Fully Unsupervised Deepfake Video Detection Via Enhanced Contrastive Learning.

Nowadays, Deepfake videos are widely spread over the Internet, which severely impairs the public trustworthiness and social security. Although more and more reliable detectors have recently sprung up for resisting against that new-emerging tampering technique, some challengeable issues still need to be addressed, such that most of Deepfake video detectors under the framework of the supervised mechanism require a large scale of samples with accurate labels for training. When the amount of the training samples with the true labels are not enough or the training data are maliciously poisoned by adversaries, the supervised classifier is probably not reliable for detection. To tackle that tough issue, it is proposed to design a fully unsupervised Deepfake detector. In particular, in the whole procedure of training or testing, we have no idea of any information about the true labels of samples. First, we novelly design a pseudo-label generator for labeling the training samples, where the traditional hand-crafted features are used to characterize both types of samples. Second, the training samples with the pseudo-labels are fed into the proposed enhanced contrastive learner, in which the discriminative features are further extracted and continually refined by iteration on the guidance of the contrastive loss. Last, relying on the inter-frame correlation, we complete the final binary classification between real and fake videos. A large scale of experimental results empirically verify the effectiveness of our proposed unsupervised Deepfake detector on the benchmark datasets including FF++, Celeb-DF, DFD, DFDC, and UADFV. Furthermore, our proposed well-performed detector is superior to the current unsupervised method, and comparable to the baseline supervised methods. More importantly, when facing the problem of the labeled data poisoned by malicious adversaries or insufficient data for training, our proposed unsupervised Deepfake detector performs its powerful superiority.

Statistical model of quantized DCT coefficients: application in the steganalysis of Jsteg algorithm.

The goal of this paper is to propose a statistical model of quantized discrete cosine transform (DCT) coefficients. It relies on a mathematical framework of studying the image processing pipeline of a typical digital camera instead of fitting empirical data with a variety of popular models proposed in this paper. To highlight the accuracy of the proposed model, this paper exploits it for the detection of hidden information in JPEG images. By formulating the hidden data detection as a hypothesis testing, this paper studies the most powerful likelihood ratio test for the steganalysis of Jsteg algorithm and establishes theoretically its statistical performance. Based on the proposed model of DCT coefficients, a maximum likelihood estimator for embedding rate is also designed. Numerical results on simulated and real images emphasize the accuracy of the proposed model and the performance of the proposed test.

Camera model identification based on the heteroscedastic noise model.

The goal of this paper is to design a statistical test for the camera model identification problem. The approach is based on the heteroscedastic noise model, which more accurately describes a natural raw image. This model is characterized by only two parameters, which are considered as unique fingerprint to identify camera models. The camera model identification problem is cast in the framework of hypothesis testing theory. In an ideal context where all model parameters are perfectly known, the likelihood ratio test (LRT) is presented and its performances are theoretically established. For a practical use, two generalized LRTs are designed to deal with unknown model parameters so that they can meet a prescribed false alarm probability while ensuring a high detection performance. Numerical results on simulated images and real natural raw images highlight the relevance of the proposed approach.

Epsilon-optimal non-Bayesian anomaly detection for parametric tomography.

The non-Bayesian detection of an anomaly from a single or a few noisy tomographic projections is considered as a statistical hypotheses testing problem. It is supposed that a radiography is composed of an imaged nonanomalous background medium, considered as a deterministic nuisance parameter, with a possibly hidden anomaly. Because the full voxel-by-voxel reconstruction is impossible, an original tomographic method based on the parametric models of the nonanomalous background medium and radiographic process is proposed to fill up the gap in the missing data. Exploiting this "parametric tomography," a new detection scheme with a limited loss of optimality is proposed as an alternative to the nonlinear generalized likelihood ratio test, which is untractable in the context of nondestructive testing for the objects with uncertainties in their physical/geometrical properties. The theoretical results are illustrated by the processing of real radiographies for the nuclear fuel rod inspection.