Merge pull request #1 from tensorflow/master

update to tensorflow/model master

slim/datasets/download_and_convert_mnist.py

@@ -125,7 +125,7 @@ def _add_to_tfrecord(data_filename, labels_filename, num_images,
         png_string = sess.run(encoded_png, feed_dict={image: images[j]})
 
         example = dataset_utils.image_to_tfexample(
-            png_string, 'png', _IMAGE_SIZE, _IMAGE_SIZE, labels[j])
+            png_string, 'png'.encode(), _IMAGE_SIZE, _IMAGE_SIZE, labels[j])
         tfrecord_writer.write(example.SerializeToString())
 
 
@@ -165,7 +165,7 @@ def _download_dataset(dataset_dir):
                                                _progress)
       print()
       with tf.gfile.GFile(filepath) as f:
-        size = f.Size()
+        size = f.size()
       print('Successfully downloaded', filename, size, 'bytes.')
 
 

slim/datasets/mnist.py

@@ -15,7 +15,7 @@
 """Provides data for the MNIST dataset.
 
 The dataset scripts used to create the dataset can be found at:
-tensorflow/models/slim/data/create_mnist_dataset.py
+tensorflow/models/slim/datasets/download_and_convert_mnist.py
 """
 
 from __future__ import absolute_import

slim/deployment/model_deploy.py

@@ -232,11 +232,9 @@ def _gather_clone_loss(clone, num_clones, regularization_losses):
       sum_loss = tf.add_n(all_losses)
   # Add the summaries out of the clone device block.
   if clone_loss is not None:
-    tf.scalar_summary(clone.scope + '/clone_loss', clone_loss,
-                      name='clone_loss')
+    tf.summary.scalar(clone.scope + '/clone_loss', clone_loss)
   if regularization_loss is not None:
-    tf.scalar_summary('regularization_loss', regularization_loss,
-                      name='regularization_loss')
+    tf.summary.scalar('regularization_loss', regularization_loss)
   return sum_loss
 
 
@@ -306,7 +304,7 @@ def optimize_clones(clones, optimizer,
       regularization_losses = None
   # Compute the total_loss summing all the clones_losses.
   total_loss = tf.add_n(clones_losses, name='total_loss')
-  # Sum the gradients accross clones.
+  # Sum the gradients across clones.
   grads_and_vars = _sum_clones_gradients(grads_and_vars)
   return total_loss, grads_and_vars
 
@@ -380,8 +378,8 @@ def deploy(config,
         update_ops.append(grad_updates)
 
         update_op = tf.group(*update_ops)
-        train_op = control_flow_ops.with_dependencies([update_op], total_loss,
-                                                      name='train_op')
+        with tf.control_dependencies([update_op]):
+          train_op = tf.identity(total_loss, name='train_op')
     else:
       clones_losses = []
       regularization_losses = tf.get_collection(
@@ -404,12 +402,11 @@ def deploy(config,
 
     if total_loss is not None:
       # Add total_loss to summary.
-      summaries.add(tf.scalar_summary('total_loss', total_loss,
-                                      name='total_loss'))
+      summaries.add(tf.summary.scalar('total_loss', total_loss))
 
     if summaries:
       # Merge all summaries together.
-      summary_op = tf.merge_summary(list(summaries), name='summary_op')
+      summary_op = tf.summary.merge(list(summaries), name='summary_op')
     else:
       summary_op = None
 
@@ -467,9 +464,9 @@ def _add_gradients_summaries(grads_and_vars):
         grad_values = grad.values
       else:
         grad_values = grad
-      summaries.append(tf.histogram_summary(var.op.name + ':gradient',
+      summaries.append(tf.summary.histogram(var.op.name + ':gradient',
                                             grad_values))
-      summaries.append(tf.histogram_summary(var.op.name + ':gradient_norm',
+      summaries.append(tf.summary.histogram(var.op.name + ':gradient_norm',
                                             tf.global_norm([grad_values])))
     else:
       tf.logging.info('Var %s has no gradient', var.op.name)
@@ -666,7 +663,7 @@ class DeploymentConfig(object):
         if op.device:
           return op.device
         node_def = op if isinstance(op, tf.NodeDef) else op.node_def
-        if node_def.op == 'Variable':
+        if node_def.op.startswith('Variable'):
           t = self._task
           self._task = (self._task + 1) % self._tasks
           d = '%s/task:%d' % (self._device, t)

slim/eval_image_classifier.py

@@ -153,14 +153,14 @@ def main(_):
     # Define the metrics:
     names_to_values, names_to_updates = slim.metrics.aggregate_metric_map({
         'Accuracy': slim.metrics.streaming_accuracy(predictions, labels),
-        'Recall@5': slim.metrics.streaming_recall_at_k(
+        'Recall_5': slim.metrics.streaming_recall_at_k(
             logits, labels, 5),
     })
 
     # Print the summaries to screen.
-    for name, value in names_to_values.iteritems():
+    for name, value in names_to_values.items():
       summary_name = 'eval/%s' % name
-      op = tf.scalar_summary(summary_name, value, collections=[])
+      op = tf.summary.scalar(summary_name, value, collections=[])
       op = tf.Print(op, [value], summary_name)
       tf.add_to_collection(tf.GraphKeys.SUMMARIES, op)
 
