., & Kautz, J. (2018) · 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/cvpr.2018.00931 · Cited by 2,801
MesoNet: a Compact Facial Video Forgery Detection Network
Afchar, D., Nozick, V., Yamagishi, J., & Echizen, I. (2018) · 2018 IEEE International Workshop on Information Forensics and Security (WIFS) · doi:10.1109/wifs.2018.8630761 · Cited by 1,872
The computation of optical flow
Beauchemin, S. S., & Barron, J. A. (1995) · ACM Computing Surveys · doi:10.1145/212094.212141 · Cited by 1,313
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
Kim, H., Garrido, P., Tewari, A., Xu, W., Thies, J., Nießner, M., Pérez, P., Richardt, C., Zollhöfer, M., & Theobalt, C. (2018) · ACM Transactions on Graphics · doi:10.1145/3197517.3201283 · Cited by 702
Thies, J., Zollhöfer, M., Stamminger, M., Theobalt, C., & Nießner, M. (2016) · ACM SIGGRAPH Emerging Technologies · doi:10.1145/2929464.2929475 · Cited by 589
Flagged first, then by citations
Cited by
Sorted by citations
Items 201–300 of 425
Fighting Malicious Media Data: A Survey on Tampering Detection and Deepfake Detection
Wang, J., Li, Z., Zhang, C., Chen, J., Wu, Z., Davis, L. S., & Jiang, Y. (2025) · Proceedings of the IEEE · doi:10.1109/jproc.2025.3576367 · Cited by 18
“…video detectors using multimodal information: Videos are embedded with multimodal information, i.e., motion (optical flow), and audio by nature, which inspires several works to explore multimodal inputs [362], [363], [363] or cross-modal alignment [364], [365], [366] for Deepfake forensics.”
Birla, L., Saikia, T., & Gupta, P. (2024) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3702232 · Cited by 9
“Alternatively, the DL networks can be employed to detect Deepfakes, like Capsule network [36], Recurrent Convolutional network [17], and VGG16 [1].”
Alkurdi, D. A., Cevik, M., & Akgündoğdu, A. (2024) · Computers, materials & continua/Computers, materials & continua (Print) · doi:10.32604/cmc.2024.057029 · Cited by 19
“[14] 98.20% 97.00% 95.50% 96.25% SVM [15] 90.24% 85.00% 80.00% 82.50% CNN [16] 98.52% 98.00% 97.00% 97.50% Fig.”
InfluentialGupta, V., Srivastava, V., Yadav, A. K., Vishwakarma, D. K., & Kumar, N. (2025) · Applied Intelligence · doi:10.1007/s10489-024-06168-5 · Cited by 9
“…et al. [70] 2020 0.9210 - 0.8390 - - - 0.9250 - 0 .7820 - - - Rössler et al. [33] 2019 0.7450 - 0.7590 - - - 0.7090 - 0 .7330 - - - Amerini et al. [20] 2019 - - 0.8161 - - - - - - - - - Afchar et al. [71] 2018 0.8727 - 0.5620 - - - 0.6117 - 0.4067 - 0.6132 - FreqFaceNet (Proposed) - 0.9503 0.9919…”
InfluentialDagar, D., & Vishwakarma, D. K. (2023) · International Conference on Mobile Networks and Wireless Communications (ICMNWC) · doi:10.1109/icmnwc60182.2023.10435983 · Cited by 9
“These techniques rely on a variety of cues, such as biological signals [1], GAN-based artifacts [2], visual cues [3] time lags between frames [4] and audio-visual discrepancies [5].”
BackgroundMethodReview: DeepFake Detection Techniques using Deep Neural Networks (DNN)
Chotaliya, H., Khatri, M. A., Kanojiya, S., et al. (2023) · doi:10.1109/icast59062.2023.10454938 · Cited by 9
Deepfake video detection using CNN and RNN with OPTICAL FLOW features
Reddy, E. M. S., Kumar, A. P., & Swetha, P. (2024) · IEEE International Students' Conference on Electrical, Electronics and Computer Science (SCEECS) · doi:10.1109/sceecs61402.2024.10482344 · Cited by 11
“The project, titled "Deep Fake Video Detection Using CNN and RNN with Optical Flow," aims to address this growing concern by leveraging the strengths of Convolutional Neural Networks [1] and Recurrent Neural Networks with optical flow techniques.”
