A Comprehensive Review on Fake Images/Videos Detection Techniques

Now that image creation and manipulation have advanced so quickly, there are serious questions about how this may affect society. At best, this leads to loss of trust in digital content. There are many existing algorithms such as Naive Bayes, CNN, RNN, Robust Hashing, GANs, SVM etc. which are being used for the detection of fake videos. Making and classifying deep fakes using Deep Neural Networks (DNN) nowadays have increased the interest of researchers in this field. Deep Fake is the regenerated media that is attained by edging in or replacing some information within the DNN model. In this work, survey withdrawn by various research groups focused the feasibility loopholes that need to be recovered for deep fakes. The use of above-mentioned techniques has been increased by a significant percentage in video game industries and cinema like enhancing visual stuff in pictures. In this paper, different types of datasets used by authors and various contemporary techniques used for fake image/video detection are described. Finally, various research gaps and the possible future directions are highlighted.