@@ -183,7 +183,7 @@ def main(_):
         checkpoint_path=checkpoint_path,
         logdir=FLAGS.eval_dir,
         num_evals=num_batches,
-        eval_op=names_to_updates.values(),
+        eval_op=list(names_to_updates.values()),
         variables_to_restore=variables_to_restore)
 
 

slim/nets/alexnet.py

@@ -113,7 +113,7 @@ def alexnet_v2(inputs,
         net = slim.conv2d(net, num_classes, [1, 1],
                           activation_fn=None,
                           normalizer_fn=None,
-                          biases_initializer=tf.zeros_initializer,
+                          biases_initializer=tf.zeros_initializer(),
                           scope='fc8')
 
       # Convert end_points_collection into a end_point dict.

slim/nets/cifarnet.py

@@ -77,7 +77,7 @@ def cifarnet(images, num_classes=10, is_training=False,
     net = slim.fully_connected(net, 192, scope='fc4')
     end_points['fc4'] = net
     logits = slim.fully_connected(net, num_classes,
-                                  biases_initializer=tf.zeros_initializer,
+                                  biases_initializer=tf.zeros_initializer(),
                                   weights_initializer=trunc_normal(1/192.0),
                                   weights_regularizer=None,
                                   activation_fn=None,

slim/nets/inception_resnet_v2.py

@@ -171,7 +171,7 @@ def inception_resnet_v2(inputs, num_classes=1001, is_training=True,
         end_points['Mixed_5b'] = net
         net = slim.repeat(net, 10, block35, scale=0.17)
 
-        # 17 x 17 x 1024
+        # 17 x 17 x 1088
         with tf.variable_scope('Mixed_6a'):
           with tf.variable_scope('Branch_0'):
             tower_conv = slim.conv2d(net, 384, 3, stride=2, padding='VALID',
@@ -191,7 +191,7 @@ def inception_resnet_v2(inputs, num_classes=1001, is_training=True,
         end_points['Mixed_6a'] = net
         net = slim.repeat(net, 20, block17, scale=0.10)
 
-        # Auxillary tower
+        # Auxiliary tower
         with tf.variable_scope('AuxLogits'):
           aux = slim.avg_pool2d(net, 5, stride=3, padding='VALID',
                                 scope='Conv2d_1a_3x3')

slim/nets/inception_resnet_v2_test.py

@@ -65,9 +65,9 @@ class InceptionTest(tf.test.TestCase):
         inception.inception_resnet_v2(inputs, num_classes)
       with tf.variable_scope('on_gpu'), tf.device('/gpu:0'):
         inception.inception_resnet_v2(inputs, num_classes)
-      for v in tf.get_collection(tf.GraphKeys.VARIABLES, scope='on_cpu'):
+      for v in tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope='on_cpu'):
         self.assertDeviceEqual(v.device, '/cpu:0')
-      for v in tf.get_collection(tf.GraphKeys.VARIABLES, scope='on_gpu'):
+      for v in tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope='on_gpu'):
         self.assertDeviceEqual(v.device, '/gpu:0')
 
   def testHalfSizeImages(self):

slim/nets/inception_v1.py

@@ -93,7 +93,7 @@ def inception_v1_base(inputs,
           with tf.variable_scope('Branch_3'):
             branch_3 = slim.max_pool2d(net, [3, 3], scope='MaxPool_0a_3x3')
             branch_3 = slim.conv2d(branch_3, 32, [1, 1], scope='Conv2d_0b_1x1')
-          net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+          net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if final_endpoint == end_point: return net, end_points
 
@@ -110,7 +110,7 @@ def inception_v1_base(inputs,
           with tf.variable_scope('Branch_3'):
             branch_3 = slim.max_pool2d(net, [3, 3], scope='MaxPool_0a_3x3')
             branch_3 = slim.conv2d(branch_3, 64, [1, 1], scope='Conv2d_0b_1x1')
-          net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+          net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if final_endpoint == end_point: return net, end_points
 
@@ -132,7 +132,7 @@ def inception_v1_base(inputs,
           with tf.variable_scope('Branch_3'):
             branch_3 = slim.max_pool2d(net, [3, 3], scope='MaxPool_0a_3x3')
             branch_3 = slim.conv2d(branch_3, 64, [1, 1], scope='Conv2d_0b_1x1')
-          net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+          net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if final_endpoint == end_point: return net, end_points
 
@@ -149,7 +149,7 @@ def inception_v1_base(inputs,
           with tf.variable_scope('Branch_3'):
             branch_3 = slim.max_pool2d(net, [3, 3], scope='MaxPool_0a_3x3')
             branch_3 = slim.conv2d(branch_3, 64, [1, 1], scope='Conv2d_0b_1x1')
-          net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+          net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if final_endpoint == end_point: return net, end_points
 
@@ -166,7 +166,7 @@ def inception_v1_base(inputs,
           with tf.variable_scope('Branch_3'):
             branch_3 = slim.max_pool2d(net, [3, 3], scope='MaxPool_0a_3x3')
             branch_3 = slim.conv2d(branch_3, 64, [1, 1], scope='Conv2d_0b_1x1')
-          net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+          net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if final_endpoint == end_point: return net, end_points
 