MethodDetection of Machine Generated Multimedia Elements Using Deep Learning
Sanghvi, B., Shelar, H., Pandey, M., & Sisodia, J. (2021) · International Conference on Computing Methodologies and Communication (ICCMC) · doi:10.1109/iccmc51019.2021.9418008 · Cited by 8
“The concept of optical flow can be used to determine tampering in videos [10].”
MethodDeepFake and its Enabling Techniques: A Review
Brooks, R., Yuan, Y., Liu, Y., & Chen, H. (2022) · APSIPA Transactions on Signal and Information Processing · doi:10.1561/116.00000024 · Cited by 12
A Survey on Video-Based Fake News Detection Techniques
Agrawal, R., & Sharma, D. K. (2021) · International Conference on Computing for Sustainable Global Development · doi:10.1109/indiacom51348.2021.00117 · Cited by 9
“Term Memory (LSTM) Architecture CNN uses the process of embedding content that is refined into a numerical matrix and then transferred to a series of layers such as a convolutional layer where key attributes are carried with the help of multi-element detectors, and then a composite layer where feature bottom mapping t…”
MethodMining Generalized Features for Detecting AI-Manipulated Fake FacesPreprint
Yu, Y., Ni, R., & Zhao, Y. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.14129 · Cited by 9
“The recurrent neural network [24, 25] and the optical flow [26] were adopted to utilize the time information for exposing fake face videos.”
MethodEfficient Identification of DeepFake Images using CNN
Gowda, N. C., Veena, H. N., & Ramani, D. R. (2024) · International Conference on Intelligent Data Communication Technologies and Internet of Things (IDCIoT) · doi:10.1109/idciot59759.2024.10467555 · Cited by 8
“The authors in [12]-[14] used the CASIA, PRINT-ATTACK, and REPLAY-ATTACK databases to compare various LBP algorithms that incorporate optical flow and fusion approaches for deepfake video detection.”
MethodA systematic literature review of video forgery detection techniques
Aulakh, M. K., Kanwal, N., & Bansal, M. (2025) · Multimedia Tools and Applications · doi:10.1007/s11042-025-20769-3 · Cited by 13
Self-Supervised Feature Disentanglement for Deepfake Detection
Yan, B., Liu, P., Yang, Y., & Guo, Y. (2025) · Mathematics · doi:10.3390/math13122024 · Cited by 9
Deepfakes in Cyber Warfare: Threats, Detection, Techniques and Countermeasures
Cinar, B. (2023) · Asian Journal of Research in Computer Science · doi:10.9734/ajrcos/2023/v16i4381 · Cited by 10
Optifake: optical flow extraction for deepfake detection using ensemble learning technique
Vashishtha, S., Gaur, H., Das, U., Sourav, S., Bhattacharjee, E., & Kumar, T. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-18641-x · Cited by 8
“The next paper [2] advocated the use of optical flow fields to exploit probable inter-frame dissimilarities, which would subsequently be used as a feature to be learned by CNN classifiers and demonstrated this using FlowCNN on VGG16 and ResNet50 networks.”
Deep Learning for Multimedia Forensics
Amerini, I., Anagnostopoulos, A., Maiano, L., & Celsi, L. R. (2021) · doi:10.1561/9781680838558 · Cited by 8
Self-citationCross-Domain Local Characteristic Enhanced Deepfake Video Detection
Liu, Z., Wang, H., & Wang, S. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-26348-4_12 · Cited by 9
Luo, Y., Ye, F., Weng, B., Du, S., & Huang, T. (2021) · Security and Communication Networks · doi:10.1155/2021/4280328 · Cited by 9
“[33] used the optical flow method to detect facial manipulation videos.”
MethodExtensive Analysis of Deep Learning-based Deepfake Video Detection
Myvizhi, D., & Pamila, J. C. M. J. (2022) · Journal of Ubiquitous Computing and Communication Technologies · doi:10.36548/jucct.2022.1.001 · Cited by 8
Generating Higher-Quality Anti-Forensics DeepFakes with Adversarial Sharpening Mask
Fan, B., Ding, F., Zhu, G., Huang, J., Kwong, S., Atrey, P., & Lyu, S. (2025) · ACM Transactions on Multimedia Computing, Communications, and Applications · doi:10.1145/3729233 · Cited by 8
“It effectively reduces the detection accuracy of forensic models with good transferability.”
