.. _imagesegmentationrst:

=================================
Segmenter une image pixel à pixel
=================================


.. only:: html

    **Links:** :download:`notebook <image_segmentation.ipynb>`, :downloadlink:`html <image_segmentation2html.html>`, :download:`PDF <image_segmentation.pdf>`, :download:`python <image_segmentation.py>`, :downloadlink:`slides <image_segmentation.slides.html>`, :githublink:`GitHub|_doc/notebooks/ai/image_segmentation.ipynb|*`


Il est aujourd’hui assez facile de construire une boîte englobante
autour d’un visage dans une image. Le deep learning permet d’extraire
précisément les pixels du visage.

.. code:: 

    from jyquickhelper import add_notebook_menu
    add_notebook_menu()






.. contents::
    :local:





.. code:: 

    %matplotlib inline

Chargement du modèle
--------------------

.. code:: 

    from code_beatrix.ai import DLImageSegmentation
    model = DLImageSegmentation(fLOG=print)


.. parsed-literal::
    [DLImageSegmentation] download model 'FCN8s'
    [DLImageSegmentation] [C:\Users\xavie/data/models/chainer/fcn8s_from_caffe.npz] Checking md5 (256c2a8235c1c65e62e48d3284fbd384)
    [DLImageSegmentation] load_npz 'C:\Users\xavie/data/models/chainer/fcn8s_from_caffe.npz'
    [DLImageSegmentation] class_name '['background' 'aeroplane' 'bicycle' 'bird' 'boat' 'bottle' 'bus' 'car'
     'cat' 'chair' 'cow' 'diningtable' 'dog' 'horse' 'motorbike' 'person'
     'potted plant' 'sheep' 'sofa' 'train' 'tv/monitor']'
    [DLImageSegmentation] cpu


Sur une petite image
--------------------

.. code:: 

    img = 'images/Tesla_circa_1890c.jpg'

.. code:: 

    feat, pred = model.predict(img)
    pred.shape




.. parsed-literal::
    (295, 220)



.. code:: 

    viz = model.plot(img, pred)  # img ou feat


.. parsed-literal::
    c:\Python363_x64\lib\site-packages\skimage\transform\_warps.py:84: UserWarning: The default mode, 'constant', will be changed to 'reflect' in skimage 0.15.
      warn("The default mode, 'constant', will be changed to 'reflect' in "


.. code:: 

    import skimage.io as skio
    skio.imshow(viz)




.. parsed-literal::
    <matplotlib.image.AxesImage at 0x18000a1a8d0>




.. image:: image_segmentation_9_1.png


Sur une image dont on change la taille
--------------------------------------

.. code:: 

    from PIL import Image
    img = 'images/Tesla_circa_1890c.jpg'
    pilimg = Image.open(img)

.. code:: 

    si = pilimg.size
    pilimg2 = pilimg.resize((si[0]//2, si[1]//2))

.. code:: 

    from skimage.io._plugins.pil_plugin import pil_to_ndarray

.. code:: 

    skimg = pil_to_ndarray(pilimg2)
    skimg.shape




.. parsed-literal::
    (147, 110, 3)



.. code:: 

    feat, pred = model.predict(skimg)
    pred.shape




.. parsed-literal::
    (147, 110)



.. code:: 

    viz = model.plot(skimg, pred)


.. parsed-literal::
    c:\Python363_x64\lib\site-packages\skimage\transform\_warps.py:84: UserWarning: The default mode, 'constant', will be changed to 'reflect' in skimage 0.15.
      warn("The default mode, 'constant', will be changed to 'reflect' in "


.. code:: 

    skio.imshow(viz)




.. parsed-literal::
    <matplotlib.image.AxesImage at 0x18001a8bd68>




.. image:: image_segmentation_17_1.png


Sur une grande image
--------------------

.. code:: 

    img = 'images/h2015_2.jpg'

.. code:: 

    pilimg = Image.open(img)

.. code:: 

    si = pilimg.size
    pilimg2 = pilimg.resize((si[0]//2, si[1]//2))

.. code:: 

    skimg = pil_to_ndarray(pilimg2)
    skimg.shape




.. parsed-literal::
    (456, 684, 3)



.. code:: 

    skio.imshow(skimg)




.. parsed-literal::
    <matplotlib.image.AxesImage at 0x1800210b9e8>




.. image:: image_segmentation_23_1.png


.. code:: 

    feat, pred = model.predict(skimg)
    pred.shape




.. parsed-literal::
    (456, 684)



.. code:: 

    viz = model.plot(feat, pred)


.. parsed-literal::
    c:\Python363_x64\lib\site-packages\skimage\transform\_warps.py:84: UserWarning: The default mode, 'constant', will be changed to 'reflect' in skimage 0.15.
      warn("The default mode, 'constant', will be changed to 'reflect' in "


.. code:: 

    import matplotlib.pyplot as plt
    fig, ax = plt.subplots(1, 1, figsize=(14, 12))
    ax.imshow(viz)




.. parsed-literal::
    <matplotlib.image.AxesImage at 0x1800216b9e8>




.. image:: image_segmentation_26_1.png