johndpope / faceshifter_tornado Goto Github PK

Faceshifter tornado server

Implementation and deployment of https://arxiv.org/abs/1912.13457.

Requirements

Python 3.6+

Step-by-step installation

1. Install MongoDB Guide
2. Create mongo database with name faceshifter: in shell run mongo and then use faceshifter;
3. cd faceshifter_tornado/
4. python3 -m venv ENV/ and source ENV/bin/activate
5. pip install -r requirements.txt
6. Download weights for AEI generator google drive link
7. Extract weights tar -xvf faceshifter_weights.tar -C faceshifter/model_weights/
8. Run tests: python -m tornado.test.runtests web.tests
9. Launch server with inference worker python app.py

Try it

1. Send photos for swapping. Make POST request on /upload_image with form-data files: source, target. Copy id from response.
2. Get swapping results. Make GET request to /result/$id/.
3. Check result_url.

Deployment notes and todos:

• Add highload testing with locust.io
• Use PyTorch jit compiled graph for saving RAM and speeding up inference
• Docker for faster deployment
• Use PostgreSQL or another relational database with migrations, transaction functionality for better data consistency
• Try to use JIT (PyPy) for faster python code execution

Model implementation reference: https://github.com/Heonozis/FaceShifter-pytorch/