CleanCrap

Description

The CleanCrap project will allow you to remove an object from the video in a couple of clicks.
To do this, you do not need to understand complex Adobe programs.

Work in Progress

• Come up with a design
• Make a landing page
• Adapt the neural network to all video sizes
• Register domain
• Upload a neural network to the cloud

Further improvements

• Create an editor to highlight the mask
• Transfer calculations to GPU
• Improve the speed of passing a sliding window in a neural network
• Fix the blurring of objects
• Learn how to remove shadows from objects

How to use it?

Run setup.py then run streamlit run app.py. Make sure that you have all the dependencies from the file installed requirements.txt. Or you can build Docker image on your system with OS Linux and installed Docker in the following way:

1. git clone https://github.com/Maxsmile123/CleanCrap.git
2. Add to current directory file Dockerfile from this repository
3. Build with command sudo docker build -t clean_crap:001 . and run container with sudo docker run --rm -p 8501:8501 clean_crap:001

Our advantages

Authors

This project was completed by the ASSS.

Note

The project does not use its own neural network technologies.
We used a ready-made neural network of our comrades, who allowed us to use it for non-commercial projects.