Detection of Manipulated Images/Videos using AI
Description
Digital image and video manipulation could potentially mislead social media for immoral purposes, promoting a propaganda by creating fake news and highly convincing disinformation. For example, in 2017, deepfakes drew the attention of the public as a user created pornographic videos by replacing the original faces with those of celebrities and posted them in the Internet. That raised awareness of the potential social issues associated with impersonating influencers, including politicians, celebrities, and other public figures. To make things worse, it was shown that it is difficult for humans to distinguish authentic from synthesized images like faces.
Therefore, it is of great importance to develop methods to detect synthesized images. In this project, we aim at investigating the differences between authentic and synthesized images and videos, in order to create effective algorithms that are suitable to identify synthesized images in the real life.