Year: 2025
Type: article
Abstract: The rising incidents of video forgery in the digital space, stemming from breaches in information security, have led to a pressing need for monitoring visual and audio content to document such forgeries. The proliferation of counterfeit films heightens the potential for disorder and security threats. With the ease of posting, downloading, and sharing multimedia files online—encompassing audio, images, and videos—the increase in viruses contributes significantly to the prevalence of video forgeries. Recent technological advancements have facilitated mass media manipulation and made the production of false information more straightforward. The integrity of media is under severe threat due to the creation and wide distribution of deepfake content on social media, and identifying such content is believed to be challenging. A method for detecting deepfakes has been introduced to identify these video forgeries. To pinpoint deepfake videos, a Convolutional Neural Network (CNN) technique known as ResNet is utilized. The objective of this model is to enhance both the performance and accuracy of the detector in recognizing videos that have been modified using a specific method. The suggested approach extracts deep features through the ResNet CNN algorithm and then employs fundamental mathematical processes.
Source: Utilitas Mathematica
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7216926580
