Eisen is a package that consists of popular datasets, model architectures, data loading functions, common image transformations, training, testing and validation work-flows which are useful when working on medical image analysis tasks using deep learning.
Eisen implements an opinionated API that builds directly on PyTorch. The goal of Eisen is to be:
- Extremely simple to use
- Extremely easy to understand
- Similar to other packages in the PyTorch echosystem
- Easy to contribute to
When developing Eisen we have therefore decided to keep the number of lines of code to the minimum, adopt a component-based architecture, avoid over-engineering and promote an easy-to-follow coding style.
Users of Eisen should be able to get started in a few minutes, by following the examples contained in this README.
Developers of Eisen should be able to grasp the entire code-base in a few hours.
Simplicity is the answer.
Find more information about this project and extensive documentation at http://docs.eisen.ai