working on the docs; adding jupyter notebook for image classify

convnet/utils.py

@@ -62,7 +62,9 @@
 __all__ = ['nor_data',
            'check_random_state',
            'extract_patches',
-           'reconstruct_patches']
+           'reconstruct_patches',
+           'img_window',
+           'extract_3d']
 
 
 def nor_data(img):

doc/demo/classify_evaluate.py

@@ -0,0 +1,66 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+"""
+Example script 
+"""
+
+from __future__ import print_function
+import dxchange
+import numpy as np
+from convnet.utils import nor_data
+from convnet.utils import extract_3d
+from convnet.utils import img_window
+from convnet.classify import model
+import matplotlib.pyplot as plt
+import time
+import glob
+
+np.random.seed(1337)
+
+dim_img = 128
+patch_size = (dim_img, dim_img)
+batch_size = 50
+nb_classes = 2
+nb_epoch = 12
+
+# number of convolutional filters to use
+nb_filters = 32
+# size of pooling area for max pooling
+nb_pool = 2
+# convolution kernel size
+nb_conv = 3
+nb_evl = 100
+
+start_time = time.time()
+fnames = glob.glob('/local/decarlo/data/databank/xiaogang/center/test_05/Hornby_APS_1000_1058_1/*.tiff')
+fnames = np.sort(fnames)
+
+mdl = model(dim_img, nb_filters, nb_conv, nb_classes)
+
+mdl.load_weights('classify_training_weights.h5')
+print('The model loading time is %s seconds'%(time.time()-start_time))
+start_time = time.time()
+Y_score = np.zeros((len(fnames)))
+
+for i in range(len(fnames)):
+    img = dxchange.read_tiff(fnames[i])
+    img = nor_data(img)
+    X_evl = np.zeros((nb_evl, dim_img, dim_img))
+
+    for j in range(nb_evl):
+        X_evl[j] = img_window(img[360:1460, 440:1440], dim_img)
+    X_evl = X_evl.reshape(X_evl.shape[0], 1, dim_img, dim_img)
+    Y_evl = mdl.predict(X_evl, batch_size=batch_size)
+    Y_score[i] = sum(np.dot(Y_evl, [0, 1]))
+    #print('The evaluate score is:', Y_score[i])
+    #Y_score = sum(np.round(Y_score))/len(Y_score)
+
+
+ind_max = np.argmax(Y_score)
+print('The well-centered reconstruction is:', fnames[ind_max])
+print('The prediction runs for %s seconds'%(time.time()-start_time))
+plt.plot(Y_score)
+plt.show()
+
+

doc/demo/classify_train.py

@@ -0,0 +1,67 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+"""
+Example script 
+"""
+
+from __future__ import print_function
+import dxchange
+import numpy as np
+from convnet.utils import nor_data
+from convnet.utils import extract_3d
+from convnet.utils import img_window
+from convnet.classify import train
+
+np.random.seed(1337)
+
+dim_img = 128
+patch_size = (dim_img, dim_img)
+batch_size = 50
+nb_classes = 2
+nb_epoch = 12
+
+# number of convolutional filters to use
+nb_filters = 32
+# size of pooling area for max pooling
+nb_pool = 2
+# convolution kernel size
+nb_conv = 3
+
+
+fname = '/local/decarlo/data/databank/xiaogang/center/test_01/Blakely_SLS_1038_1058_1/1038.tiff'
+ind_uncenter1 = range(1038, 1047)
+ind_uncenter2 = range(1049, 1057)
+uncenter1 = dxchange.read_tiff_stack(fname, ind=ind_uncenter1, digit=4)
+uncenter2 = dxchange.read_tiff_stack(fname, ind=ind_uncenter2, digit=4)
+uncenter = np.concatenate((uncenter1, uncenter2), axis=0)
+uncenter = nor_data(uncenter)
+print (uncenter.shape)
+uncenter = img_window(uncenter[:, 360:1460, 440:1440], 200)
+print (uncenter.shape)
+uncenter_patches = extract_3d(uncenter, patch_size, 1)
+np.random.shuffle(uncenter_patches)
+print (uncenter_patches.shape)
+# print uncenter_patches.shape
+center_img = dxchange.read_tiff('/local/decarlo/data/databank/xiaogang/center/test_01/Blakely_SLS_1038_1058_1/1048.tiff')
+center_img = nor_data(center_img)
+print (center_img.shape)
+center_img = img_window(center_img[360:1460, 440:1440], 400)
+center_patches = extract_3d(center_img, patch_size, 1)
+np.random.shuffle(center_patches)
+print (center_patches.shape)
+# plt.imshow(center_img, cmap='gray', interpolation= None)
+# plt.show()
+
+x_train = np.concatenate((uncenter_patches[0:50000], center_patches[0:50000]), axis=0)
+x_test = np.concatenate((uncenter_patches[50000:60000], center_patches[50000:60000]), axis=0)
+x_train = x_train.reshape(x_train.shape[0], 1, dim_img, dim_img)
+x_test = x_test.reshape(x_test.shape[0], 1, dim_img, dim_img)
+y_train = np.zeros(100000)
+y_train[50000:99999] = 1
+y_test = np.zeros(20000)
+y_test[10000:19999] = 1
+
+model = train(x_train, y_train, x_test, y_test, dim_img, nb_filters, nb_conv, batch_size, nb_epoch, nb_classes)
+model.save_weights('classify_training_weights.h5')
+

