NearestCentroidAssigner

Note

This page is a reference documentation. It only explains the class signature, and not how to use it. Please refer to the Habitat Guide and Python API guide (v2.0) for usage.

class NearestCentroidAssigner(model: HabitatModel)

Bases: object

Assign each supervoxel to the habitat of its nearest centroid.

The fitted model is bound at construction time – the ordinary way to obtain this assigner is model.assigner(). Prediction then has no way to re-learn anything, because everything it needs is inside the model; this is what enforces train/predict consistency structurally.

Habitat ids are the centroid row indices plus one, so label 0 is reserved for background, matching the v0.1 label-image convention.

Parameters:

model – The fitted habitat definition to project.

__init__(model: HabitatModel) → None

property model: HabitatModel

Return the fitted habitat definition this assigner projects.

property spec: Spec

Return the algorithm specification, bound to the model id.

__call__(supervoxel_map: Supervoxelization) → HabitatMap

Project the fitted habitat definition onto one subject.

Parameters:

supervoxel_map – Supervoxelization of the subject to label.

Returns:

The subject’s habitat label image, tagged with the model’s id.

Raises:

CompatibilityError – If the supervoxel features lack a feature the model requires, or a label has no feature row.

Examples using habit.habitat_model.NearestCentroidAssigner

/auto_examples/06_parallel_runs/plot_06_reuse_workers

Reusing workers across passes