Packaged form of ZeroCostDl4Mic to make the process more platform agnostic. Attempts to bundle reusable code and structure model training and prediction into a no-code config file toolset.
pip install git+https://github.com/ctr26/dl4mic
Currently working with Noise2Void and Care2D
This project uses poetry to build, test and manage dependnecies:
Quick start: peotry build poetry install poetry run pytest
Note that testing is (rightly) slow due to running model epochs for testing
- Find all the bugs
- Implement the full roster of ZeroCostDL4Mic models.
- Command line interface
- Implement lazy loading of large/uninstalled packages (looking at you pyTorch)