False Media Detection by Using Deep Learning
Renu Dalal, Manju Khari, Archit Garg, Dheeraj Gupta, Anubhav Gautam · 2021
Peoples meet across numerous images and videos in everyday life, and most of them are digitally manipulated in some way to some extent. Most of these images are subjected to simple Instagram effects or have undergone photoshop to completely transform the original picture. Since humans are not good enough to recognize the originality of these morphed images so they might be exposed to wrong messages or content which may be disastrous at individual, community level in some aspect. These morphed contents are called deep fakes. The illegal use of these deep fakes is prevalent worldwide to get political advantage. So, the main objective of this chapter is to detect any kind of transformation/effects in an image that creates questions about the originality of its content. As for most of the human pictures, videos and images are trustworthy sources of information and at the same time the internet is flooded with morphed images and videos, created by face re-enactment algorithms. The proposed detection algorithm is efficient enough and has a good degree of accuracy to detect any kind of transformation in a given picture or video.