An Enhanced Deep Learning-Based DeepFake Video Detection and Classification System

Abstract

The privacy of individuals and entire countries is currently threatened by the widespread use of face-swapping DeepFake models, which result in a sizable number of fake videos that seem extraordinarily genuine. Because DeepFake production tools have advanced so much and since so many researchers and businesses are interested in testing their limits, fake media is spreading like wildfire over the internet. Therefore, this study proposes five-layered convolutional neural networks (CNNs) for a DeepFake detection and classification model. The CNN enhanced with ReLU is used to extract features from these faces once the model has extracted the face region from video frames. To guarantee model accuracy while maintaining a suitable weight, a CNN enabled with ReLU model was used for the DeepFake-detection-influenced video. The performance evaluation of the proposed model was tested using Face2Face, and first-order motion DeepFake datasets. Experimental results revealed that the proposed model has an average prediction rate of 98% for DeepFake videos and 95% for Face2Face videos under actual network diffusion circumstances. When compared with systems such as Meso4, MesoInception4, Xception, EfficientNet-B0, and VGG16 which utilizes the convolutional neural network, the suggested model produced the best results with an accuracy rate of 86%.

A review of deep learning‐based approaches for deepfake content detection

“…(Awotunde et al, 2023) addresses the considerable increase in fake videos appearing genuine thanks to advances in deepfake production tools. This investigation suggests five‐layer CNNs for a DeepFake detection and classification model.…”

“…combined a CNN model with a Vision Transformer (ViT) architecture to detect videos with evidence of face manipulation. The authors rely on the VGG-16 CNN model for feature extraction from the video frames and the ViT model on such feature maps to classify the video as real or fake, obtaining significant results over the DFDC dataset (Awotunde et al, 2023). addresses the considerable increase in fake videos appearing genuine thanks to advances in deepfake production tools.…”

Deep Fake Video Detection

“…The research methodology involves the use of a substantial manipulated face dataset, MANFA, testing a customized deep learning model, XGB-MANFA, and achieving state-of-theart performance with an AUC value of 93.4%. The proposed model showcases superiorities over existing systems, emphasizing high [18] performance, flexibility, robustness, and its ability to operate without specialized tools or expert knowledge. Table 1 shows the Methodology Overview Table . The study addresses the rising concern of facial manipulation in videos, proposing a network architecture utilizing five (CNNs) and (RNN) to effectively detect manipulations with low computational cost.…”

Transfer Learning-Based Models for Comparative Evaluation for the Detection of AI-Generated Images

“…The publication also advances the field by releasing a curated dataset for more research in this area and insights into the features that these models learn during the detection process. Figure 1 shows the Classification layer combination of the proposed model in [6]. Figure 2 shows the Face morphing attack detection proposed in [7], The authors suggest a brand-new methodology that fuses progressive enhancement learning with high-frequency feature analysis.…”