https://github.com/simonwinther/rare-unet

This repository features RARE-UNet — a resolution-aware 3D U-Net for adaptive medical segmentation. It uses multi-scale entry blocks and resolution-based routing to dynamically adjust the inference path to input resolution. Combined with consistency-based training, RARE-UNet delivers accurate, efficient segmentation across resolutions.

3d-unet adaptive-inference brain-mri consistency-training dynamic-routing hippocampus-segmentation medical-image-segmentation multi-scale rare-unet resolution-aware tumor-segmentation

Last synced: 13 Oct 2025