ImageNet Service

This is a small exploratory project to play around with ImageNet models and explore the Pytorch library.

The Deploying PyTorch in Python via a REST API with Flask has been used as a reference and starting point for this project.

Setup

Local Install

First, you will need to configure a Python environment.

An environment.yml file has been provided to that effect.

This project assumes the use of micromamba and has only been tested under Linux OS.

To create the environment:

make create-env

If you are using a different Conda tool

make CONDA_TOOL="<YOUR_CONDA_TOOL>" create-env

The application and Python dependencies can then be installed

make install

Docker Install

To build the application as a Docker container

make docker-build

To run the application

make docker-run

Or, with the application module mounted to the Docker container for live-reload development

make docker-run-dev

Development

This project uses Nox to automate linting checks, autoformatting and running tests.

Check lint

make check-lint

Fix lint

make fix-lint

Run tests

make test

To run all checks and fixes

nox