/nuclei_segmentation

Primary LanguagePythonMIT LicenseMIT

nuclei_segmentation

Requirements

  • python 2.7
  • numpy
  • keras
  • tensorflow
  • scipy
  • cv2
  • skimage
  • pillow
  • h6py
  • configParser

Training and Testing

First run:

python prepare_dataset.py

to prepare dataset

  • Training Directly run:
python run_training.py

At the begining of the training, a folder name "test1" will be created in "experiment". During the training, models will be saved in it.

  • Testing
python run_training.py

It will generate a folder name '1' in 'test1'. The predictions will be shown in it. You may modify configuration.txt to change the experiement settings.

Licence

The code is licensed under MIT. Copyright (c) 2018. The dataset is obtained from https://nucleisegmentationbenchmark.weebly.com/dataset.html.