kennbroorg/shit-bucket/fecalfaceThis project (PoC for now, and part of Shit Bucket) involves face detection, face recognition and adversarial input to protect avatars
Description | Installation | Website | Issues
This project (PoC for now) involves :
This project is part of Shit Bucket tools
It collects the avatar from a RRSS, tries to determine the position of a face if it exists and then attacks it in order to make the final comparison between the original face and the attacked one not recognizable by an AI, but for human eye will look similar.
This project will not cause our faces to be confused with panda bear's one. And at this moment the project doesn't work with avatars with more than one face.
The attack process is very processing intensive, taking between 5 and 7 minutes. Be patient or buy GPUs.
The above clarification is important and is also in the disclaimer
Visit the website of the project.
Go to our website. Download the ZIP file and unzip it.
unzip latest.zip
cd Fecalface-pack
pip install -r requirements.txt
cd backend
python app.py -e prod
And, finally, browse it.
git clone https://gitlab.com/kennbroorg/iKy.git
You must install the libraries inside requirements.txt
python3 -m pip install -r requirements.txt
First of all, install nodejs.
Inside the directory frontend install the dependencies
cd frontend
npm install
Turn on backend in a terminal, within the directory backend
python3 app.py
Finally, to run frontend server, execute the following command from directory frontend
npm start
Open the browser in this url
If you experience any problems or have any suggestions, you are welcome to create an issue.
Anyone who contributes or contributed to the project, including me, is not responsible for the use of the tool (Neither the legal use nor the illegal use, nor the "other" use).
Keep in mind that this software was initially written for educational purposes (to educate ourselves), and now the goal is to collaborate with the community making quality free software, and while the quality is not excellent (sometimes not even good) we strive to pursue excellence.
The attack process is very processing intensive, taking between 5 and 7 minutes. Be patient or buy GPUs.
We do not refund your money if you are not satisfied. I hope you enjoy using the tool as much as we enjoy doing it. The effort was and is enormous (Time, knowledge, coding, tests, reviews, etc.) but we would do it again. Do not use the tool if you cannot read the instructions and / or this disclaimer clearly.
By the way, for those who insist on remembering my mother, she died many years ago but I love her as if she were right here.