deepextractor.utils.io
======================

.. py:module:: deepextractor.utils.io






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

.. py:data:: logger

.. py:function:: get_loaders(train_dir, train_target_dir, val_dir, val_target_dir, batch_size, train_transform=False, val_transform=False, num_workers=4, pin_memory=True, time_domain=True)

   Return train and validation DataLoaders.


.. py:function:: check_accuracy(loader, model, model_name, device='cuda')

   Compute MSE loss on the validation set and return average losses.


.. py:function:: numpy_to_gwf(strain, sample_times, channel, output_filename)

   Write a strain time series to a GWF (frame) file.

   :param strain: The time-domain strain data.
   :type strain: array-like
   :param sample_times: The corresponding time array.
   :type sample_times: array-like
   :param channel: Channel name, e.g. ``'L1:STRAIN'``.
   :type channel: str
   :param output_filename: Path for the output GWF file.
   :type output_filename: str


.. py:function:: gwf_to_lcf(start_time, duration, channel_name, gwf_file_location)

   Write a minimal LCF (frame cache) file alongside the GWF file.


.. py:function:: load_tf_model(path, model_name)

   Load a TensorFlow/Keras SavedModel.

   Requires the ``[generative]`` optional dependencies:
   ``pip install deepextractor[generative]``.