STM Journals Peer-reviewed research from Consortium e-Learning Network

Journal Of Network security

Deepfake Detection Using Resnext50

  1. Vaishali Jadhav
  2. Taabish Sutriwala
  3. Raunak Singh
  4. Sohel Siraj Mukadam

Abstract

Due to the recent advancements in Artificial Intelligence (AI), technologies it has led to the rapid development in the modification of audio, video, and image manipulation techniques. This AIgenerated media content is referred to as "Deep Fakes". AI-based tools can exploit and manipulate media in increasingly believable ways like super-imposing a person’s face on another person’s body or creating copies of a public person’s voice. A range of papers has discovered significant signs of deep fakes, including unnatural blinking patterns, inconsistent head poses, distortion in facial features, incongruities between the speech and mouth movements, and even learning to note the absence. Deepfake detection solutions use multimodal detection techniques to determine whether target media has been manipulated or generated artificially. Existing detection methods are classified into two types: manual and algorithmic methods. Human media forensic practitioners, often armed with software tools, use manual techniques. Algorithmic detection identifies manipulated media using an AI-based algorithm. Our Deepfake detection system was created to identify deep fakes and manipulations in images and videos.

Keywords

Support