mikful / streamlit-fastapi-model-serving

Simple web app example serving a PyTorch model using streamlit and FastAPI

streamlit-fastapi-model-serving

Simple example of usage of streamlit and FastAPI for ML model serving described on this blogpost and PyConES 2020 video.

When developing simple APIs that serve machine learning models, it can be useful to have both a backend (with API documentation) for other applications to call and a frontend for users to experiment with the functionality.

In this example, we serve an image semantic segmentation model using FastAPI for the backend service and streamlit for the frontend service. docker-compose orchestrates the two services and allows communication between them.

To run the example in a machine running Docker and docker-compose, run:

docker-compose build
docker-compose up

To visit the FastAPI documentation of the resulting service, visit http://localhost:8000 with a web browser.
To visit the streamlit UI, visit http://localhost:8501.

Logs can be inspected via:

docker-compose logs

Deployment

To deploy the app, one option is deployment on Heroku (with Dockhero). To do so:

• rename docker-compose.yml to dockhero-compose.yml
• create an app (we refer to its name as <my-app>) on a Heroku account
• install locally the Heroku CLI, and enable the Dockhero plugin with heroku plugins:install dockhero
• add to the app the DockHero add-on (and with a plan allowing enough RAM to run the model!)
• in a command line enter heroku dh:compose up -d --app <my-app> to deploy the app
• to find the address of the app on the web, enter heroku dh:open --app <my-app>
• to visualize the api, visit the address adding port 8000/docs, e.g. http://dockhero-<named-assigned-to-my-app>-12345.dockhero.io:8000/docs(not https)
• visit the address adding :8501 to visit the streamlit interface
• logs are accessible via heroku logs -p dockhero --app <my-app>

Debugging

To modify and debug the app, development in containers can be useful (and kind of fun!).