DeepFake

A spare-time research project I started for my interest, challenge and experience - experimenting with auto-encoders, GANs, etc. to transfer gestures and maybe even voice between source and target videos.

Project Roadmap

(last update: 2019/10/15)

1. (100%) developing a script to extract frames from a video.

2. (100%) developing a simple auto-encoder for images (see notebook).

3. (10%) developing a variational auto-encoder for images.

4. (10%) developing a GAN for images.

5. (0%) developing a visual DeepFake with crossed auto-encoders sharing the same encoder.

6. (0%) developing a visual DeepFake with crossed GAN-based auto-encoders.

7. (0%) developing a script to extract voice from a video.

8. (0%) developing a simple auto-encoder for voice.

9. (0%) developing a voice DeepFake with crossed auto-encoders sharing the same encoder.

10. (0%) combining everything, creating a complete DeepFake video.

---

I would be glad for any comment, question or discussion! You can contact me privately at oz.livneh@gmail.com or www.linkedin.com/in/oz-livneh.

Yours,

Oz Livneh