ImageTranslatorFGR - Feature Generation&Regression

class aydin.it.fgr.ImageTranslatorFGR(feature_generator: Optional[aydin.features.base.FeatureGeneratorBase] = None, regressor: Optional[aydin.regression.base.RegressorBase] = None, balance_training_data: bool = False, voxel_keep_ratio: float = 1, max_voxels_for_training: Optional[int] = None, favour_bright_pixels: float = 0, blind_spots: Optional[Union[str, List[Tuple[int]]]] = None, tile_min_margin: int = 8, tile_max_margin: Optional[int] = None, max_memory_usage_ratio: float = 0.9, max_tiling_overhead: float = 0.1)

Feature Generation & Regression (FGR) based Image TranslatorFGR Image Translator.

add_transform(transform: aydin.it.transforms.base.ImageTransformBase, sort: bool = True)

Adds the given transform to the self.transforms_list

Parameters

transformImageTransformBase

clear_transforms()

Clears the transforms list

static load(path: str)

Returns an ‘all-batteries-included’ image translation model at a given path (folder).

Parameters

pathstr

path to load from.

static parse_axes_args(batch_axes: Union[List[int], List[bool]], chan_axes: Union[List[int], List[bool]], ndim: int)

Parameters

batch_axesUnion[List[int], List[bool]]

chan_axesUnion[List[int], List[bool]]

ndimint

save(path: str)

Saves a ‘all-batteries-included’ image translation model at a given path (folder).

Parameters

pathstr

path to save to

Returns

frozen

stop_training()

Stops currently running training within the instance by calling the corresponding stop_fit() method on the regressor.

train(input_image, target_image=None, batch_axes=None, channel_axes=None, train_valid_ratio=0.1, callback_period=3, jinv=None)

Train to translate a given input image to a given output image. This has a lot of the machinery for batching and more…