Abstract
In recent years, the introduction of deep learning-based methods and huge reductions in computational cost have made it possible to create hyper realistic video of humans and beyond, usually referred to as DeepFakes. The availability of open-source tools for creating DeepFakes presents a challenge to the credibility of online media especially after online social media platforms also being the source of news for the majority. In this thesis, we create an open-source web platform called DeepSafe, which incorporates cutting-edge DeepFake detection algorithms and provides a user-friendly interface. DeepSafe provides APIs and makes adding new better detection algorithms to the platform possible. In this thesis, we discuss the design and operation of the DeepSafe and the experiments which led to design it as a webapp. The thesis also discusses potential solutions to media credibility beyond DeepFakes and how block chain can be used to keep a check on media authenticity and ownership. All the source code files are open-source and can be accessed free of charge here https://github.com/siddharthksah/DeepSafe