Kime Engine

Kime is a web app for upscaling and enhancing images using generative adversarial networks. This repository contains the engine which wraps the deep learning model and performs the bulk of the computation.

Getting Started

Step 1: Create a virtual environment and install dependencies.

Create a new Virtual Environment for the project and activate it. If you don't have the virtualenv command yet, you can find installation instructions here.

$ virtualenv venv
$ source venv/bin/activate

Next, install the project dependencies, which are listed in requirements.txt.

$ pip install -r requirements.txt

Step 2: Update environment variables and start the server.

Create a new file named config.json and update the new file with your Azure credentials. It should look similar to this:

{
  "storage": {
    "azure": {
      "account_name": "...",
      "access_key": "..."
    }
  },

  "backend": {
    "username": "...",
    "secret_hash": "..."
  },
}

Now you are ready to start the backend service:

$ gunicorn --timeout 120 --max-requests 10 -b 127.0.0.1:3785 engine.__main__:app