smujiang / digipath_mltk

Parameterized large image pre-processing with pychunklbl.toolkit

The package module pychunklbl.toolkit provides eight parameterized functions designed to work with large image files and provide pre-processing for machine learning with tensorflow 2.0.

The package module is intended developers to create machine learning datasets leveraging the openslide library for usage with tensorflow 2.0.

---

Installation

pip install -i https://test.pypi.org/simple/ pychunklbl==0.0.5

# requires python 3.5 or later
pip3 install -r requirements.txt

---

Package usage:

Example .yml files are in the DigiPath_MLTK/data/run_files directory and may be used as templates to run with your data.

Each parameters (.yml) file is a template for running one of the eight methods.

When edited with valid data they run from the command line using the example main function in this repository:
src/python/digipath_tk_run.py

Command line example:

python3 ../DigiPath_MLTK/src/digipath_tk_run.py -run_directory . -run_file annotations_to_dir.yml

---

Installation Test Notes:

The DigiPath_MLTK/test directory contains the Makefile that may be used for either development or post-installation testing.

The README.md details how to run the integration test suite after installing the package as a module.

By changing the first line of the Makefile SCRIPT = ../src/python/digipath_tk_run.py to SCRIPT = ../src/python/digipath_mltk.py the code in this (cloned) repository can be run for development purposes.

Note: the run_parameters from the .yml files are just regular python type dict when passed to the toolkit module.