Fake Videos Research Articles

Dual Attention Network Approaches to Face Forgery Video Detection

Forged videos are commonly spread online. Most have malicious content and cause serious information security problems. The most critical issue in deepfake detection is the identification of traces of tampering in fake videos. This study designs a Dual Attention Forgery Detection Network (DAFDN), which embeds a spatial reduction attention block (SRAB) and a forgery feature attention module (FFAM) to the backbone network. DAFDN embeds the two proposed attention mechanisms and enables the convolution neural network to extract peculiar traces left by images’ warping. This study uses two benchmark datasets, DFDC and FaceForensics++, to compare the performance of the proposed DAFDN with other methods. The results show that the proposed DAFDN mechanism achieves AUC scores of 0.911 and 0.945 in the datasets DFDC and FaceForensics++, respectively. These results are better than those of previously developed methods, such as XceptionNet and EfficientNet-related methods.

Deepfake Detection: A Systematic Literature Review

Over the last few decades, rapid progress in AI, machine learning, and deep learning has resulted in new techniques and various tools for manipulating multimedia. Though the technology has been mostly used in legitimate applications such as for entertainment and education, etc., malicious users have also exploited them for unlawful or nefarious purposes. For example, high-quality and realistic fake videos, images, or audios have been created to spread misinformation and propaganda, foment political discord and hate, or even harass and blackmail people. The manipulated, high-quality and realistic videos have become known recently as Deepfake. Various approaches have since been described in the literature to deal with the problems raised by Deepfake. To provide an updated overview of the research works in Deepfake detection, we conduct a systematic literature review (SLR) in this paper, summarizing 112 relevant articles from 2018 to 2020 that presented a variety of methodologies. We analyze them by grouping them into four different categories: deep learning-based techniques, classical machine learning-based methods, statistical techniques, and blockchain-based techniques. We also evaluate the performance of the detection capability of the various methods with respect to different datasets and conclude that the deep learning-based methods outperform other methods in Deepfake detection.

Semiotic approaches to big data visualization

With a few exceptions – focusing on journalism (Compagno 2017), on works of art (Chartier, Pulizzotto, Chartrand & Meunier 2019, Dondero 2020), on deep fake videos (Leone 2021, Dondero 2021), and on data epistemology (Bachimont 2018) – semiotics has been late to approach big data. In contrast to information and communication studies, which have been quick to recognize the big data revolution in medical, biological, political, urbanistic, and journalistic practices, and while digital humanities and digital art history continue to offer crucial insight with their analysis of large collections of artworks and heritage objects, semiotics has yet to live up to the big data challenge. In this Punctum issue we have attempted to (1) contribute to the global discussion of possible semiotic approaches to big data theory and methodology; (2) explore specifically an epistemological approach to visualizations of big data, i.e., to their manipulation, design, display, and interpretation; (3) study, from a critical vantage point, the ideologies and epistemic perspectives that underlie the acts of collecting and visualizing big data, prior to their analysis. Our overall objective has been to trace the path that leads from visualization, regarded as a unifying representation of disparate data, to its full elevation to an interpretative device.

Deep fake videos identification using ANN and LSTM

With the demand for deep learning (DL) and artificial intelligence (AI) techniques, fake digital content has multiplied in the current days. Like fake movies, images, audio and video can be an unfavorable and dangerous phenomenon and can have the ability to alter trust in the truth by providing false reality content. But the explanations provided lack of ability to provide the historical tracing and origin of electronic media. In the current research, we try to provide a general solution using different methods like ANN, CNN and LSTM with smart contracts to copy and trace the root of origin and history of the computerized content to its unique source, regardless of whether the advanced substance is replicated on numerous occasions. The shrewd agreement utilizes the hash upsides of the interplanetary document framework used to store advanced substance and related metadata. Our answer is generally centered on video content.

Deepfake Detection Using the Rate of Change between Frames Based on Computer Vision.