doc/source/api.rst

@@ -6,6 +6,7 @@ API reference
 .. toctree::
 
    api/convnet.transform
+   api/convnet.classify
    api/convnet.utils
 
 

doc/source/api/convnet.classify.rst

@@ -0,0 +1,15 @@
+:mod:`convnet.classify`
+========================
+
+.. automodule:: convnet.classify
+   :members:
+   :show-inheritance:
+   :undoc-members:
+
+   .. rubric:: **Functions:**
+
+   .. autosummary::
+
+      model
+      train
+

doc/source/api/convnet.utils.rst

@@ -14,4 +14,7 @@
       check_random_state
       extract_patches
       reconstruct_patches
+      img_window
+      extract_3d
+
 

doc/source/demo.rst

@@ -13,11 +13,6 @@ scripts from `here <https://github.com/XiaogangYangAPS/convnet/tree/master/doc/d
 
 .. toctree::
 
-   ipynb/transform_train.rst
-   ipynb/transform_predict.rst
-
-.. automodule:: convnet
-   :members:
-   :undoc-members:
-   :show-inheritance: 
+   ipynb/transform.rst
+   ipynb/classify.rst
 

doc/source/ipynb/classify.rst

@@ -0,0 +1,12 @@
+Classify
+--------
+
+.. toctree::
+
+   classify_train.rst
+   classify_evaluate.rst
+
+.. automodule:: convnet
+   :members:
+   :undoc-members:
+   :show-inheritance: 

doc/source/ipynb/classify_evaluate.rst

@@ -0,0 +1,57 @@
+Evaluate
+--------
+
+Here is an example on how to use an already trained convolutional neural netwrok
+on a new image. The output will be estract features based on the training received.
+
+You can download the python scritp :download:`here<../../../doc/demo/transform_predict.py>`
+or the Jupyter notebook :download:`here<../../../doc/demo/transform_predict.ipynb>`
+
+.. code:: python
+
+    %pylab inline
+
+
+.. parsed-literal::
+
+    Populating the interactive namespace from numpy and matplotlib
+
+
+.. code:: python
+
+    import dxchange
+
+Image data I/O in ConvNet is supported by 
+`DXchange <http://dxchange.readthedocs.io>`__.
+
+.. code:: python
+
+    import matplotlib.pyplot as plt
+
+matplotlib provide plotting of the result in this notebook.
+
+`Install <http://convnet.readthedocs.io/en/latest/install.html>`__ ConvNet
+then:
+
+.. code:: python
+
+    import numpy as np
+    from convnet.utils import nor_data
+    from convnet.utils import extract_3d
+    from convnet.utils import img_window
+    from convnet.classify import model
+    import matplotlib.pyplot as plt
+    import time
+    import glob
+
+.. code:: python
+
+    np.random.seed(1337)
+
+    dim_img = 128
+    patch_size = (dim_img, dim_img)
+    batch_size = 50
+    nb_classes = 2
+    nb_epoch = 12
+    
+to be completed

doc/source/ipynb/classify_train.rst

@@ -0,0 +1,61 @@
+Train
+-----
+
+Here is an example on how to train a convolutional neural network.
+
+You can download the python scritp :download:`here<../../../doc/demo/transform_train.py>`
+or the Jupyter notebook :download:`here<../../../doc/demo/transform_train.ipynb>`
+
+.. code:: python
+
+    %pylab inline
+
+
+.. parsed-literal::
+
+    Populating the interactive namespace from numpy and matplotlib
+
+
+.. code:: python
+
+    import dxchange
+
+Image data I/O in ConvNet is supported by 
+`DXchange <http://dxchange.readthedocs.io>`__.
+
+.. code:: python
+
+    import matplotlib.pyplot as plt
+
+matplotlib provide plotting of the result in this notebook.
+
+`Install <http://convnet.readthedocs.io/en/latest/install.html>`__ ConvNet
+then:
+
+.. code:: python
+
+    import numpy as np
+    from convnet.utils import nor_data
+    from convnet.utils import extract_3d
+    from convnet.utils import img_window
+    from convnet.classify import train
+
+.. code:: python
+
+    np.random.seed(1337)
+    dim_img = 128
+    patch_size = (dim_img, dim_img)
+    batch_size = 50
+    nb_classes = 2
+    nb_epoch = 12
+
+.. code:: python
+
+    nb_filters = 32
+    nb_pool = 2
+    nb_conv = 3
+
+    img_y = dxchange.read_tiff('../../test/test_data/training_output.tiff')
+
+
+to be completed

doc/source/ipynb/transform.rst

@@ -0,0 +1,12 @@
+Transform
+---------
+
+.. toctree::
+
+   transform_train.rst
+   transform_predict.rst
+
+.. automodule:: convnet
+   :members:
+   :undoc-members:
+   :show-inheritance: 

doc/source/ipynb/transform_predict.rst

@@ -1,5 +1,5 @@
-Transform Predict
------------------
+Predict
+-------
 
 Here is an example on how to use an already trained convolutional neural netwrok
 on a new image. The output will be estract features based on the training received.

doc/source/ipynb/transform_train.rst

@@ -1,5 +1,5 @@
-Transform Training
-------------------
+Train
+-----
 
 Here is an example on how to train a convolutional neural network.