CaBRNet is an open source library that offers an API to use state-of-the-art prototype-based architectures (also called case-based reasoning models), or easily add a new one.

Currently, CaBRNet supports the following architectures:

• ProtoPNet, as described in Chaofan Chen, Oscar Li, Chaofan Tao, Alina Jade Barnett, Jonathan Su and Cynthia Rudin. This Looks like That: Deep Learning for Interpretable Image Recognition. Proceedings of the 33rd International Conference on Neural Information Processing Systems, page 8930–8941, 2019.
• ProtoTree, as described in Meike Nauta, Ron van Bree and Christin Seifert. Neural Prototype Trees for Interpretable Fine-grained Image Recognition. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 14928–14938, 2021.

Index

• Build and install
• Quickstart

Authors

This library is collaboratively maintained by members of CEA-LIST. The current point of contact is Romain Xu-Darme. The following authors contributed in a significant manner to the code base and/or the documentation of the library:

• Romain Xu-Darme (CEA-LIST)
• Aymeric Varasse (CEA-LIST)
• Alban Grastien (CEA-LIST)
• Julien Girard-Satabin (CEA-LIST)

The following authors contributed in a significant manner to the experiments and the publication of trained models:

• Jules Soria (CEA-LIST)
• Alban Grastien (CEA-LIST)
• Romain Xu-Darme (CEA-LIST)