Recently, artificial intelligence has been successfully used in fields, such as computer vision, voice, and big data analysis. However, various problems, such as security, privacy, and ethics, also occur owing to the development of artificial intelligence. One such problem are deepfakes. Deepfake is a compound word for deep learning and fake. It refers to a fake video created using artificial intelligence technology or the production process itself. Deepfakes can be exploited for political abuse, pornography, and fake information. This paper proposes a method to determine integrity by analyzing the computer vision features of digital content. The proposed method extracts the rate of change in the computer vision features of adjacent frames and then checks whether the video is manipulated. The test demonstrated the highest detection rate of 97% compared to the existing method or machine learning method. It also maintained the highest detection rate of 96%, even for the test that manipulates the matrix of the image to avoid the convolutional neural network detection method.

Deepfake tweets classification using stacked Bi-LSTM and words embedding.

The spread of altered media in the form of fake videos, audios, and images, has been largely increased over the past few years. Advanced digital manipulation tools and techniques make it easier to generate fake content and post it on social media. In addition, tweets with deep fake content make their way to social platforms. The polarity of such tweets is significant to determine the sentiment of people about deep fakes. This paper presents a deep learning model to predict the polarity of deep fake tweets. For this purpose, a stacked bi-directional long short-term memory (SBi-LSTM) network is proposed to classify the sentiment of deep fake tweets. Several well-known machine learning classifiers are investigated as well such as support vector machine, logistic regression, Gaussian Naive Bayes, extra tree classifier, and AdaBoost classifier. These classifiers are utilized with term frequency-inverse document frequency and a bag of words feature extraction approaches. Besides, the performance of deep learning models is analyzed including long short-term memory network, gated recurrent unit, bi-direction LSTM, and convolutional neural network+LSTM. Experimental results indicate that the proposed SBi-LSTM outperforms both machine and deep learning models and achieves an accuracy of 0.92.

Digitally-Signed Video/Audio Streams as Prevention of AI-Based Attacks

The main purpose of the research is to provide the solution that allows digitally signing video and audio records, which is extremely important not only in the frame of copyright protection but also when it comes to understanding whether a video is fake. In the world where neural networks are used more and more often, it becomes easy to make a fake video using the face of some famous politician, maybe even the president of the US, and broadcast such a video, causing different kinds of negative events in the situation of the political crises.

Deepfake video detection: YOLO-Face convolution recurrent approach.

Recently, the deepfake techniques for swapping faces have been spreading, allowing easy creation of hyper-realistic fake videos. Detecting the authenticity of a video has become increasingly critical because of the potential negative impact on the world. Here, a new project is introduced; You Only Look Once Convolution Recurrent Neural Networks (YOLO-CRNNs), to detect deepfake videos. The YOLO-Face detector detects face regions from each frame in the video, whereas a fine-tuned EfficientNet-B5 is used to extract the spatial features of these faces. These features are fed as a batch of input sequences into a Bidirectional Long Short-Term Memory (Bi-LSTM), to extract the temporal features. The new scheme is then evaluated on a new large-scale dataset; CelebDF-FaceForencics++ (c23), based on a combination of two popular datasets; FaceForencies++ (c23) and Celeb-DF. It achieves an Area Under the Receiver Operating Characteristic Curve (AUROC) 89.35% score, 89.38% accuracy, 83.15% recall, 85.55% precision, and 84.33% F1-measure for pasting data approach. The experimental analysis approves the superiority of the proposed method compared to the state-of-the-art methods.

Deterrence of Piracy Employing IR Transmitter and Steganography System

Moving picture entertainment is a foremost source of amusement for populace in today’s existence. To entertain populace a lot of investment is put into film production by the film makers. Their endeavour is being ruined by few people by pirate the movies substance. They do it by capture the video recording in mobile phone camera and upload it to websites or put up for sale it to people which cause huge loss. In this research work we are proposea novel technique for reduction of film piracy by avoiding fake video recordings of video in theatres. An indistinguishable luminosity is projected from the display to the whole spectators that falls on the camera lens which is sensitive to infrared light rays Makes the recorded video unfit to watch. A method is developed for anti-piracy system for film industry using steganography technique in MATLAB.

