Detection of Machine Generated Multimedia Elements Using Deep Learning

With recent advancements in Deep Learning [DL], many useful tasks like text recognition, speech recognition and many more tasks can be easily done using machines. Such advancements have led to the development of deepfakes (coined from combining the words "Deep learning" and "fake"), face swaps, facial motion transfer, and other such media content. Deepfakes are synthetically generated media in which a person in an image or video is replaced with someone else’s likeness. Deepfakes and other such content were intended to help in the field of cinematography, conceal the identity of the witness, and many such positive uses, but they can easily be used to spread false news, spread fake propaganda, and spread hate. There has come a time now that the ill uses of such technologies heavily outweigh the positive uses. With the passing time, deepfakes are becoming more and more realistic and it is becoming increasingly difficult to identify such media from real ones. This paper intends to present a system that would make its users to distinguish such media from authentic media. The user can simply select the image or video file which they would like to authenticate, and the system would predict them to be doctored or pristine.