Deep Fake Evolution is an API that can perform arbitrary face swapping on videos using one single trained model.

That allows you to perform face-swapping on videos using a single and efficient trained model.

Examples of usage:

Team:

• Anna Gulich
• Sergey Shvets
• Alexey Naymushin
• Liudmila Mihaylichenko

Key features:

1. Arbitrary single-face and multi-face swapping using modern, fast and efficient SimSwap Framework
2. High-quality output video for professional use
3. Easy-to-use interface for users

Installation on CPU

API usage

Project structure description:

├── FaceDetectAntelopeModel.py              - Face detection model class
├── LICENSE                                 - License file
├── README.MD                               - Readme
├── SwapManager.py                          - Class that manages face swapping using SimSwap framework
├── init.py                                 - Module initialization class
├── demo                                    - Demo files folder
│ ├── multi                                 - Demo folder for multi swap testing
│ └── single                                - Demo folder for single swap testing
├── downloads                               - Folder used for keeping result videos files available for download
├── guides                                  - Guides and instructions for users
│ ├── INSTALLATION.md                       - Installation guide
│ └── USAGE.md                              - Usage guide
├── init.sh                                 - SimSwap installation shell script
├── main.py                                 - API file and program entry point
├── main_test.py                            - API tests
├── patch_cpu.ps1                           - SimSwap patch for CPU usage
├── requirements                            - Project requirements
│ ├── common_mac.txt                        - MacOS common requirements
│ ├── git_actions_ubuntu.txt                - Linux Ubuntu requirements
│ ├── mac_arm64.txt                         - MacOS ARM64 specific requirements
│ └── mac_x86_64.txt                        - MacOS x86-64 specific requirements
├── uploads                                 - API users' uploads folder
│ ├── images                                - Image files uploads
│ ├── multispecific.                        - Multispecific archive files uploads
│ └── videos                                - Video files uploads
└── utilities                               - Utilities module
├── init.py                                 - Module initialization files
├── api_constants.py                        - Constants used in API
├── api_utils.py                            - Helper functions used by API
├── enums.py                                - Enumeration classes
└── hasher.py                               - Class used for calculating video file hashes

Dependencies:

• Python 3.7+
• opencv-python~=4.7.0.72
• torch~=1.12.0
• torchvision~=0.13.0
• torchaudio
• onnxruntime
• moviepy
• insightface==0.2.1
• imageio==2.5.0
• numpy~=1.21.6
• uvicorn==0.21.1
• starlette==0.26.1
• Werkzeug==2.2.3
• fastapi==0.95.1
• python-multipart
• pytest
• httpx