Streamlit Deep Fake Generator

Inspiration

This project was inspired by a previous deep fake generator that I had. I wanted to build one on Streamlit so it has more user accessibility

What it does

To put it simply, it takes an input image and video and returns a deep fake of the image using the inputted video as a template

How I built it

I built it using the First Order Motion Model from https://github.com/AliaksandrSiarohin/first-order-model with minor modifications to allow it to work with only a CPU and without a Graphics Processor with CUDA cores so that it can be run on streamlit. This was then integrated into Streamlit for effective user interfacing.

Challenges I ran into

Because of the program's complexity, it cannot run on streamlit because of the required computing resources. Because of this, for this project to work, it must be run on a local machine. If the project had more computing resources, then this issue would not be present.

Accomplishments that I am proud of

I am proud that I was able to finish this project on my own and also being a first-year university student. This project was quite complex and I am quite proud of the fact that I was able to finish it.

Built With

• python
• streamlit