@@ -183,7 +183,7 @@ def inception_v1_base(inputs,
           with tf.variable_scope('Branch_3'):
             branch_3 = slim.max_pool2d(net, [3, 3], scope='MaxPool_0a_3x3')
             branch_3 = slim.conv2d(branch_3, 64, [1, 1], scope='Conv2d_0b_1x1')
-          net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+          net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if final_endpoint == end_point: return net, end_points
 
@@ -200,7 +200,7 @@ def inception_v1_base(inputs,
           with tf.variable_scope('Branch_3'):
             branch_3 = slim.max_pool2d(net, [3, 3], scope='MaxPool_0a_3x3')
             branch_3 = slim.conv2d(branch_3, 128, [1, 1], scope='Conv2d_0b_1x1')
-          net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+          net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if final_endpoint == end_point: return net, end_points
 
@@ -222,7 +222,7 @@ def inception_v1_base(inputs,
           with tf.variable_scope('Branch_3'):
             branch_3 = slim.max_pool2d(net, [3, 3], scope='MaxPool_0a_3x3')
             branch_3 = slim.conv2d(branch_3, 128, [1, 1], scope='Conv2d_0b_1x1')
-          net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+          net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if final_endpoint == end_point: return net, end_points
 
@@ -239,7 +239,7 @@ def inception_v1_base(inputs,
           with tf.variable_scope('Branch_3'):
             branch_3 = slim.max_pool2d(net, [3, 3], scope='MaxPool_0a_3x3')
             branch_3 = slim.conv2d(branch_3, 128, [1, 1], scope='Conv2d_0b_1x1')
-          net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+          net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if final_endpoint == end_point: return net, end_points
     raise ValueError('Unknown final endpoint %s' % final_endpoint)
@@ -270,7 +270,7 @@ def inception_v1(inputs,
     is_training: whether is training or not.
     dropout_keep_prob: the percentage of activation values that are retained.
     prediction_fn: a function to get predictions out of logits.
-    spatial_squeeze: if True, logits is of shape is [B, C], if false logits is
+    spatial_squeeze: if True, logits is of shape [B, C], if false logits is
         of shape [B, 1, 1, C], where B is batch_size and C is number of classes.
     reuse: whether or not the network and its variables should be reused. To be
       able to reuse 'scope' must be given.
@@ -289,7 +289,7 @@ def inception_v1(inputs,
                         is_training=is_training):
       net, end_points = inception_v1_base(inputs, scope=scope)
       with tf.variable_scope('Logits'):
-        net = slim.avg_pool2d(net, [7, 7], stride=1, scope='MaxPool_0a_7x7')
+        net = slim.avg_pool2d(net, [7, 7], stride=1, scope='AvgPool_0a_7x7')
         net = slim.dropout(net,
                            dropout_keep_prob, scope='Dropout_0b')
         logits = slim.conv2d(net, num_classes, [1, 1], activation_fn=None,

slim/nets/inception_v2.py

@@ -145,7 +145,7 @@ def inception_v2_base(inputs,
               branch_3, depth(32), [1, 1],
               weights_initializer=trunc_normal(0.1),
               scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if end_point == final_endpoint: return net, end_points
       # 28 x 28 x 256
@@ -175,7 +175,7 @@ def inception_v2_base(inputs,
               branch_3, depth(64), [1, 1],
               weights_initializer=trunc_normal(0.1),
               scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if end_point == final_endpoint: return net, end_points
       # 28 x 28 x 320
@@ -200,7 +200,7 @@ def inception_v2_base(inputs,
         with tf.variable_scope('Branch_2'):
           branch_2 = slim.max_pool2d(
               net, [3, 3], stride=2, scope='MaxPool_1a_3x3')
-        net = tf.concat(3, [branch_0, branch_1, branch_2])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2])
         end_points[end_point] = net
         if end_point == final_endpoint: return net, end_points
       # 14 x 14 x 576
@@ -230,7 +230,7 @@ def inception_v2_base(inputs,
               branch_3, depth(128), [1, 1],
               weights_initializer=trunc_normal(0.1),
               scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if end_point == final_endpoint: return net, end_points
       # 14 x 14 x 576
@@ -260,7 +260,7 @@ def inception_v2_base(inputs,
               branch_3, depth(128), [1, 1],
               weights_initializer=trunc_normal(0.1),
               scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if end_point == final_endpoint: return net, end_points
       # 14 x 14 x 576
@@ -290,7 +290,7 @@ def inception_v2_base(inputs,
               branch_3, depth(96), [1, 1],
               weights_initializer=trunc_normal(0.1),
               scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if end_point == final_endpoint: return net, end_points
 