Predicting video engagement using heterogeneous DeepWalk

Video engagement is important in online advertisements where there is no physical interaction with the consumer. Engagement can be directly measured as the number of seconds after which a consumer skips an advertisement. In this paper, we propose a model to predict video engagement of an advertisement using only a few samples. This allows for early identification of poor quality videos. This can also help identify advertisement frauds where a robot runs fake videos behind the name of well-known brands. We leverage on the fact that videos with high engagement have similar viewing patterns over time. Hence, we can create a similarity network of videos and use a graph-embedding model called DeepWalk to cluster videos into significant communities. The learned embedding is able to identify viewing patterns of fraud and popular videos. In order to assess the impact of a video, we also consider how the view counts increase or decrease over time. This results in a heterogeneous graph where an edge indicates similar video engagement or history of view counts between two videos. Since it is difficult to find labelled samples for ‘fraud’ video, we leverage on a one-class model that can determine ‘fraud’ videos with outlier or abnormal behavior. The proposed model outperforms baselines in F-measure by over 20%.

Disinformation, Deepfakes and Democracies: The Need for Legislative Reform

Rapid technological advancement is changing the way that political parties, voters, and media platforms engage with each other. This along with cultural change has led to an emerging era of disinformation and misinformation driven by both domestic and foreign actors. Political deepfakes, videos created through the use of artificial intelligence, allow individuals to rapidly create fake videos indistinguishable from true content. These videos have the capacity to undermine voter trust and could alter electoral outcomes. Regulating disinformation however raises significant free speech concerns, as well as questions about where liability should fall. In particular, holding large technology and media platforms accountable for content could lead to unintended chilling effects around freedom of expression, harming rather than protecting democratic institutions. Proposed regulations should therefore be carefully analysed through the framework of the implied freedom of political communication, ensuring that any new laws are proportionate and tailored to the threat they seek to prevent. This article analyses how current Australian law interacts with political deepfakes and proposes two targeted amendments to our federal electoral regulations to reduce the threat they pose to elections.

Real time object-based video forgery detection using YOLO (V2)

Video forgery detection is a challenging task nowadays due to fake video forwarding. Copy-Move type of attack is especially mostly practiced to tamper with the original contents of a video or an image. Copy-Move attack mainly deals with object-based video forgery. Traditional methods are quiet slow and not strong enough to detect complex Copy-Move attacks. So, automatic tamper detection in videos related to speed and accuracy is a challenging task. This paper proposes a new approach for the detection of Copy-Move attack in passive blind videos. Object-based forgery detection approach is implemented using fast and real-time object detector “You Only Look Once -Version 2″:YOLO (V2). The system is trained on the benchmark dataset videos for the automatic detection of forged objects within the video with a 0.99 confidence score. Trained YOLO (V2) model is accurately able to classify and localize the forged and non-forged objects within the given input video. The results and experimental analysis demonstrates that the proposed YOLO (V2) model achieved excellent results for detecting plain and complex Copy-Move attacks such as scaling, rotation, flipping. The performance excellent for object-based forgery detection for speed and accuracy than existing similar state-of-art deep learning approaches.

A New Deep Learning-Based Methodology for Video Deepfake Detection Using XGBoost.

Currently, face-swapping deepfake techniques are widely spread, generating a significant number of highly realistic fake videos that threaten the privacy of people and countries. Due to their devastating impacts on the world, distinguishing between real and deepfake videos has become a fundamental issue. This paper presents a new deepfake detection method: you only look once–convolutional neural network–extreme gradient boosting (YOLO-CNN-XGBoost). The YOLO face detector is employed to extract the face area from video frames, while the InceptionResNetV2 CNN is utilized to extract features from these faces. These features are fed into the XGBoost that works as a recognizer on the top level of the CNN network. The proposed method achieves 90.62% of an area under the receiver operating characteristic curve (AUC), 90.73% accuracy, 93.53% specificity, 85.39% sensitivity, 85.39% recall, 87.36% precision, and 86.36% F1-measure on the CelebDF-FaceForencics++ (c23) merged dataset. The experimental study confirms the superiority of the presented method as compared to the state-of-the-art methods.

