Face changer

This is an initial version for test task. Future iterations are expected to implement Celery and Docker integration, utility function segmentation, code coverage with tests and detailed code commenting.

---

Instalation

0. Init the database

mkdir db_data
touch db_data/db.sqlite3

1. Download inswapper_128.onnx to ~/.insightface/models/

mkdir -p ~/.insightface/models/ ; wget -P ~/.insightface/models/ -O ~/.insightface/models/inswapper_128.onnx https://huggingface.co/Devia/G/resolve/main/inswapper_128.onnx?download=true

2. Clone the repository

$ git clone https://github.com/IlyaTorch/test-face-swapper
$ cd test-face-swapper

3. Create virtual environment and install the dependencies. Python3.11 required

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

4. Fill in .env file or use .env_example:

$ mv .env_example .env

5. Run the application
   The initial run initiates the download of buffalo_l to .insightface\models

uvicorn src.main:app --reload

Endpoints

POST /api/v1/swapper/in-swapping : Perform face swapping using the provided images or URLs.
Supports a combination of different types

Data required:
face_image: The image or URL with face
source_image: The image or URL for replacing faces

curl --location 'http://127.0.0.1:8000/api/v1/swapper/in-swapping' \
--form 'face_image="<LINK WITH IMAGE>"' \
--form 'source_image="<LINK WITH IMAGE>"'

or

curl --location 'http://127.0.0.1:8000/api/v1/swapper/in-swapping' \
--form 'face_image=@"<PATH TO FILE>"' \
--form 'source_image=@"<PATH TO FILE>"'