jupyanno

Overview

The focus of the project is to make acquiring more training data as easy as possible for data-scientists. The project uses Jupyter notebooks and widgets to allow for custom GUI tools to be made and deployed with JupyterHub / BinderHub. The tools are also designed to make tracker user interaction easy as well (how long did a user look, what did they change / zoom into / click). The toolkit is made to be easily extensible and suggestions / improvements are welcome.

Tasks Addressed

Multiple Choice

Binary Choice (with unknown option)

Bounding Box (in progress)

Example Tasks

Chest X-Ray for Pneumonia Task

Here we have a demo chest X-Ray task where the annotation and viewing of a sample dataset can be done entirely with these tools.

Installation

The widgets can be installed by running

pip install git+https://github.com/chestrays/jupyanno/

For a more complete environment you can use repo2docker to create a Docker Image from the repository