DiffFake: Exposing Deepfakes Using Differential Anomaly Detection
Stamnas, S., & Sánchez, V. (2025) · IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw65960.2025.00079 · Cited by 8
“It is important to note that differential anomaly detection does not exploit inter-frame dissimilarities between subsequent frames (temporal coherence) as done in [9,45], but rather leverages information extracted from changes of pose or facial expressions.”
Fang, J., Ding, X., Feng, J., Yu, Y., & He, L. (2025) · IEEE Transactions on Consumer Electronics · doi:10.1109/tce.2025.3580624 · Cited by 7
“On the other hand, CNN-based models [26], [27], [28], [29] typically focus on detecting pixel-level inconsistencies, such as unnatural textures, abnormal face landmarks, and artifacts introduced during the image manipulation process.”
Channel-Wise Spatiotemporal Aggregation Technology for Face Video Forensics
Lu, Y., Liu, Y., Fei, J., & Xia, Z. (2021) · Security and Communication Networks · doi:10.1155/2021/5524930 · Cited by 7
A Novel Face Forgery Detection Method Based on Edge Details and Reused-Network
Wang, J., Li, X., & Zhao, Y. (2021) · International Conference on Computer Science and Network Technology (ICCSNT) · doi:10.1109/iccsnt53786.2021.9615487 · Cited by 9
Comparative Analysis of Deep Fake Detection Techniques
Alanazi, F. (2022) · International Conference on Computational Intelligence and Communication Networks (CICN) · doi:10.1109/cicn56167.2022.10008363 · Cited by 7
“A spatial-based method was proposed in the research [17] , which adopts end-to-end convolutional neural network classifiers and a sequence-based methodology, applying a long short-term memory (LSTM) model.”
MethodDeep Scene Flow Learning: From 2D Images to 3D Point Clouds
Xiang, X., Abdein, R., Li, W., & Saddik, A. E. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3319448 · Cited by 7
How Do Deepfakes Move? Motion Magnification for Deepfake Source Detection
Demir, İ., & Çiftçi, U. A. (2024) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00471 · Cited by 13
“Initial deepfake detection research focused on finding pixel-level artifacts directly from data, proposing “blind” detectors [11, 13, 16, 33, 34, 41, 47, 54, 65, 77, 78, 78].”
BackgroundAn Experimental Evaluation of Recent Face Recognition Losses for Deepfake Detection
Liu, Y., Chang, C., Chen, I.-H., Ku, Y.-R., & Chen, J.-C. (2021) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr48806.2021.9413323 · Cited by 7
“On the other hand, for manipulated video content, based on unusual motion detection, [1] proposes to leverage the optical flow of facial expression as the detection cue since the motion of real face acts differently from forged one.”
BackgroundLearning Diverse Local Patterns for Deepfake Detection with Image-level Supervision
Ling, H., Huang, J., Zhao, C., Yao, Y., Chen, J., & Li, P. (2021) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn52387.2021.9533912 · Cited by 7
“Heart rate [16] and optical flow pattern [5] also contribute.”
BackgroundAttention-based Multimodal learning framework for Generalized Audio-Visual Deepfake Detection
Masood, M., Javed, A., & Irtaza, A. (2023) · Preprint (Research Square) · doi:10.21203/rs.3.rs-3415144/v1 · Cited by 6
Mining Generalized Multi-timescale Inconsistency for Detecting Deepfake Videos
Yu, Y., Ni, R., Yang, S., Ni, Y., Zhao, Y., & Kot, A. C. (2025) · International Journal of Computer Vision · doi:10.1007/s11263-024-02249-7 · Cited by 7
“Initial approaches leverage 3D CNN (Ganiyusufoglu et al., 2020; Choi et al., 2020; Zhang et al., 2021), RNN (Güera & Delp, 2018; Sabir et al., 2019), optical flow (Amerini et al., 2019), or LSTM (Masi et al., 2020; Chintha et al., 2020) to mine these temporal inconsistencies.”
X3A: Efficient Multimodal Deepfake Detection with Score-Level Fusion
Park, C. S., Moon, B., Jeon, M., Jung, J.-W., & Woo, S. S. (2025) · ACM Symposium on Applied Computing · doi:10.1145/3672608.3707934 · Cited by 8
“One can model interactions between every frame with dense sampling techniques [22] or model the difference between two adjacent frames using optical flow analysis [5].”
