/unet

Primary LanguagePython

Implementation of an U-NET

Implemetation of the U-NET Structure. Built based on https://www.youtube.com/watch?v=IHq1t7NxS8k and modified further.

Usage

The image shape must be (n, m, ny, nx) where

  • n: number of images
  • m: image depts (e.g. rgb=3, gray=1)
  • ny: vertical image size [pixel]
  • nx: horizontal image size [pixel]

The label shape must be (n, m, ny, nx) where

  • n: number of images
  • m: number of features, typically 1
  • ny: vertical image size [pixel]
  • nx: horizontal image size [pixel]

Besides the obvious parameters like device, batch_size etc, you can adjust/vary the following parameters using the hyperparameters.yaml:

  • number of features for each level and the number of levels
  • stride and kernel_size for down- and up-path and for the bottleneck

Monitoring Unet

If your runs are managed using hydra, the results can by nicely monitored using tensorboard. There's an example pytho file using hydra in the example directory of this repo: example/using_hydra.py.

For that, navigate to the respective directory and run tensorboar --logdir