@@ -321,7 +321,7 @@ def inception_v2_base(inputs,
               branch_3, depth(96), [1, 1],
               weights_initializer=trunc_normal(0.1),
               scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if end_point == final_endpoint: return net, end_points
       # 14 x 14 x 576
@@ -346,7 +346,7 @@ def inception_v2_base(inputs,
         with tf.variable_scope('Branch_2'):
           branch_2 = slim.max_pool2d(net, [3, 3], stride=2,
                                      scope='MaxPool_1a_3x3')
-        net = tf.concat(3, [branch_0, branch_1, branch_2])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2])
         end_points[end_point] = net
         if end_point == final_endpoint: return net, end_points
       # 7 x 7 x 1024
@@ -376,7 +376,7 @@ def inception_v2_base(inputs,
               branch_3, depth(128), [1, 1],
               weights_initializer=trunc_normal(0.1),
               scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if end_point == final_endpoint: return net, end_points
 
@@ -407,7 +407,7 @@ def inception_v2_base(inputs,
               branch_3, depth(128), [1, 1],
               weights_initializer=trunc_normal(0.1),
               scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
         end_points[end_point] = net
         if end_point == final_endpoint: return net, end_points
     raise ValueError('Unknown final endpoint %s' % final_endpoint)
@@ -443,7 +443,7 @@ def inception_v2(inputs,
       usage will be to set this value in (0, 1) to reduce the number of
       parameters or computation cost of the model.
     prediction_fn: a function to get predictions out of logits.
-    spatial_squeeze: if True, logits is of shape is [B, C], if false logits is
+    spatial_squeeze: if True, logits is of shape [B, C], if false logits is
         of shape [B, 1, 1, C], where B is batch_size and C is number of classes.
     reuse: whether or not the network and its variables should be reused. To be
       able to reuse 'scope' must be given.

slim/nets/inception_v3.py

@@ -158,7 +158,7 @@ def inception_v3_base(inputs,
           branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
           branch_3 = slim.conv2d(branch_3, depth(32), [1, 1],
                                  scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
 
@@ -182,7 +182,7 @@ def inception_v3_base(inputs,
           branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
           branch_3 = slim.conv2d(branch_3, depth(64), [1, 1],
                                  scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
 
@@ -205,7 +205,7 @@ def inception_v3_base(inputs,
           branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
           branch_3 = slim.conv2d(branch_3, depth(64), [1, 1],
                                  scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
 
@@ -224,7 +224,7 @@ def inception_v3_base(inputs,
         with tf.variable_scope('Branch_2'):
           branch_2 = slim.max_pool2d(net, [3, 3], stride=2, padding='VALID',
                                      scope='MaxPool_1a_3x3')
-        net = tf.concat(3, [branch_0, branch_1, branch_2])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
 
@@ -253,7 +253,7 @@ def inception_v3_base(inputs,
           branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
           branch_3 = slim.conv2d(branch_3, depth(192), [1, 1],
                                  scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
 
@@ -282,7 +282,7 @@ def inception_v3_base(inputs,
           branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
           branch_3 = slim.conv2d(branch_3, depth(192), [1, 1],
                                  scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
       # mixed_6: 17 x 17 x 768.
@@ -310,7 +310,7 @@ def inception_v3_base(inputs,
           branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
           branch_3 = slim.conv2d(branch_3, depth(192), [1, 1],
                                  scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
 
@@ -339,7 +339,7 @@ def inception_v3_base(inputs,
           branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
           branch_3 = slim.conv2d(branch_3, depth(192), [1, 1],
                                  scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
 
@@ -361,7 +361,7 @@ def inception_v3_base(inputs,
         with tf.variable_scope('Branch_2'):
           branch_2 = slim.max_pool2d(net, [3, 3], stride=2, padding='VALID',
                                      scope='MaxPool_1a_3x3')
-        net = tf.concat(3, [branch_0, branch_1, branch_2])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
       # mixed_9: 8 x 8 x 2048.
@@ -371,21 +371,21 @@ def inception_v3_base(inputs,
           branch_0 = slim.conv2d(net, depth(320), [1, 1], scope='Conv2d_0a_1x1')
         with tf.variable_scope('Branch_1'):
           branch_1 = slim.conv2d(net, depth(384), [1, 1], scope='Conv2d_0a_1x1')
-          branch_1 = tf.concat(3, [
+          branch_1 = tf.concat(axis=3, values=[
               slim.conv2d(branch_1, depth(384), [1, 3], scope='Conv2d_0b_1x3'),
               slim.conv2d(branch_1, depth(384), [3, 1], scope='Conv2d_0b_3x1')])
         with tf.variable_scope('Branch_2'):
           branch_2 = slim.conv2d(net, depth(448), [1, 1], scope='Conv2d_0a_1x1')
           branch_2 = slim.conv2d(
               branch_2, depth(384), [3, 3], scope='Conv2d_0b_3x3')
-          branch_2 = tf.concat(3, [
+          branch_2 = tf.concat(axis=3, values=[
               slim.conv2d(branch_2, depth(384), [1, 3], scope='Conv2d_0c_1x3'),
               slim.conv2d(branch_2, depth(384), [3, 1], scope='Conv2d_0d_3x1')])
         with tf.variable_scope('Branch_3'):
           branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
           branch_3 = slim.conv2d(
               branch_3, depth(192), [1, 1], scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
 
