Lightning-Covid19

A detector for covid-19 chest X-ray images
using PyTorch Lightning (for educational purposes)

Overview

This project came with actual global situation and reciently release X-ray dataset.

Data

The baseline models uses only covid-chestxray-dataset
which is rather small (79 images at the time of writing).

They are also accessible with following python packages:

• https://github.com/ieee8023/covid-chestxray-dataset
• https://github.com/mlmed/torchxrayvision

Installation

1. Download the project: git clone https://github.com/PyTorchLightning/lightning-Covid19.git

2. Setup your development environment:

From Python Virtual Environments

python3.6 -m venv venv  # from repo root, use python3.6+
source venv/bin/activate
pip install -r requirements

From conda

To install all the needed dependencies, or update the conda environment: ./setup_dev_env.sh

Then to activate the conda environment: source path.bash.inc

1. You are good to go to run your experiments -> See section Experiments.

Experiments

See the Experiments on Wiki.

Contribution

Anyone is welcome to contribute and use this project!

The project’s features, bugs, and questions are discussed and planned in the project’s Github issue’s.

For informal, longer discussion we use PyTorch-Lightning (PL) Slack Workspace. See how to join it - See point 4 in Section “Asking for Help” in PL README.

For easier combination and coordination we are using separate channel in PL Slack.
Join our open-source community!

Notes on contributing code and results:

• when mering PR, do “Squash & merge” so in master is only one commit and does not overlap with other commits from different PRs in parallel
• for writing particular result or how to launch anything, lets ise Wiki for now as it much more flexible

References

TBD