Paper Title
Deep Fake Video Detection using Deep Learning

Latest advances and improvements in machine learning algorithms have contributed to the evolution of excellent quality manipulated images which forms video frames, and has a lot of resemblance with the real images which forms video frames. This can have a fatal impact on the way one perceives digitally available knowledge or facts. Current advances in machine learning and camera function have made it possible for images which forms video frames and audio to be manipulated convincingly. These deep-fake videos differ from the real videos via either replacement of audio or the mouth movement or the place where the video was recorded (background). Detection of profound falsehoods with just a little spatial and temporal distortion is particularly difficult. These falsehoods in the distorted videos have now arisen as a threat to humanity and can even serve as war tools in the new digital age. There have been many useful methods and automation procedures for the detection of such deep fake videos. Pose estimation, Facial artifacts, Temporal Pattern Analysis, Background comparison, Eye blinking and Mesoscopic Analysis are some of the techniques used by researchers. We aim to provide a descriptive review of these classical deep Fake detection methods in a distinctive statistical study and hence facilitate the creation and enhancement of a new and a way better approach to handle such deepFake videos through a CNN architecture based on the extreme Inception or Xception model. Keywords - Deep fake, Neural Networks, Xception