Exploring the Advancements and Challenges of Deepfake Face-swap: A Survey
Bakar, S. A. R. B. S. A., Waseem, S., Omar, Z., & Bilalashfaqahmed (2026) · Multimedia Tools and Applications · doi:10.1007/s11042-026-21285-8 · Cited by 13
Interpol review of forensic video analysis, 2019–2022
Geradts, Z., & Riphagen, Q. (2023) · Forensic Science International Synergy · doi:10.1016/j.fsisyn.2022.100309 · Cited by 9
“[3] is different from other spatio-temporal based methods in that you only need 2 frames to classify a video.”
InfluentialUnmasking DeepFake Visual Content with Generative AI
Roy, M., & Raval, M. S. (2023) · IEEE Region Humanitarian Technology Conference · doi:10.1109/r10-htc57504.2023.10461811 · Cited by 7
“In particular, unlike state-of-the-art methods, which usually act in a frame-based fashion, [22] presents a sequence-based approach dedicated to investigating possible dissimilarities in the temporal structure of a video.”
BackgroundA 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
Attentional Local Contrastive Learning for Face Forgery Detection
Dai, Y., Fei, J., Wang, H., & Xia, Z. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-15919-0_59 · Cited by 6
Improvised detection of deepfakes from visual inputs using light weight deep ensemble model
Panda, S. K., Diwan, T., Kakde, O. G., & Tembhurne, J. V. (2022) · Multimedia Tools and Applications · doi:10.1007/s11042-022-14307-8 · Cited by 6
“In [3], a new forensic technique is introduced to differentiate between fake and real images within the videos.”
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
Deep Fake Detection: Unmasking the Illusion using CNN and LSTM
Niranjani, V., S, A. S., Devamitra, T., & Jagapreetha, B. (2023) · International Conference on Innovative Mechanisms for Industry Applications · doi:10.1109/icimia60377.2023.10426523 · Cited by 6
“The DeepFake Detection Challenge (DFDC) [10] dataset, which is freely available and downloaded from a repository, in our deep fake detection studies.”
MethodUnknown presentation attack detection against rational attackers
Khodabakhsh, A., & Akhtar, Z. (2021) · IET Biometrics · doi:10.1049/bme2.12053 · Cited by 6
“Some methods take into account the temporal aspect of videos and propose the use of long short-term memory (LSTM) networks [52], recurrent CNNs [53], and optical fields [54].”
MethodExploiting Inconsistencies in Object Representations for Deepfake Video Detection
Bhaumik, K. K., & Woo, S. S. (2023) · doi:10.1145/3595353.3595885 · Cited by 5
“[1] focus on the inconsistency of optical flow between video frames.”
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
PhilaeX: Explaining the Failure and Success of AI Models in Malware Detection
Lu, Z., & Thing, V. L. L. (2022) · International Conference on Internet of Things, Big Data and Security (IoTBDS) · doi:10.5220/0010986700003194 · Cited by 5
“(Zeiler and Fergus, 2014) proposed a qualitative explanation method through the visualization and observation on the neurons in a convolutional neural networks (CNNs) that shows how each neuron responds to different data instances.”
MethodInfluentialDual-Modality Co-Learning for Unveiling Deepfake in Spatio-Temporal Space
Guan, J., Zhou, H., Guo, Z., Hu, T., Deng, L., Quan, C., Fang, M., & Zhao, Y. (2023) · International Conference on Multimedia Retrieval · doi:10.1145/3591106.3592284 · Cited by 5
“[3] propose to detect deepfakes by directly replacing the RGB frame with the flow as input.”
MethodA survey on deepfake video detection datasets
Sohan, M. F., Solaiman, M., & Hasan, M. A. (2023) · Indonesian Journal of Electrical Engineering and Computer Science · doi:10.11591/ijeecs.v32.i2.pp1168-1176 · Cited by 5
“2 Faceforensics++ [24], [25], [26], [27], [28],”
Mining Temporal Inconsistency with 3D Face Model for Deepfake Video Detection
Cheng, Z., Chen, C., Zhou, Y., & Hu, X. (2023) · Lecture notes in computer science · doi:10.1007/978-981-99-8540-1_19 · Cited by 6
Attending Generalizability in Course of Deep Fake Detection by Exploring Multi-task Learning
Balaji, P., Das, A., Das, S., & Dantcheva, A. (2023) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw60793.2023.00054 · Cited by 5
“This for the detection of temporal inconsistencies with respect to the lip movement [2], jitters between frames [21] and optical flow [4].”
