vertebrae.extractors.spatial

Explicit spatial extractor contracts and adapters.

Classes

SpatialLayout

Mapping from a spatial feature tensor to a uniform image grid.

SpatialOutputSpec

Declarative description of one spatial extractor output.

SpatialEmbeddingOutput

Per-image spatial feature tensors from one named output.

CallableSpatialExtractor

Wrap a callable returning explicit spatial outputs.

PrecomputedSpatialExtractor

Spatial extractor for tensors already stored in the dataset inputs.

Module Contents

class vertebrae.extractors.spatial.SpatialLayout[source]

Mapping from a spatial feature tensor to a uniform image grid.

class vertebrae.extractors.spatial.SpatialOutputSpec[source]

Declarative description of one spatial extractor output.

class vertebrae.extractors.spatial.SpatialEmbeddingOutput[source]

Per-image spatial feature tensors from one named output.

class vertebrae.extractors.spatial.CallableSpatialExtractor(name, transform_fn, output_specs, recipe_data=None, streaming_safe=True, cache_identity=None)[source]

Wrap a callable returning explicit spatial outputs.

Parameters:

  • name (str)

  • transform_fn (Callable[[Any], Any])

  • output_specs (Iterable[SpatialOutputSpec])

  • recipe_data (Optional[Dict[str, Any]])

  • streaming_safe (bool)

  • cache_identity (Optional[str])

class vertebrae.extractors.spatial.PrecomputedSpatialExtractor(output_specs, name='precomputed_spatial')[source]

Bases: CallableSpatialExtractor

Spatial extractor for tensors already stored in the dataset inputs.

Parameters: