Deep Learning Technique for Recognition of De...

Deep Learning Technique for Recognition of Deep Fake Videos

Author Name : Fahad Mira

ABSTRACT

New computer methods and digital content production have been made possible by recent advancements in digital media technology. They have also aided in the development of contemporary AI-based breakthroughs and offer simple tools for creating incredibly lifelike video modifications. These "Deep Fakes" or fake videos might pose a serious danger to the public's impression of a case or the public's opinion of society as a whole. The effects of these videos serving as accurate representations are substantial, especially in the propagation of false news. However, these kinds of fraudulent videos can be produced by software manipulation. In cybersecurity, deep fake detection is crucial because it aids in data protection, the identification of deep fakes, and media manipulation. Therefore, it is crucial and necessary for a person to be able to spot this type of erroneous knowledge. The purpose of this research is to review current literature on the topic of deep learning algorithms for deep fake video recognition in order to identify the most effective new approaches. It evaluated data from two investigations and produced the following recommendations: using (CNN) and (LSTM) to distinguish between fake and real video frames. It recommended employing these methods of detection as well as using the new approach for identifying deepfakesthat uses the YOLO face detector to identify facial video frames(YOLO-CNN-XGBoost),and recommends searching other new ways of detection.

Key Word: Deep Learning Technique, Deep Fake videos.