BackgroundImage plagiarism detection using GAN - (Generative Adversarial Network)
Gayadhankar, K., Patel, R., Lodha, H., & Shinde, S. K. (2021) · ITM Web of Conferences · doi:10.1051/itmconf/20214003013 · Cited by 5
A shared updatable method of content regulation for deepfake videos based on blockchain
Mao, D., Zhao, S., & Hao, Z. (2022) · Applied Intelligence · doi:10.1007/s10489-021-03156-x · Cited by 5
“used the change of pixel points of video frames in the time domain to find the correspondence between the existence of two adjacent frames before and after, so as to calculate the optical flow vector of adjacent frames [16].”
MethodImpact of Deepfake Technology on Digital World Authenticity: A Review
Ahmed, S. (2022) · International Journal of Engineering and Management Research · doi:10.31033/ijemr.12.3.10 · Cited by 6
Unmasking Deepfakes: A Review of Current Datasets, Tools, and Detection Features
Garg, D., & Gill, R. (2025) · Procedia Computer Science · doi:10.1016/j.procs.2025.04.129 · Cited by 6
It Wasn’t Me: Irregular Identity in Deepfake Videos
Liu, H., Bestagini, P., Huang, L., Zhou, W., Tubaro, S., Zhang, W., & Yu, N. (2023) · Proceedings - International Conference on Image Processing · doi:10.1109/icip49359.2023.10222654 · Cited by 5
“[23] adopts image noise or video motion traces to detect DeepFake videos with better interpretability.”
MethodA Performance Analysis to Detect Synthetic Images with Deep Learning-Based Image Classification
Gautam, V., Joshi, M., Tiwari, R. G., Agarwal, A. K., & Trivedi, N. K. (2024) · Lecture notes in networks and systems · doi:10.1007/978-981-97-3594-5_3 · Cited by 4
GFADE: generalized feature adaptation and discrimination enhancement for deepfake detection
Tian, Z., & Yi, J. (2025) · PeerJ Computer Science · doi:10.7717/peerj-cs.2879 · Cited by 5
“Video forgery detection efforts are focused on the temporal inconsistency ( Wang et al., 2023b ), eye blinking ( Li, Chang & Lyu, 2018 ), and optical flow ( Amerini et al., 2019 ).”
Melouk, M. M., Shao, M., Basit, A., Abouzahir, C., Zhou, R., & Shafique, M. (2025) · IEEE Access · doi:10.1109/access.2025.3614455 · Cited by 6
“Although existing works [17], [20], [70], [71], [72] have explored using optical flow in deepfake detection, our approach uniquely integrates (i) transformer-based optical flow estimation (FlowFormer ++ ) rather than a simpler flow computation pipeline, (ii) multiple specialized detectors…”
Video Forgery Detection Using Spatio-Temporal Dual Transformer
Liu, C., Li, J., Duan, J., & Huang, H. (2022) · International Conference on Computing and Pattern Recognition (ICCPR) · doi:10.1145/3581807.3581847 · Cited by 4
“Video detectionmethods can discover differences in the temporal dimension of video frames [16, 39], such as [49] using spatial and motion information, and [2, 3, 6] using optical flow for video forgery detection.”
BackgroundDiv-Df: A Diverse Manipulation Deepfake Video Dataset
Dagar, D., & Vishwakarma, D. K. (2023) · Global Conference on Information Technologies and Communications (GCITC) · doi:10.1109/gcitc60406.2023.10426446 · Cited by 4
“These methods rely on a variety of cues, including biological signals [7], GAN-based artifacts [8], visual cues [9], temporal inconsistencies between frames [10], Audiovisual discrepancy [11], while others allowed the DNN models [12], to understand the innate characteristics of the distribution.”
MethodTex-Net: texture-based parallel branch cross-attention generalized robust Deepfake detector
Dagar, D., & Vishwakarma, D. K. (2024) · Multimedia Systems · doi:10.1007/s00530-024-01424-7 · Cited by 4
“Sequential data samples can be detected using features like temporal inconsistencies between the frames [12, 13], audio-visual inconsistencies [14, 15], etc.”
FGSM Adversarial Attack Detection On Deepfake Videos
Mohamed, S. N., Ahmed, A. A., & Elsersy, W. F. (2024) · Intelligent Methods, Systems, and Applications (IMSA) · doi:10.1109/imsa61967.2024.10652708 · Cited by 4
“In order to effectively tackle this complex issue, a multidisciplinary team effort involving legis-lators, legal professionals, and media practitioners is needed, in addition to an emphasis on technology developments in detection[9].”
