approximate_crappifier#

from pssr.train import approximate_crappifier
pssr.train.approximate_crappifier(crappifier: Crappifier, space: list[Dimension], dataset: Dataset, max_images=None, opt_kwargs=None)#

Approximates Crappifier parameters from ground truth paired images. Uses Bayesian optimization because Crappifier functions are not differentiable.

Parameters:

  • crappifier (Crappifier) – Crappifier whose parameter space will be optimized.

  • space (list[Dimension]) – List of parameter spaces for each crappifier parameter.

  • dataset (Dataset) – Paired image dataset to load data from.

  • max_images (int) – Number of image samples to average computations over for each optimization step. Default is None, using all images in dataset.

  • opt_kwargs (dict[str, Any]) – Keyword arguments for skopt gp_minimize. Default is None