Fake face image detection using feature network

Abstract

In the recent times, the image data in social networks such as Instagram, Whatsapp, Facebook, Snapchat, twitter etc has an exponential growth in terms of volume, variety due to the velocity of the data stream. On the other hand, the advancements in the image and video processing led to increase in the fake images relatively in huge volumes. Due to the involvement in spreading fake news and leading mob incitements fake images became major concern to handle that demands an efficient a fake image detector is of at most concern to entire social networking organizations. In this paper, a deep learning framework is proposed that differentiates fabricated parts of the image from the real image using supervised learning strategies. Also a modified neural network structure called the Fake Feature Network is proposed in this work which consists of advanced convolution networks. In order to make model effective, the proposed methodology has a two major steps in learning which combines a modified neural structure that uses classifier and pairwise learning for the fake image detection.