Detection of fake-video uploaders on social media using Naive Bayesian model with social cues

With the rapid development of the Internet, the wide circulation of disinformation has considerably disrupted the search and recognition of information. Despite intensive research devoted to fake text detection, studies on fake short videos that inundate the Internet are rare. Fake videos, because of their quick transmission and broad reach, can increase misunderstanding, impact decision-making, and lead to irrevocable losses. Therefore, it is important to detect fake videos that mislead users on the Internet. Since it is difficult to detect fake videos directly, we probed the detection of fake video uploaders in this study with a vision to provide a basis for the detection of fake videos. Specifically, a dataset consisting of 450 uploaders of videos on diabetes and traditional Chinese medicine was constructed, five features of the fake video uploaders were proposed, and a Naive Bayesian model was built. Through experiments, the optimal feature combination was identified, and the proposed model reached a maximum accuracy of 70.7%.

A Novel Defensive Strategy for Facial Manipulation Detection Combining Bilateral Filtering and Joint Adversarial Training

Facial manipulation enables facial expressions to be tampered with or facial identities to be replaced in videos. The fake videos are so realistic that they are even difficult for human eyes to distinguish. This poses a great threat to social and public information security. A number of facial manipulation detectors have been proposed to address this threat. However, previous studies have shown that the accuracy of these detectors is sensitive to adversarial examples. The existing defense methods are very limited in terms of applicable scenes and defense effects. This paper proposes a new defense strategy for facial manipulation detectors, which combines a passive defense method, bilateral filtering, and a proactive defense method, joint adversarial training, to mitigate the vulnerability of facial manipulation detectors against adversarial examples. The bilateral filtering method is applied in the preprocessing stage of the model without any modification to denoise the input adversarial examples. The joint adversarial training starts from the training stage of the model, which mixes various adversarial examples and original examples to train the model. The introduction of joint adversarial training can train a model that defends against multiple adversarial attacks. The experimental results show that the proposed defense strategy positively helps facial manipulation detectors counter adversarial examples.

DeepFake Detection using 3D-Xception Net with Discrete Fourier Transformation

The videos are more popular for sharing content on social media to capture the audience’s attention. The artificial manipulation of videos is growing rapidly to make the videos flashy and interesting but they can easily misuse to spread false information on social media platforms. Deep Fake is a problematic method for the manipulation of videos in which artificial components are added to the video using emerging deep learning techniques. Due to the increase in the accuracy of deep fake generation methods, artificially created videos are no longer detectable and pose a major threat to social media users. To address this growing problem, we have proposed a new method for detecting deep fake videos using 3D Inflated Xception Net with Discrete Fourier Transformation. Xception Net was originally designed for application on 2D images only. The proposed method is the first attempt to use a 3D Xception Net for categorizing video-based data. The advantage of the proposed method is, it works on the whole video rather than the subset of frames while categorizing. Our proposed model was tested on the popular dataset Celeb-DF and achieved better accuracy.

A Fight Against the Dissemination of Deepfakes in Other Countries: Criminal and Criminological Aspects

The authors analyze a relatively new phenomenon of spreading realistic audiovisual fake materials (deepfakes). This socially dangerous phenomenon is not reflected in the Russian criminal legislation as a separate offence. At the same time, some countries have started developing a criminal policy in this sphere. The methodology of the study presupposes a comparative law analysis of current legislations of the USA, China and the European union regarding the liability for the dissemination of realistic audiovisual fakes. The analysis of criminal legislation is aimed at the identification and systematization of key approaches to criminalizing the dissemination of realistic audiovisual fakes in the countries that are the leaders in digitizing their social and economic life. It showed that there are radically different approaches to regulating criminal liability for the actions under consideration. The authors analyzed criminal policy of the United States at the federal and state levels on the criminal law protection against infringements through deepfakes. They found that the first action to be recognized as criminal is the use of realistic audiovisual fakes for electoral intervention. The legislations of some states strictly regulate the procedure of posting such content before elections, the most serious violations leading to criminal liability. Besides, the United States recognizes as criminally punishable the use of deepfakes for creating materials of intimate nature and for identity theft. The People’s Republic of China establishes liability, including criminal liability, for posting any fake realistic-looking audiovisual materials without mentioning that they are fake. Currently there are no special criminal law norms regulating liability for the dissemination of deepfakes in the law of the European Union. This action should be viewed as infringement of the lawful use of personal data. The authors give their assessment of some criminological characteristics of the analyzed publicly dangerous phenomenon in Russia and in the world. In spite of the relative novelty of the deepfake technology, realistic fake videos are quite common. The society supports the necessity of criminalizing this publicly dangerous action.

Forensic Verification and Detection of Fake Video using Deep Fake Algorithm

In the course of the most recen years, the ascent in cell phones and interpersonal organizations has made computerized pictures and recordings basic advanced articles. per reports, right around two billion pictures are transferred every day on the web. This gigantic utilization of computerized pictures has been trailed by an increment of methods to change picture substance, utilizing altering programming like Photoshop for instance. Counterfeit recordings and pictures made by deepFake methods turned into a decent open issue as of late. These days a few procedures for facial control in recordings are effectively evolved like FaceSwap, deepFake, and so on On one side, this innovative progression increment degree to new regions (e.g., film making, special visualization, visual expressions, and so on) On the contrary side, repudiating, it likewise expands the advantage inside the age of video frauds by malignant clients. In this manner by utilizing profound learning strategies we can distinguish the video is phony or not. to recognize these malevolent pictures, we are visiting foster a framework which will naturally identify and survey the trustworthiness of advanced visual media is in this way crucial. Deepfake could be a procedure for human picture union upheld AI, i.e., to superimpose the predominant (source) pictures or recordings onto objective pictures or recordings utilizing neural organizations (NNs). Deepfake aficionados are utilizing NNs to give persuading face trades. Deepfakes are a sort of video or picture imitation created to spread deception, attack protection, and veil the truth utilizing cutting edge innovations like prepared calculations, profound learning applications, and figuring. they need become an irritation to online media clients by distributing counterfeit recordings made by melding a big name's face over a precise video. The effect of deepFakes is disturbing, with lawmakers, senior corporate officials, and world pioneers being focused by loathsome entertainers. A way to deal with distinguish deepFake recordings of legislators utilizing transient consecutive edges is proposed. The proposed approach utilizes the strong video to separate the edges at the essential level followed by a profound profundity based convolutional long memory model to recognize the phony casings at the subsequent level. Additionally, the proposed model is assessed on our recently gathered ground truth dataset of produced recordings utilizing source and objective video edges of renowned lawmakers. Trial results exhibit the viability of our strategy.

Countering Spoof: Towards Detecting Deepfake with Multidimensional Biological Signals

The deepfake technology is conveniently abused with the low technology threshold, which may bring the huge social security risks. As GAN-based synthesis technology is becoming stronger, various methods are difficult to classify the fake content effectively. However, although the fake content generated by GANs can deceive the human eyes, it ignores the biological signals hidden in the face video. In this paper, we proposed a novel video forensics method with multidimensional biological signals, which extracting the difference of the biological signal between real and fake videos from three dimensions. The experimental results show that our method achieves 98% accuracy on the main public dataset. Compared with other technologies, the proposed method only extracts fake video information and is not limited to a specific generation method, so it is not affected by synthetic methods and has good adaptability.

Detecting and Locating Passive Video Forgery Based on Low Computational Complexity Third-Order Tensor Representation.

Great attention is paid to detecting video forgeries nowadays, especially with the widespread sharing of videos over social media and websites. Many video editing software programs are available and perform well in tampering with video contents or even creating fake videos. Forgery affects video integrity and authenticity and has serious implications. For example, digital videos for security and surveillance purposes are used as evidence in courts. In this paper, a newly developed passive video forgery scheme is introduced and discussed. The developed scheme is based on representing highly correlated video data with a low computational complexity third-order tensor tube-fiber mode. An arbitrary number of core tensors is selected to detect and locate two serious types of forgeries which are: insertion and deletion. These tensor data are orthogonally transformed to achieve more data reductions and to provide good features to trace forgery along the whole video. Experimental results and comparisons show the superiority of the proposed scheme with a precision value of up to 99% in detecting and locating both types of attacks for static as well as dynamic videos, quick-moving foreground items (single or multiple), zooming in and zooming out datasets which are rarely tested by previous works. Moreover, the proposed scheme offers a reduction in time and a linear computational complexity. Based on the used computer’s configurations, an average time of 35 s. is needed to detect and locate 40 forged frames out of 300 frames.