BackgroundLearning Meta Model for Strong Generalization Deepfake Detection
Huang, D., & Zhang, Y. (2024) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn60899.2024.10651482 · Cited by 6
“A plethora of research [6], [7], [28] has concentrated on deepfake detection via convolutional neural networks (CNNs), leveraging their robust feature learning capabilities especially for image and video analysis.”
BackgroundDetecting Deepfakes Through Inconsistencies in Local Camera Surface Frames
Ciamarra, A., Caldelli, R., & Bimbo, A. D. (2024) · IEEE International Conference on Image Processing Workshops (ICIPW) · doi:10.1109/icipcw64161.2024.10769135 · Cited by 4
“Optical flow [7] has also been studied as a motion feature capable to highlight synthetic facial movements intrinsically found in fake images.”
Self-citationDouble salted HMAC signature with blockchain for faster and secure video integrity verification
Lawrence, L., & Shreelekshmi, R. (2025) · The Journal of Supercomputing · doi:10.1007/s11227-025-06996-3 · Cited by 4
“Amerini et al. [10] present a method for detecting deepfakes based on the optical flow of many frames.”
Detecting Localized Deepfake Manipulations Using Action Unit-Guided Video Representations
Anand, T., Sankar, S., & Nair, P. (2025) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw67362.2025.00419 · Cited by 7
“CNN-based recurrent models [59] incorporated recurrent units after CNNs to capture temporal dynamics, while other methods directly learned spatiotemporal features [3, 13, 72].”
FST-Net: Exploiting Frequency Spatial Temporal Information for Low-Quality Fake Video Detection
Zhang, M., Liu, X., Liu, C., Zhang, X., & Xie, H. (2021) · IEEE International Conference on Tools with Artificial Intelligence (ICTAI) · doi:10.1109/ictai52525.2021.00087 · Cited by 5
Kumari, P., & Kaur, M. (2022) · Algorithms for intelligent systems · doi:10.1007/978-981-19-0332-8_8 · Cited by 4
MetaFake: Few-shot Face Forgery Detection with Meta Learning
Xu, N., & Feng, W. (2023) · ACM Workshop on Information Hiding and Multimedia Security (IH&MMSec) · doi:10.1145/3577163.3595108 · Cited by 4
“Convolution Neural Networks (CNNs) [21] are commonly used to discriminate forged images from real ones [1, 2], which have achieved remarkable progress, even reaching more than 95% in-dataset detection accuracy 1.”
BackgroundSiamLBP: Exploiting Texture Discrepancies for Deepfake Detection
Kingra, S., Aggarwal, N., & Kaur, N. (2023) · Lecture notes in electrical engineering · doi:10.1007/978-981-99-0085-5_36 · Cited by 5
Forged facial video detection framework based on multi-region temporal relationship feature
Fang, X., Hao, Y., Luo, Y., Xu, N., & Cao, J. (2023) · AIP Advances · doi:10.1063/5.0125032 · Cited by 4
Mining collaborative spatio-temporal clues for face forgery detection
Ding, B. Y., Fan, Z., Zhao, Z., & Xia, S. (2023) · Multimedia Tools and Applications · doi:10.1007/s11042-023-16173-4 · Cited by 4
Efficient Temporally-aware DeepFake Detection using H.264 Motion Vectors
Grönquist, P., Ren, Y., He, Q., Verardo, A., & Susstrunk, S. E. (2024) · Electronic Imaging · doi:10.2352/ei.2024.36.4.mwsf-335 · Cited by 8
A survey of Deepfake and related digital forensics
Ding, F., 况任生, Zhou, Y., Sun, L., Zhu, X., & Zhu, G. (2024) · Journal of Image and Graphics · doi:10.11834/jig.230088 · Cited by 7
Peng, C., Chen, Y., Liu, D., Wang, N., & Gao, X. (2025) · Neural Networks · doi:10.1016/j.neunet.2025.107899 · Cited by 5
Review of AIGC passive detection and proactive protection
Liu, X., Liu, H., Zhao, Y., Ni, R., Li, X., & Guo, M. (2025) · Scientia Sinica Informationis · doi:10.1360/ssi-2025-0043 · Cited by 3
“Amerini 等 [150] 提出利用光流 帧间差异检测伪造视频 , 但在快速运动或亮度后处理情况下效果较差 .”
