unittest in models/

README.md

@@ -8,4 +8,5 @@ In development. No document, don't use.
 + Provide callbacks to control training behavior (as in [Keras](http://keras.io)).
 + Use `Dataflow` to gain fine-grained control on data preprocessing.
 + Automatically use the Queue operator in tensorflow to speed up input.
-+ Training and testing graph are modeled together, automatically. Just need to follow the conventions to setup stuffs.
++ Training and testing graph are modeled together. Just need to follow the conventions to setup stuffs.
++ Use tensorboard easily.

example_mnist.py

@@ -25,7 +25,7 @@ CAPACITY = MIN_AFTER_DEQUEUE + 3 * BATCH_SIZE
 def get_model(inputs, is_training):
     """
     Args:
-        inputs: a list of input variable,
+        inputs: a list of input variable for training
         e.g.: [image_var, label_var] with:
             image_var: bx28x28
             label_var: bx1 integer

tensorpack/dataflow/README.md

 
 ## Dataflow
 
-Dataflow: use `yield` to build a efficient data processing pipeline.
+Dataflow: use `yield` to build a simple and efficient data processing pipeline.
 
 NOTE: Dataflow aims to be independent of tensorflow.
 It should be useful for other python-based learning libraries.
+
+Documents to be done.

tensorpack/models/_test.py

+#!/usr/bin/env python2
+# -*- coding: UTF-8 -*-
+# File: _test.py
+# Author: Yuxin Wu <ppwwyyxx@gmail.com>
+
+import tensorflow as tf
+import numpy as np
+from . import *
+import unittest
+
+subs = unittest.TestCase.__subclasses__()
+
+def run_test_case(case):
+    suite = unittest.TestLoader().loadTestsFromTestCase(case)
+    unittest.TextTestRunner(verbosity=2).run(suite)
+
+for cls in subs:
+    if 'tensorpack.models' in str(cls):
+        run_test_case(cls)
+
+

tensorpack/models/pool.py

@@ -57,3 +57,23 @@ def FixedUnPooling(x, shape, unpool_mat=None):
                             input_shape[2] * shape[1],
                             input_shape[3]])
     return prod
+
+import unittest
+class TestPool(unittest.TestCase):
+    def test_fixed_unpooling(self):
+        h, w = 3, 4
+        mat = np.random.rand(h, w).astype('float32')
+        inp = tf.Variable(mat)
+        inp = tf.reshape(inp, [1, h, w, 1])
+        output = FixedUnPooling('unpool', inp, 2)
+        sess = tf.Session()
+        sess.run(tf.initialize_all_variables())
+        res = sess.run([output])[0]
+        self.assertEqual(res.shape, (1, 2*h, 2*w, 1))
+
+        # mat is on cornser
+        ele = res[0,::2,::2,0]
+        self.assertTrue((ele == mat).all())
+        # the rest are zeros
+        res[0,::2,::2,0] = 0
+        self.assertTrue((res == 0).all())