@@ -396,21 +396,21 @@ def inception_v3_base(inputs,
           branch_0 = slim.conv2d(net, depth(320), [1, 1], scope='Conv2d_0a_1x1')
         with tf.variable_scope('Branch_1'):
           branch_1 = slim.conv2d(net, depth(384), [1, 1], scope='Conv2d_0a_1x1')
-          branch_1 = tf.concat(3, [
+          branch_1 = tf.concat(axis=3, values=[
               slim.conv2d(branch_1, depth(384), [1, 3], scope='Conv2d_0b_1x3'),
               slim.conv2d(branch_1, depth(384), [3, 1], scope='Conv2d_0c_3x1')])
         with tf.variable_scope('Branch_2'):
           branch_2 = slim.conv2d(net, depth(448), [1, 1], scope='Conv2d_0a_1x1')
           branch_2 = slim.conv2d(
               branch_2, depth(384), [3, 3], scope='Conv2d_0b_3x3')
-          branch_2 = tf.concat(3, [
+          branch_2 = tf.concat(axis=3, values=[
               slim.conv2d(branch_2, depth(384), [1, 3], scope='Conv2d_0c_1x3'),
               slim.conv2d(branch_2, depth(384), [3, 1], scope='Conv2d_0d_3x1')])
         with tf.variable_scope('Branch_3'):
           branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
           branch_3 = slim.conv2d(
               branch_3, depth(192), [1, 1], scope='Conv2d_0b_1x1')
-        net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+        net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
       end_points[end_point] = net
       if end_point == final_endpoint: return net, end_points
     raise ValueError('Unknown final endpoint %s' % final_endpoint)
@@ -453,7 +453,7 @@ def inception_v3(inputs,
       usage will be to set this value in (0, 1) to reduce the number of
       parameters or computation cost of the model.
     prediction_fn: a function to get predictions out of logits.
-    spatial_squeeze: if True, logits is of shape is [B, C], if false logits is
+    spatial_squeeze: if True, logits is of shape [B, C], if false logits is
         of shape [B, 1, 1, C], where B is batch_size and C is number of classes.
     reuse: whether or not the network and its variables should be reused. To be
       able to reuse 'scope' must be given.

slim/nets/inception_v4.py

@@ -49,7 +49,7 @@ def block_inception_a(inputs, scope=None, reuse=None):
       with tf.variable_scope('Branch_3'):
         branch_3 = slim.avg_pool2d(inputs, [3, 3], scope='AvgPool_0a_3x3')
         branch_3 = slim.conv2d(branch_3, 96, [1, 1], scope='Conv2d_0b_1x1')
-      return tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+      return tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
 
 
 def block_reduction_a(inputs, scope=None, reuse=None):
@@ -69,7 +69,7 @@ def block_reduction_a(inputs, scope=None, reuse=None):
       with tf.variable_scope('Branch_2'):
         branch_2 = slim.max_pool2d(inputs, [3, 3], stride=2, padding='VALID',
                                    scope='MaxPool_1a_3x3')
-      return tf.concat(3, [branch_0, branch_1, branch_2])
+      return tf.concat(axis=3, values=[branch_0, branch_1, branch_2])
 
 
 def block_inception_b(inputs, scope=None, reuse=None):
@@ -93,7 +93,7 @@ def block_inception_b(inputs, scope=None, reuse=None):
       with tf.variable_scope('Branch_3'):
         branch_3 = slim.avg_pool2d(inputs, [3, 3], scope='AvgPool_0a_3x3')
         branch_3 = slim.conv2d(branch_3, 128, [1, 1], scope='Conv2d_0b_1x1')
-      return tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+      return tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
 
 
 def block_reduction_b(inputs, scope=None, reuse=None):
@@ -115,7 +115,7 @@ def block_reduction_b(inputs, scope=None, reuse=None):
       with tf.variable_scope('Branch_2'):
         branch_2 = slim.max_pool2d(inputs, [3, 3], stride=2, padding='VALID',
                                    scope='MaxPool_1a_3x3')
-      return tf.concat(3, [branch_0, branch_1, branch_2])
+      return tf.concat(axis=3, values=[branch_0, branch_1, branch_2])
 
 
 def block_inception_c(inputs, scope=None, reuse=None):
@@ -128,20 +128,20 @@ def block_inception_c(inputs, scope=None, reuse=None):
         branch_0 = slim.conv2d(inputs, 256, [1, 1], scope='Conv2d_0a_1x1')
       with tf.variable_scope('Branch_1'):
         branch_1 = slim.conv2d(inputs, 384, [1, 1], scope='Conv2d_0a_1x1')
-        branch_1 = tf.concat(3, [
+        branch_1 = tf.concat(axis=3, values=[
             slim.conv2d(branch_1, 256, [1, 3], scope='Conv2d_0b_1x3'),
             slim.conv2d(branch_1, 256, [3, 1], scope='Conv2d_0c_3x1')])
       with tf.variable_scope('Branch_2'):
         branch_2 = slim.conv2d(inputs, 384, [1, 1], scope='Conv2d_0a_1x1')
         branch_2 = slim.conv2d(branch_2, 448, [3, 1], scope='Conv2d_0b_3x1')
         branch_2 = slim.conv2d(branch_2, 512, [1, 3], scope='Conv2d_0c_1x3')
-        branch_2 = tf.concat(3, [
+        branch_2 = tf.concat(axis=3, values=[
             slim.conv2d(branch_2, 256, [1, 3], scope='Conv2d_0d_1x3'),
             slim.conv2d(branch_2, 256, [3, 1], scope='Conv2d_0e_3x1')])
       with tf.variable_scope('Branch_3'):
         branch_3 = slim.avg_pool2d(inputs, [3, 3], scope='AvgPool_0a_3x3')
         branch_3 = slim.conv2d(branch_3, 256, [1, 1], scope='Conv2d_0b_1x1')
-      return tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
+      return tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
 
