Deepfake detector

Deepfake can create realistic AI-generated videos show real people doing and saying fictional things. Can we build a tool to detect such the fake videos?
Master

“Deepfake” techniques, which present realistic AI-generated videos of real people doing and saying fictional things, have significant implications for determining the legitimacy of information presented online. With the constant development of Deepfake technologies the society need new ways of detecting and preventing media manipulated via AI from being used to mislead others.

“Manipulated media being put out on the internet, to create bogus conspiracy theories and to manipulate people for political gain, is becoming an issue of global importance, as it is a fundamental threat to democracy, and hence freedom. I believe we urgently need new tools to detect and characterize this misinformation, so I am happy to be part of an initiative that seeks to mobilize the research community around these goals — both to preserve the truth whilst pushing the frontiers of science.” — Professor Philip H. S. Torr, Department of Engineering Science, University of Oxford

Goal

The goal of the thesis is to produce an algorithm and technology that can be used to detect when AI has been used to alter a video in order to mislead the viewer. The developed algorithm will be evaluated via participation in upcoming The Deepfake Detection Challenge.

Learning outcome

During the course of this master thesis, the candidate will gain in-depth knowledge of Machine and Deep learning, Deepfake manipulation algorithms and Deepfake detection methods.

Qualifications

Python programming is required. Knowledge about machine and deep learning is an advantage.

Supervisors

  • Carsten Griwodz
  • Johannes Langguth
  • Konstantin Pogorelov

References

ai.facebook.com/blog/deepfake-detection-challenge
deepfakedetectionchallenge.ai/