Loading Extensions (machine-learning and plugin features)

Since version 2.9.0, Shape-Out 2 allows loading dclab plugin features and machine-learning features. You may need those if you need to quantify anything that is not covered by the default features.

Note

If you installed Shape-In via installer (not via pip), then many extensions might not work due to software dependencies that those extensions might have. If this happens, please create an issue in the Shape-Out 2 repository so we can find a solution.

Warning

Extensions can be harmful. Please only load extensions that you received first-hand from people you trust.

You can load and manage extensions via the Edit | Preferences dialog in the Extensions tab.

_images/qg_extensions.png

Example: Fluorescence volume

Download the extension_fl1_density.py extension and add it to Shape-Out. You will see a new scalar feature named “FL-1 density [a.u.]” that quantifies the collected fluorescence signal per object volume for the fluorescence channel 1. Using this extension as a template, you could create the density features for the other fluorescence channels as well.

Example: RBC-detection with machine-learning

Note

For this example you will need to have tensorflow installed.

Download the extension_naive_rbc_score.modc extension and add it to Shape-Out. You will see a new scalar feature named “RBC score (naive)”. This showcase model only consists of one dense layer of size six, but already does a better-than-random job in identifying red blood cells.