 
 def inception_v4_base(inputs, final_endpoint='Mixed_7d', scope=None):
@@ -192,7 +192,7 @@ def inception_v4_base(inputs, final_endpoint='Mixed_7d', scope=None):
         with tf.variable_scope('Branch_1'):
           branch_1 = slim.conv2d(net, 96, [3, 3], stride=2, padding='VALID',
                                  scope='Conv2d_0a_3x3')
-        net = tf.concat(3, [branch_0, branch_1])
+        net = tf.concat(axis=3, values=[branch_0, branch_1])
         if add_and_check_final('Mixed_3a', net): return net, end_points
 
       # 73 x 73 x 160
@@ -207,7 +207,7 @@ def inception_v4_base(inputs, final_endpoint='Mixed_7d', scope=None):
           branch_1 = slim.conv2d(branch_1, 64, [7, 1], scope='Conv2d_0c_7x1')
           branch_1 = slim.conv2d(branch_1, 96, [3, 3], padding='VALID',
                                  scope='Conv2d_1a_3x3')
-        net = tf.concat(3, [branch_0, branch_1])
+        net = tf.concat(axis=3, values=[branch_0, branch_1])
         if add_and_check_final('Mixed_4a', net): return net, end_points
 
       # 71 x 71 x 192
@@ -218,12 +218,12 @@ def inception_v4_base(inputs, final_endpoint='Mixed_7d', scope=None):
         with tf.variable_scope('Branch_1'):
           branch_1 = slim.max_pool2d(net, [3, 3], stride=2, padding='VALID',
                                      scope='MaxPool_1a_3x3')
-        net = tf.concat(3, [branch_0, branch_1])
+        net = tf.concat(axis=3, values=[branch_0, branch_1])
         if add_and_check_final('Mixed_5a', net): return net, end_points
 
       # 35 x 35 x 384
       # 4 x Inception-A blocks
-      for idx in xrange(4):
+      for idx in range(4):
         block_scope = 'Mixed_5' + chr(ord('b') + idx)
         net = block_inception_a(net, block_scope)
         if add_and_check_final(block_scope, net): return net, end_points
@@ -235,7 +235,7 @@ def inception_v4_base(inputs, final_endpoint='Mixed_7d', scope=None):
 
       # 17 x 17 x 1024
       # 7 x Inception-B blocks
-      for idx in xrange(7):
+      for idx in range(7):
         block_scope = 'Mixed_6' + chr(ord('b') + idx)
         net = block_inception_b(net, block_scope)
         if add_and_check_final(block_scope, net): return net, end_points
@@ -247,7 +247,7 @@ def inception_v4_base(inputs, final_endpoint='Mixed_7d', scope=None):
 
       # 8 x 8 x 1536
       # 3 x Inception-C blocks
-      for idx in xrange(3):
+      for idx in range(3):
         block_scope = 'Mixed_7' + chr(ord('b') + idx)
         net = block_inception_c(net, block_scope)
         if add_and_check_final(block_scope, net): return net, end_points
@@ -269,7 +269,7 @@ def inception_v4(inputs, num_classes=1001, is_training=True,
     reuse: whether or not the network and its variables should be reused. To be
       able to reuse 'scope' must be given.
     scope: Optional variable_scope.
-    create_aux_logits: Whether to include the auxilliary logits.
+    create_aux_logits: Whether to include the auxiliary logits.
 
   Returns:
     logits: the logits outputs of the model.

slim/nets/inception_v4_test.py

@@ -146,9 +146,9 @@ class InceptionTest(tf.test.TestCase):
       inception.inception_v4(inputs, num_classes)
     with tf.variable_scope('on_gpu'), tf.device('/gpu:0'):
       inception.inception_v4(inputs, num_classes)
-    for v in tf.get_collection(tf.GraphKeys.VARIABLES, scope='on_cpu'):
+    for v in tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope='on_cpu'):
       self.assertDeviceEqual(v.device, '/cpu:0')
-    for v in tf.get_collection(tf.GraphKeys.VARIABLES, scope='on_gpu'):
+    for v in tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope='on_gpu'):
       self.assertDeviceEqual(v.device, '/gpu:0')
 
   def testHalfSizeImages(self):

slim/nets/nets_factory.py

@@ -97,10 +97,10 @@ def get_network_fn(name, num_classes, weight_decay=0.0, is_training=False):
   """
   if name not in networks_map:
     raise ValueError('Name of network unknown %s' % name)
-  arg_scope = arg_scopes_map[name](weight_decay=weight_decay)
   func = networks_map[name]
   @functools.wraps(func)
   def network_fn(images):
+    arg_scope = arg_scopes_map[name](weight_decay=weight_decay)
     with slim.arg_scope(arg_scope):
       return func(images, num_classes, is_training=is_training)
   if hasattr(func, 'default_image_size'):

slim/nets/nets_factory_test.py

@@ -19,11 +19,12 @@ from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
 
-
 import tensorflow as tf
 
 from nets import nets_factory
 
+slim = tf.contrib.slim
+
 
 class NetworksTest(tf.test.TestCase):
 
