neuro_morpho.model.base
=======================

.. py:module:: neuro_morpho.model.base

.. autoapi-nested-parse::

   Base class for all models.



Attributes
----------

.. autoapisummary::

   neuro_morpho.model.base.ERR_NOT_IMPLEMENTED


Classes
-------

.. autoapisummary::

   neuro_morpho.model.base.BaseModel


Module Contents
---------------

.. py:data:: ERR_NOT_IMPLEMENTED
   :value: 'The {name} method is not implemented'


.. py:class:: BaseModel

   Base class for all models.

   This class defines the interface for all models. All models should inherit
   from this class and implement the methods defined here.


   .. py:method:: fit(data_dir: str | pathlib.Path) -> BaseModel
      :abstractmethod:


      Fit the model to the data.

      :param data_dir: The directory containing the data files to fit the model
                       images should have the size (n_samples, channels, height, width)
      :type data_dir: str|Path

      :returns: The fitted model
      :rtype: BaseModel



   .. py:method:: predict_dir(in_dir: str | pathlib.Path, out_dir: str | pathlib.Path, threshold: float, mode: str, tile_size: tuple[int, int], tile_assembly: str, binarize: bool, fix_breaks: bool) -> None
      :abstractmethod:


      Predict the output for all images in the given directory.

      :param in_dir: The directory containing the data files to predict
                     images should have the size (n_samples, channels, height, width)
      :type in_dir: str|Path
      :param out_dir: The directory to save the output
      :type out_dir: str|Path
      :param threshold: Use to get the hard prediction (binary output)
      :type threshold: float
      :param mode: The mode of the prediction, can be 'test' or 'infer'
                   'test' - runs the model on the test set (same size images) and saves the statistics
                   'infer' - runs the model on the inference set (images may be of different size) and saves the output
      :type mode: str
      :param tile_size: The size of the tiles to use for tiling the input images
      :type tile_size: tuple[int, int]
      :param tile_assembly: The method for assembling the tiles, can be 'nn' (nearest neighbor), 'mean', or 'max'
      :type tile_assembly: str
      :param binarize: Whether to binarize the output
      :type binarize: bool
      :param fix_breaks: Whether to fix breaks in the binarized output
      :type fix_breaks: bool



   .. py:method:: predict(x: numpy.ndarray) -> numpy.ndarray
      :abstractmethod:


      Predict the output given the input x

      :param x: The input data should be size of (n_samples, channels, height, width)
      :type x: np.ndarray
      :param thresh: The threshold to use for the prediction
      :type thresh: float

      :returns: The predicted output
      :rtype: np.ndarray



   .. py:method:: predict_proba(x: numpy.ndarray, tiler: neuro_morpho.model.tiler.Tiler) -> numpy.ndarray
      :abstractmethod:


      Predict a soft version of the output given the input x and tiling params as an option

      :param x: The input data should be size of (n_samples, channels, height, width)
      :type x: np.ndarray
      :param tiler: The tiler object to use for tiling the input data
      :type tiler: Tiler

      :returns: The predicted output
      :rtype: np.ndarray



   .. py:method:: find_threshold(in_dir: str | pathlib.Path, out_dir: str | pathlib.Path, model_dir: str | pathlib.Path, model_out_val_y_dir: str | pathlib.Path, min_thresh: float, max_thresh: float, thresh_step: float) -> float
      :abstractmethod:


      Predict the output for all images in the given directory.

      :param in_dir: The directory containing images (validation set)
      :type in_dir: str|Path
      :param out_dir: The directory containing labels (validation set)
      :type out_dir: str|Path
      :param model_dir: The directory containing model checkpoints
      :type model_dir: str|Path



   .. py:method:: save(path: pathlib.Path | str) -> None
      :abstractmethod:


      Save the model to the given path.

      :param path: The path to save the model
      :type path: Path|str



   .. py:method:: load(path: pathlib.Path | str) -> None
      :abstractmethod:


      Load the model from the given path.

      :param path: The path to load the model
      :type path: Path|str