Scene API¶

class hugin.engine.scene.ArrayExporter(zarr_dataset, destination_array, *args, **kwargs)¶

Bases: hugin.engine.scene.SceneExporter

destination_array¶: A trait for unicode strings.

flow_prediction_from_array_loader(loader, predictor)¶

class hugin.engine.scene.ArrayModel(name: str, model: hugin.engine.core.RasterModel, *args, **kwargs)¶: Bases: hugin.engine.scene.BaseSceneModel

class hugin.engine.scene.ArrayModelPredictor(*args, **kwargs)¶

Bases: hugin.engine.scene.ArrayModel

predict(data_source: hugin.io.zarr_loader.ZarrArrayLoader)¶

class hugin.engine.scene.ArrayModelTrainer(*args, **kwargs)¶

Bases: hugin.engine.scene.ArrayModel

save(destination: str = None)¶

train(data_source: hugin.io.zarr_loader.ZarrArrayLoader)¶

class hugin.engine.scene.AvgEnsembleScenePredictor(predictors, *args, name=None, resume=False, cache_file=None, **kwargs)¶

Bases: hugin.engine.scene.BaseEnsembleScenePredictor

predict_scene_proba(scene, *args, **kwargs)¶

class hugin.engine.scene.BaseEnsembleScenePredictor(predictors, *args, name=None, resume=False, cache_file=None, **kwargs)¶

Bases: hugin.engine.scene.BaseSceneModel, hugin.engine.scene.MultipleSceneModel

predict_scenes_proba(scenes)¶

Run the predictor on all input scenes

Parameters:	scenes – An iterable object yielding tuples like (scene_id, type_mapping) predictor – The predictor to use for predicting scenes (defaults to self)
Returns:	a list of predictions according to model configuration

class hugin.engine.scene.BaseSceneModel(base_directory=None, post_processors=None, pre_processors=None, metrics=None, gti_component=None)¶

Bases: traitlets.traitlets.HasTraits

base_directory¶: A trait for unicode strings.

predict_scene_proba(*args, **kwargs)¶

class hugin.engine.scene.CoreScenePredictor(predictor, name=None, mapping=None, stride_size=None, window_size=None, output_shape=None, prediction_merger=<class 'hugin.engine.core.NullMerger'>, post_processors=None, pre_processors=None, format_converter=<hugin.io.loader.NullFormatConverter object>, metrics=None)¶

Bases: hugin.engine.scene.BaseSceneModel

predict_scene_proba(scene, *args, **kwargs)¶

class hugin.engine.scene.MultipleFormatExporter(*args, exporters=[], **kwargs)¶

Bases: hugin.engine.scene.SceneExporter

save_scene(*args, destination=None, **kwargs)¶

class hugin.engine.scene.MultipleSceneModel(scene_id_filter=None, randomize_training=True, threaded=True, prefetch_queue_size=None)¶

Bases: object

This class is intended to be inherited by classes aimed to predict on multiple scenes

predict_scenes_proba(scenes, predictor=None)¶

Run the predictor on all input scenes

Parameters:	scenes – An iterable object yielding tuples like (scene_id, type_mapping) predictor – The predictor to use for predicting scenes (defaults to self)
Returns:	a list of predictions according to model configuration

train_scenes(scenes, validation_scenes=None, trainer=None)¶

class hugin.engine.scene.RasterIOSceneExporter(*args, srs_source_component=None, rasterio_options={}, rasterio_creation_options={}, filename_pattern='{scene_id}.tif', **kwargs)¶

Bases: hugin.engine.scene.SceneExporter

save_scene(scene_id, scene_data, prediction, destination=None, destination_file=None)¶

class hugin.engine.scene.RasterScenePredictor(model, *args, scene_id_filter=None, **kwargs)¶: Bases: hugin.engine.scene.CoreScenePredictor, hugin.engine.scene.MultipleSceneModel

class hugin.engine.scene.RasterScenePredictorMaxClass(*args, **kwargs)¶

Bases: hugin.engine.scene.RasterScenePredictor

predict_scene_proba(*args, **kwargs)¶

class hugin.engine.scene.RasterSceneTrainer(model, *args, destination=None, scene_id_filter=None, **kwargs)¶

Bases: hugin.engine.scene.CoreScenePredictor, hugin.engine.scene.MultipleSceneModel

predict_scene_proba(scene, dataset_loader=None)¶

save(destination=None)¶

class hugin.engine.scene.SceneExporter(destination=None, metric_destination=None, _format_options={})¶

Bases: traitlets.traitlets.HasTraits

destination¶: A trait for unicode strings.

flow_prediction_from_source(loader, predictor)¶

save_scene(scene_id, scene_data, prediction, destination=None)¶