A Comparative Evaluation of Local Feature Descriptors for DeepFakes Detection

Akhtar, Zahid; Dasgupta, Dipankar

doi:10.1109/hst47167.2019.9033005

Cited by 31 publications

(31 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(2) From the experiments, it was seen that the best average accuracy score was obtained for BGP local descriptors and the best accuracy score was obtained with PHOG local features and Quadratic SVM classifier. (3) As LBP, CENTRIST and FDLBP have similar conceptual structures, their achievements were quite close [26]. (4) The comparison of the proposed method with the state-of-the-art methods revealed that the proposed method has the potential in the use of various ECG based disease detection applications.…”

Section: Discussionmentioning

confidence: 77%

Local feature descriptors based ECG beat classification

Abdullah

Akpınar

Şengür

2020

Health Inf Sci Syst

View full text Add to dashboard Cite

ECG beat type analysis is important in the detection of various heart diseases. The ECG beats give useful information about the status of the monitored heart condition. Up to now, various artificial intelligence-based methods have been proposed for ECG based heart failure detection. These methods were generally based on either time or frequency domain signal processing routines. In this study, we propose a different approach for ECG beat classification. The proposed approach is based on image processing. Thus, the initial step of the proposed work is converting the ECG beat signals to the ECG beat images. To do that, the ECG beat snapshots are initially saved as ECG beat images and then local feature descriptors are considered for feature extraction from ECG beat images. Eight local feature descriptors namely Local Binary Patterns, Frequency Decoded LBP, Quaternionic Local Ranking Binary Pattern, Binary Gabor Pattern, Local Phase Quantization, Binarized Statistical Image Features, CENsus TRansform hISTogram and Pyramid Histogram of Oriented Gradients are considered for feature extraction. The Support Vector Machines (SVM) classifier is used in the classification stage of the study. Linear, Quadratic, Cubic and Gaussian kernel functions are used in the SVM classifier. Five types of ECG beats from the MIT-BIH arrhythmia dataset are considered in experiments and the classification accuracy is used for performance measure. To construct a balanced training and test sets, 5000 and 10,000 ECG beat samples are randomly selected and are used in experiments in tenfold cross-validation fashion. The obtained resultsshow that the proposed method is quite efficient where the calculated accuracy score is 99.9% and the comparisons with the state-of-the-art method show that the proposed method outperforms other methods.

show abstract

Section: Discussionmentioning

confidence: 77%

Local feature descriptors based ECG beat classification

Abdullah

Akpınar

Şengür

2020

Health Inf Sci Syst

View full text Add to dashboard Cite

show abstract

“…As with other technologies, the same algorithms used for creating deepfakes could have a beneficial application in the field of psychology, building digital synthetic identities for voiceless users; or in robot sketches through advanced facial recognition for law enforcement, for example (Akhtar & Dasgupta, 2019;Zhu, Fang, Sui, & Li, 2020). Notwithstanding, its use seems to be more harmful than beneficial nowadays with examples of the use of these technologies in acts of fraud and crime (Stupp, 2019).…”

Section: Deepfake: a Novel Form Of Fake Newsmentioning

confidence: 99%

Fighting Deepfakes: Media and Internet Giants’ Converging and Diverging Strategies Against Hi-Tech Misinformation

Vizoso

Vaz-Álvarez

García

2021

MaC

View full text Add to dashboard Cite

Deepfakes, one of the most novel forms of misinformation, have become a real challenge in the communicative environment due to their spread through online news and social media spaces. Although fake news have existed for centuries, its circulation is now more harmful than ever before, thanks to the ease of its production and dissemination. At this juncture, technological development has led to the emergence of deepfakes, doctored videos, audios or photos that use artificial intelligence. Since its inception in 2017, the tools and algorithms that enable the modification of faces and sounds in audiovisual content have evolved to the point where there are mobile apps and web services that allow average users its manipulation. This research tries to show how three renowned media outlets—The Wall Street Journal, The Washington Post, and Reuters—and three of the biggest Internet-based companies—Google, Facebook, and Twitter—are dealing with the spread of this new form of fake news. Results show that identification of deepfakes is a common practice for both types of organizations. However, while the media is focused on training journalists for its detection, online platforms tended to fund research projects whose objective is to develop or improve media forensics tools.

show abstract

“…Akhtar et al. [4] selected 10 traditional local feature descriptors to describe image features, and used SVM to build a classification model.…”

Section: Related Workmentioning

confidence: 99%

“…Thus, the application range of these single‐type forgery detection models is greatly restricted. What is more, most models [4, 12–14] achieve high accuracy on rough and low‐quality (LQ) forged databases, while the detection of high‐quality (HQ) forged contents becomes particularly difficult. The target of this research is to build a deep forgery discriminator with better comprehensive performance.…”

Section: Introductionmentioning

confidence: 99%

Deep forgery discriminator via image degradation analysis

Zhang

et al. 2021

IET Image Processing

View full text Add to dashboard Cite

Generative adversarial network-based deep generative model is widely applied in creating hyper-realistic face-swapping images and videos. However, its malicious use has posed a great threat to online contents, thus making detecting the authenticity of images and videos a tricky task. Most of the existing detection methods are only suitable for one type of forgery and only work for low-quality tampered images, restricting their applications. This paper concerns the construction of a novel discriminator with better comprehensive capabilities. Through analysis of the visual characteristics of manipulated images from the perspective of image quality, it is revealed that the synthesized face does have different degrees of quality degradation compared to the source content. Therefore, several kinds of image quality-related handicraft features are extracted, including texture, sharpness, frequency domain features, and deep features, to unveil the inconsistent information and modification traces in the fake faces. In this way, a 1065-dimensional vector of each image is obtained through multi-feature fusion, and it is then fed into RF to train a targeted binary classification detector. Extensive experiments have shown that the proposed scheme is superior to the previous methods in recognition accuracy on multiple manipulation databases including the Celeb-DF database with better visual quality.

show abstract

A Comparative Evaluation of Local Feature Descriptors for DeepFakes Detection

Cited by 31 publications

References 27 publications

Local feature descriptors based ECG beat classification

Local feature descriptors based ECG beat classification

Fighting Deepfakes: Media and Internet Giants’ Converging and Diverging Strategies Against Hi-Tech Misinformation

Deep forgery discriminator via image degradation analysis

Contact Info

Product

Resources

About