The purpose of the Global Disease Database (GDD) Webapp is to provide an interface for scientists to validate images of diseases in the GDD so that the images can be used to train AI models more effectively.

The app provides users with an interface where they are presented with an image. They then can take one of the following actions:

• CONFIRM - if the user believes the image should be used as part of the GDD
• SKIP - if the user is unsure whether the image should be in the GDD
• Mark as INVALID - if the user believes the image should not be used in the GDD

Every time a user CONFIRMS an image is credible they also add labelling for any disease that is present. It is this labelling that will be used for training the AI model. Every image will be confirmed by 4 different labellers. Once 4 labellers have confirmed the image, the data will be stored for the AI model.

If three users mark an image as INVALID, this image won't be displayed to any new labeller.

The app is a React application (using create-react-app and TypeScript). It uses the firebase client to authenticate users and to fetch and store images and their associated data.

The app is hosted on firebase and can be found at https://aiscope-labeling-app.web.app/

You must have node and yarn installed.

It is recommended to use the Node Version Manager to manage your node installations and to use the version of node that is specified in the .nvmrc file.

To run the app locally:

1. Install all dependencies.

yarn install

2. Start the application.

yarn start

3. Go to localhost:3000 in the browser.

To run the tests:

yarn test

Uses Jest as the test runner and React Testing Library for interacting with the DOM

To run the tests with Firebase Emulators:

yarn test:emulator

To run linting (ESLint):

yarn lint

To auto format the code (Prettier):

yarn prettier

The project uses husky (Git Hooks) to run tests and lint before committing.

The app is currently deployed in test environment, on every push to main branch. Github actions are used to run automated tests, and if they are successful, a new version of the application is deployed using the firebase cli.

When a new tag is generated, it is published on production environment.

Contributions should be made by creating a new branch from main branch, and then make a PR against main. Maintainers of the project can then review and merge these PRs.

All PRs should aim to have good unit test coverage and all tests must be passing before a PR can be merged.