@@ -42,5 +43,19 @@ class NetworksTest(tf.test.TestCase):
         self.assertEqual(logits.get_shape().as_list()[0], batch_size)
         self.assertEqual(logits.get_shape().as_list()[-1], num_classes)
 
+  def testGetNetworkFnArgScope(self):
+    batch_size = 5
+    num_classes = 10
+    net = 'cifarnet'
+    with self.test_session(use_gpu=True):
+      net_fn = nets_factory.get_network_fn(net, num_classes)
+      image_size = getattr(net_fn, 'default_image_size', 224)
+      with slim.arg_scope([slim.model_variable, slim.variable],
+                          device='/CPU:0'):
+        inputs = tf.random_uniform((batch_size, image_size, image_size, 3))
+        net_fn(inputs)
+      weights = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, 'CifarNet/conv1')[0]
+      self.assertDeviceEqual('/CPU:0', weights.device)
+
 if __name__ == '__main__':
   tf.test.main()

slim/nets/overfeat.py

@@ -41,7 +41,7 @@ def overfeat_arg_scope(weight_decay=0.0005):
   with slim.arg_scope([slim.conv2d, slim.fully_connected],
                       activation_fn=tf.nn.relu,
                       weights_regularizer=slim.l2_regularizer(weight_decay),
-                      biases_initializer=tf.zeros_initializer):
+                      biases_initializer=tf.zeros_initializer()):
     with slim.arg_scope([slim.conv2d], padding='SAME'):
       with slim.arg_scope([slim.max_pool2d], padding='VALID') as arg_sc:
         return arg_sc
@@ -107,7 +107,7 @@ def overfeat(inputs,
         net = slim.conv2d(net, num_classes, [1, 1],
                           activation_fn=None,
                           normalizer_fn=None,
-                          biases_initializer=tf.zeros_initializer,
+                          biases_initializer=tf.zeros_initializer(),
                           scope='fc8')
       # Convert end_points_collection into a end_point dict.
       end_points = slim.utils.convert_collection_to_dict(end_points_collection)

slim/nets/resnet_v1.py

@@ -119,6 +119,7 @@ def resnet_v1(inputs,
               global_pool=True,
               output_stride=None,
               include_root_block=True,
+              spatial_squeeze=True,
               reuse=None,
               scope=None):
   """Generator for v1 ResNet models.
@@ -158,6 +159,8 @@ def resnet_v1(inputs,
       ratio of input to output spatial resolution.
     include_root_block: If True, include the initial convolution followed by
       max-pooling, if False excludes it.
+    spatial_squeeze: if True, logits is of shape [B, C], if false logits is
+        of shape [B, 1, 1, C], where B is batch_size and C is number of classes.
     reuse: whether or not the network and its variables should be reused. To be
       able to reuse 'scope' must be given.
     scope: Optional variable_scope.
@@ -197,11 +200,15 @@ def resnet_v1(inputs,
         if num_classes is not None:
           net = slim.conv2d(net, num_classes, [1, 1], activation_fn=None,
                             normalizer_fn=None, scope='logits')
+        if spatial_squeeze:
+          logits = tf.squeeze(net, [1, 2], name='SpatialSqueeze')
+        else:
+          logits = net
         # Convert end_points_collection into a dictionary of end_points.
         end_points = slim.utils.convert_collection_to_dict(end_points_collection)
         if num_classes is not None:
-          end_points['predictions'] = slim.softmax(net, scope='predictions')
-        return net, end_points
+          end_points['predictions'] = slim.softmax(logits, scope='predictions')
+        return logits, end_points
 resnet_v1.default_image_size = 224
 
 
@@ -210,6 +217,7 @@ def resnet_v1_50(inputs,
                  is_training=True,
                  global_pool=True,
                  output_stride=None,
+                 spatial_squeeze=True,
                  reuse=None,
                  scope='resnet_v1_50'):
   """ResNet-50 model of [1]. See resnet_v1() for arg and return description."""
@@ -225,7 +233,9 @@ def resnet_v1_50(inputs,
   ]
   return resnet_v1(inputs, blocks, num_classes, is_training,
                    global_pool=global_pool, output_stride=output_stride,
-                   include_root_block=True, reuse=reuse, scope=scope)
+                   include_root_block=True, spatial_squeeze=spatial_squeeze,
+                   reuse=reuse, scope=scope)
+resnet_v1_50.default_image_size = resnet_v1.default_image_size
 
