.. note::
    :class: sphx-glr-download-link-note

    Click :ref:`here <sphx_glr_download_auto_examples_apply_segm.py>` to download the full example code
.. rst-class:: sphx-glr-example-title

.. _sphx_glr_auto_examples_apply_segm.py:


Example 04: Apply trained network for segmentation
==================================================

This script applies a trained MS-D network for segmentation (i.e. labeling)
Run generatedata.py first to generate required training data and train_segm.py to train
a network.


.. code-block:: default


    # Import code
    import msdnet
    import glob
    import tifffile
    import os
    import numpy as np

    # Make folder for output
    os.makedirs('results', exist_ok=True)

    # Load network from file
    n = msdnet.network.SegmentationMSDNet.from_file('segm_params.h5', gpu=True)

    # Process all test images
    flsin = sorted(glob.glob('test/noisy/*.tiff'))
    for i in range(len(flsin)):
        # Create datapoint with only input image
        d = msdnet.data.ImageFileDataPoint(flsin[i])
        # Compute network output
        output = n.forward(d.input)
        # Save labels with maximum probability to file (i.e. prediceted labels for each pixel)
        tifffile.imsave('results/segm_label_{:05d}.tiff'.format(i), np.argmax(output,0).astype(np.uint8))
        # Save probability map of a single channel (here, channel 2) to file
        tifffile.imsave('results/segm_prob_lab2_{:05d}.tiff'.format(i), output[2])


.. _sphx_glr_download_auto_examples_apply_segm.py:


.. only :: html

 .. container:: sphx-glr-footer
    :class: sphx-glr-footer-example



  .. container:: sphx-glr-download

     :download:`Download Python source code: apply_segm.py <apply_segm.py>`



  .. container:: sphx-glr-download

     :download:`Download Jupyter notebook: apply_segm.ipynb <apply_segm.ipynb>`


.. only:: html

 .. rst-class:: sphx-glr-signature

    `Gallery generated by Sphinx-Gallery <https://sphinx-gallery.github.io>`_