DeepPrivacy Demo

This repository includes the code for the demo of the advanced program for further research on the paper 'User-centric Perspectives Toward DeepPrivacy Framework: Exploring the Influence of Social Relationships and Facial Substitution Similarity on User Experience' on CHI 2025.

1. File Structure

├── README.md
├── backend
│   ├── LICENSE
│   ├── README.md
│   ├── arcface_model
│   ├── checkpoints
│   ├── crop_224
│   ├── data
│   ├── deepPrivacy_service.py
│   ├── demo_file
│   ├── docs
│   ├── insightface_func
│   ├── main.py
│   ├── models
│   ├── options
│   ├── output
│   ├── parsing_model
│   ├── people
│   ├── pg_modules
│   ├── simswaplogo
│   ├── test.png
│   ├── (miscellaneous files, not used on current demo)
│   ├── util
│   └── workaround.ipynb
└── frontend
    ├── README.md
    ├── build
    ├── node_modules
    ├── package-lock.json
    ├── package.json
    ├── public
    └── src

The frontend directory contains the code for React-based web application, and the backend directory contains the code for Flask-based server-side application. If you want to make a fix, please do not forget to npm run build in the frontend directory before executing the python main.py in the backend directory.

2. Prerequisites

The demo requires the following software and libraries:

numpy==1.23.5
pillow==10.4.0
requests==2.32.3
scikit-learn==1.3.2
sympy==1.13.2
flask==3.0.3
torch==1.12.0
torchvision==0.14.1
matplotlib==3.7.5
networkx==3.1

Also, you need to download several pre-trained models to run the demo. You can download such weights following the instructions in the SimSwap repository.

3. How to start

To start this application, you have to execute follow commands.

cd backends
python main.py

Then, you can access to the demo page from your browser.

http://localhost:5001/