Deepfake video detection: YOLO-Face convolution recurrent approach

Evaluation of deepfake detection using YOLO with local binary pattern histogram

Recently, deepfake technology has become a popularly used technique for swapping faces in images or videos that create forged data to mislead society. Detecting the originality of the video is a critical process due to the negative pattern of the image. In the detection of forged images or videos, various image processing techniques were implemented. Existing methods are ineffective in detecting new threats or false images. This article has proposed You Only Look Once-Local Binary Pattern Histogram (YOLO-LBPH) to detect fake videos. YOLO is used to detect the face in an image or a frame of a video. The spatial features are extracted from the face image using a EfficientNet-B5 method. Spatial feature extractions are fed as input in the Local Binary Pattern Histogram to extract temporal features. The proposed YOLO-LBPH is implemented using the large scale deepfake forensics (DF) dataset known as CelebDF-FaceForensics++(c23), which is a combination of FaceForensics++(c23) and Celeb-DF. As a result, the precision score is 86.88% in the CelebDF-FaceForensics++(c23) dataset, 88.9% in the DFFD dataset, 91.35% in the CASIA-WebFace data. Similarly, the recall is 92.45% in the Celeb-DF-Face Forensics ++(c23) dataset, 93.76% in the DFFD dataset, and 94.35% in the CASIA-Web Face dataset.

An integrated spatiotemporal-based methodology for deepfake detection

Rapid advances in deep learning models have made it easier for public and crackers to generate hyper-realistic deepfake videos in which faces are swapped. Such deepfake videos may constitute a significant threat to the world if they are misused to blackmail public figures and to deceive systems of face recognition. As a result, distinguishing these fake videos from real ones has become fundamental. This paper introduces a new deepfake video detection method. You Only Look Once (YOLO) face detector is used to detect faces from video frames. A proposed hybrid method based on proposing two different feature extraction methods is applied to these faces. The first feature extraction method, a proposed Convolution Neural Network (CNN), is based on the Histogram of Oriented Gradient (HOG) method. The second one is an ameliorated XceptionNet CNN. The two extracted sets of features are merged together and fed as input to a sequence of Gated Recurrent Units (GRUs) to extract the spatial and temporal features and then individuate the authenticity of videos. The proposed method is trained on the CelebDF-FaceForencics++ (c23) dataset and evaluated on the CelebDF test set. The experimental results and analysis confirm the superiority of the suggested method over the state-of-the-art methods.