Package `napari_blender`

napari-blender is a plugin for napari that allows you to render 3D scenes using Blender.

This is a system that combines optical validation of model predictions by having different 3D visualizations and quantitative evaluation. Utilizing Blender’s rendering capabilities for deepening the understanding of nuclei segmentation for users with different levels of expertise. Examples are time-lapse animations (aimed to be generated from label data), opaque ground truth visualizations with solid prediction objects to compare prediction quality optically and 3D images where nuclei are coloured according to their prediction quality. To facilitate meaningful quantitative evaluation, different metrics are calculated for these predictions and displayed within the animation, such as the Jaccard index, Intersection over Union and the F1-score. The emphasis in this system is on operating it with limited technical knowledge, allowing for bridging between researchers and developers; but includes metrics that allow for a deeper understanding of performance for expert users.

Installation

The installation of napari-blender is made easier by the inclusion of a virtual environment containing specified versions of each library. These versions should be followed, since dependencies could otherwise break the build.

To set up the environment, download the environment.yml file and use it to create a new conda environment. A raw text version is made available as well: environment.yml

conda env create -f environment.yml
conda activate napari-blender-env

Now you can install the plug-in using:

pip install napari-blender

Note:

In Windows, an error might occur in the installation of the mathutils package, stating 'Microsoft Visual C++ 14.0 or greater is required'. A fix can be found on StackOverflow.