MrGiovanni/UNetPlusPlus
[IEEE TMI Best Paper Award] Official Implementation for UNet++
What it solves
It addresses two primary limitations of the original U-Net architecture used in image segmentation: the uncertainty regarding the optimal depth of the network and the restrictive nature of its skip connections.
How it works
UNet++ uses a nested U-Net architecture consisting of multiple U-Nets of varying depths. Its decoders are densely connected at the same resolution through redesigned skip pathways, allowing the model to better exploit multiscale features.
Who it’s for
This project is designed for researchers and developers working on medical image segmentation.
Highlights
- Official implementations provided in both Keras and PyTorch.
- Redesigned skip connections to improve feature exploitation.
- Nested architecture that allows for varying depths.
Related
- Project
- Project
- Project
- Project