DeepFake Image/video Detection

I want to try to make a deepfake classifier. It is my first time doing practical deep learning and I am really struggled to start. Generally, I have the following questions.
Let’s say Im given a set of deepfake images/videos.

  1. What are the differences in handling such data types?
  2. Is there any sources for powerful(high accuracy) DeepFake Image/video classifier?
  3. What existing (e.g. CNN) model is good at handling such classification?
  4. Generally, what approach will be used in making such a classifier?

Thank you.

I believe detecting deepfakes with Generative Adversarial Networks (GAN) is possible. Maybe the following course can help you with that:

1 Like