Adversarial Reality for Evading Deepfake Image Detectors
Çiftçi, U. A., Solar, N., Greene, E., Saremsky, S. R., & Demir, İ. (2025) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw69036.2025.00171 · Cited by 4
“Blind detectors [7, 9, 11, 37, 56, 63, 66, 71, 76, 78, 82, 88, 92, 104, 106] are trained or ensembled, directly operating on raw data, though they may perform poorly in-the-wild despite high accuracy on training datasets [104].”
Uncovering visual attention-based multi-level tampering traces for face forgery detection
Yadav, A. K., Gupta, D., & Vishwakarma, D. K. (2023) · Signal Image and Video Processing · doi:10.1007/s11760-023-02774-x · Cited by 3
“Amerini et al. [20] use optical flow maps to CNNs to classify them as real or fake.”
A Comprehensive Study on Mitigating Synthetic Identity Threats Using Deepfake Detection Mechanisms
Uppal, S., Banga, V., Neeraj, S., et al. (2024) · doi:10.1109/confluence60223.2024.10463324 · Cited by 3
Video Morphing Attack Detection using Convolutional Neural Networks on Deep Fake Algorithm
Boovaneswari, S., Palanivel, N., NihilRS, S., Madhavan, R., & Daniyel, A. (2024) · International Conference on Knowledge Engineering and Communication Systems (ICKECS) · doi:10.1109/ickecs61492.2024.10616494 · Cited by 3
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
Rahman, A., Karim, M. R., Rahman, M. A., Chowdhury, P., Hossain, A., & Islam, R. (2024) · International Conference on Computer and Information Technology (ICCIT) · doi:10.1109/iccit64611.2024.11022369 · Cited by 4
“Another study [22] explored the use of optical flow-based CNNs for detecting deepfake videos.”
Wang, J., Lei, J., Li, S., & Zhang, J. (2025) · Symmetry · doi:10.3390/sym17071037 · Cited by 4
Robust Deepfake Detection for Electronic Know Your Customer Systems Using Registered Images
Amada, T., Kakizaki, K., Miyagawa, T., Ebihara, A. F., Shiohara, K., & Yamasaki, T. (2025) · International Conference on Automatic Face and Gesture Recognition · doi:10.1109/fg61629.2025.11099118 · Cited by 3
Deep Fake Face Detection Using Machine Learning and Various GAN Models
Dharmendra, K. R. D. K. R., & Tiwari, U. T. U. (2022) · International Journal of Engineering Technology and Management Sciences · doi:10.46647//ijetms.2022.v06i02.001 · Cited by 3
Akbar, M., Suaib, M., & Hussain, M. S. (2022) · International Journal of Computer and Information Technology(2279-0764) · doi:10.24203/ijcit.v11i2.282 · Cited by 6
Versoza, E. B., Tecson, G. J., Viaje, J. B., Andrada, T. A., Flores, L., Deluta, C., Melendres, E., Andres, B. B., Salvador, G. T., Bandala, A. A., et al. (2022) · IEEE International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management (HNICEM) · doi:10.1109/hnicem57413.2022.10109408 · Cited by 3
“CNN was chosen because it is considered as a standard framework for image classifications [36,37] and is known to produce state-of-theart results for many different applications [38-55].”
BackgroundLocal similarity anomaly for general face forgery detection
Dai, Y., Fei, J., Xia, Z., Liu, J.-N., & Weng, J. (2023) · Journal of Image and Graphics · doi:10.11834/jig.221006 · Cited by 4
AI Generated Art: Latent Diffusion-Based Style and Detection
Bird, J. J., Barnes, C. M., & Lotfi, A. (2024) · Advances in intelligent systems and computing · doi:10.1007/978-3-031-47508-5_13 · Cited by 3
Enhancing DeepFake Detection: Leveraging Mesonet for Video Fraud Identification
Shanmuganathan, C., Thamizharasi, M., Anish, T. P., & Sivasankari, K. (2024) · SN Computer Science · doi:10.1007/s42979-024-02629-3 · Cited by 3
“DeepFakes are a form of artificial intelligence which is a compilation of doctored images and audio put together with machine learning algorithms to produce the desired result [3].”
Noise-aware progressive multi-scale deepfake detection
Ding, X., Pang, S., & Guo, W. Q. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-18836-2 · Cited by 4
An Automated Workflow For Deepfake Detection
Joshi, A., & Chavan, C. P. (2023) · International Conference on Advances in Electronics Computers and Communications · doi:10.1109/icaecc59324.2023.10560271 · Cited by 4
“Optical Flow-based CNN approach [9], proposed by Amerini et al, detects deepfake videos using optical flow-based CNNs.”
