DETECTING FAKE VIDEOS

Abstract

As the World Wide Web usage continues to grow, people all over the world are relying more and more everyday on it in different ways like social networking, making online payments, entertainment purposes like watching videos and content sharing, educational purposes as well as professional uses too. One such common use is watching videos on the web or having update feeds in form of videos from social networking websites. This work is an effort towards helping the users in identifying the content in the videos as fake or real. Thus the user is alerted from believing false information which might lead to unwanted outcomes for the user like money loss for instance if the video was regarding share market or identifying rumors circulating on web. For the above stated aim, an application using Python has been developed. The application follows supervised learning with a training data set of 574 videos having fake as well as real videos. The technique used is taking into consideration the audio component of the video in addition to the video component. Also, the accuracy percentage of the subsequent test results using LSTM, CNN and Naïve Bayes model is displayed.