Method3D data augmentation and dual-branch model for robust face forgery detection
Zhou, C., Li, F. W. B., Song, C., Dong, Z., & Yang, B. L. (2025) · Graphical Models · doi:10.1016/j.gmod.2025.101255 · Cited by 3
Ekler, P., & Pásztor, D. (2020) · Scientia et Securitas · doi:10.1556/112.2020.00006 · Cited by 3
Face Manipulation Detection in Remote Operational Systems
Pic, M. M., Mahfoudi, G., Trabelsi, A., & Dugelay, J. (2022) · Advances in computer vision and pattern recognition · doi:10.1007/978-3-030-87664-7_19 · Cited by 3
“For detection methods based on image or texture analysis, the authors mainly look for inconsistencies in the optical flow [24] or the presence of artefacts [25].”
MethodSpatial-temporal transformer network for protecting person-of-interest from deepfaking
Lu, D., Liu, Z., Zhang, D., Zhang, J., & Jin, G. (2025) · Multimedia Systems · doi:10.1007/s00530-024-01655-8 · Cited by 3
“…heart rate, etc.), the correlation between the visual and audio of faces in the video [24–27], and detecting temporal anomaly via inter-frame inconsistency [28–31] or optical flow [32] in fake video, which usually use 3D CNN and attention mechanism [33, 34] to locate the temporal fake clues.”
Leveraging Acoustic Features and Deep Neural Architectures for Audio Deepfake Detection
Sundaram, V., Babitha, S., & Vekkot, S. (2024) · International Conference on Advances in Computing and Communications (ICACC) · doi:10.1109/icacc63692.2024.10845392 · Cited by 2
Khan, F., Sareen, A., Kumar, A. S., & Bhuvaneswari, M. (2026) · IEEE Access · doi:10.1109/access.2026.3659021 · Cited by 3
“For example, Li et al. [14], have employed optical flow to identify temporal inconsistencies in manipulated video streams.”
Trident : Detecting Face Forgeries With Domain-Adversarial Triplet Learning
Kara, M. H., Dündar, A., & Güdükbay, U. (2026) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2026.3703428 · Cited by 2
“To capture these cues, Gu et al. [18] proposed spatiotemporal inconsistency learning, while Amerini et al. [19] utilized optical-flow convolutional neural networks (CNNs) to uncover compression-robust artifacts.”
Artificial Intelligence Against Disinformation: The FANDANGO Practical Case (short paper).
Nucci, F. S., Boi, S., & Magaldi, M. (2020) · Cited by 2
Deepfake Detection on Videos Based on Ratio Images
Testa, R. L., Machado‐Lima, A., & Nunes, F. L. S. (2022) · IIAI International Congress on Advanced Applied Informatics (IIAI-AAI) · doi:10.1109/iiaiaai55812.2022.00086 · Cited by 2
“It can also use temporal features: detecting blood volume [15], monitoring optical flow [16], and others.”
Background
Metadata
- Authors
- Amerini, I., Galteri, L., Caldelli, R., & Del Bimbo, A.
- Affiliations
- University of Florence (IT)
- Consorzio Nazionale Interuniversitario per le Telecomunicazioni (IT)
- Venue
- 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW)
- Publisher
- IEEE
- Published
- 2019-10
- Volume
- 2019
- Pages
- 1205-1207
- ISSN
- 2473-9936 (ISSN-L)
- OpenAlex
- W2995516027
- Cited by
- 424
- References
- 14
- Citations by year
- 2026: 35 · 2025: 70 · 2024: 77 · 2023: 95 · 2022: 71 · 2021: 49 · 2020: 27
- FWCI
- 11.98
- Type
- conference paper
- Language
- en
- Open access
- green
- hdl.handle.net
- Licence
- Locations
- IEEE International Conference on Computer Vision Workshops (publishedVersion)
- IRIS Research product catalog (Sapienza University of Rome) (submittedVersion, other-oa)
- IEEE Conference Proceedings
- 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
- Social Sciences › Sociology and Political Science › Misinformation and Its Impacts
- Keywords
- deepfake detection, optical flow, convolutional neural networks, manipulated video, video forensics, FaceForensics++, fake information
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