 
 def resnet_v1_101(inputs,
@@ -233,6 +243,7 @@ def resnet_v1_101(inputs,
                   is_training=True,
                   global_pool=True,
                   output_stride=None,
+                  spatial_squeeze=True,
                   reuse=None,
                   scope='resnet_v1_101'):
   """ResNet-101 model of [1]. See resnet_v1() for arg and return description."""
@@ -248,7 +259,9 @@ def resnet_v1_101(inputs,
   ]
   return resnet_v1(inputs, blocks, num_classes, is_training,
                    global_pool=global_pool, output_stride=output_stride,
-                   include_root_block=True, reuse=reuse, scope=scope)
+                   include_root_block=True, spatial_squeeze=spatial_squeeze,
+                   reuse=reuse, scope=scope)
+resnet_v1_101.default_image_size = resnet_v1.default_image_size
 
 
 def resnet_v1_152(inputs,
@@ -256,6 +269,7 @@ def resnet_v1_152(inputs,
                   is_training=True,
                   global_pool=True,
                   output_stride=None,
+                  spatial_squeeze=True,
                   reuse=None,
                   scope='resnet_v1_152'):
   """ResNet-152 model of [1]. See resnet_v1() for arg and return description."""
@@ -270,7 +284,9 @@ def resnet_v1_152(inputs,
           'block4', bottleneck, [(2048, 512, 1)] * 3)]
   return resnet_v1(inputs, blocks, num_classes, is_training,
                    global_pool=global_pool, output_stride=output_stride,
-                   include_root_block=True, reuse=reuse, scope=scope)
+                   include_root_block=True, spatial_squeeze=spatial_squeeze,
+                   reuse=reuse, scope=scope)
+resnet_v1_152.default_image_size = resnet_v1.default_image_size
 
 
 def resnet_v1_200(inputs,
@@ -278,6 +294,7 @@ def resnet_v1_200(inputs,
                   is_training=True,
                   global_pool=True,
                   output_stride=None,
+                  spatial_squeeze=True,
                   reuse=None,
                   scope='resnet_v1_200'):
   """ResNet-200 model of [2]. See resnet_v1() for arg and return description."""
@@ -292,4 +309,6 @@ def resnet_v1_200(inputs,
           'block4', bottleneck, [(2048, 512, 1)] * 3)]
   return resnet_v1(inputs, blocks, num_classes, is_training,
                    global_pool=global_pool, output_stride=output_stride,
-                   include_root_block=True, reuse=reuse, scope=scope)
+                   include_root_block=True, spatial_squeeze=spatial_squeeze,
+                   reuse=reuse, scope=scope)
+resnet_v1_200.default_image_size = resnet_v1.default_image_size

slim/nets/resnet_v2.py

@@ -25,8 +25,6 @@ introduced by:
 
 The key difference of the full preactivation 'v2' variant compared to the
 'v1' variant in [1] is the use of batch normalization before every weight layer.
-Another difference is that 'v2' ResNets do not include an activation function in
-the main pathway. Also see [2; Fig. 4e].
 
 Typical use:
 
@@ -117,6 +115,7 @@ def resnet_v2(inputs,
               global_pool=True,
               output_stride=None,
               include_root_block=True,
+              spatial_squeeze=True,
               reuse=None,
               scope=None):
   """Generator for v2 (preactivation) ResNet models.
@@ -157,6 +156,8 @@ def resnet_v2(inputs,
     include_root_block: If True, include the initial convolution followed by
       max-pooling, if False excludes it. If excluded, `inputs` should be the
       results of an activation-less convolution.
+    spatial_squeeze: if True, logits is of shape [B, C], if false logits is
+        of shape [B, 1, 1, C], where B is batch_size and C is number of classes.
     reuse: whether or not the network and its variables should be reused. To be
       able to reuse 'scope' must be given.
     scope: Optional variable_scope.
@@ -206,11 +207,15 @@ def resnet_v2(inputs,
         if num_classes is not None:
           net = slim.conv2d(net, num_classes, [1, 1], activation_fn=None,
                             normalizer_fn=None, scope='logits')
+        if spatial_squeeze:
+          logits = tf.squeeze(net, [1, 2], name='SpatialSqueeze')
+        else:
+          logits = net
         # Convert end_points_collection into a dictionary of end_points.
         end_points = slim.utils.convert_collection_to_dict(end_points_collection)
         if num_classes is not None:
-          end_points['predictions'] = slim.softmax(net, scope='predictions')
-        return net, end_points
+          end_points['predictions'] = slim.softmax(logits, scope='predictions')
+        return logits, end_points
 resnet_v2.default_image_size = 224
 
 
@@ -219,6 +224,7 @@ def resnet_v2_50(inputs,
                  is_training=True,
                  global_pool=True,
                  output_stride=None,
+                 spatial_squeeze=True,
                  reuse=None,
                  scope='resnet_v2_50'):
   """ResNet-50 model of [1]. See resnet_v2() for arg and return description."""
@@ -233,7 +239,9 @@ def resnet_v2_50(inputs,
           'block4', bottleneck, [(2048, 512, 1)] * 3)]
   return resnet_v2(inputs, blocks, num_classes, is_training=is_training,
                    global_pool=global_pool, output_stride=output_stride,
-                   include_root_block=True, reuse=reuse, scope=scope)
+                   include_root_block=True, spatial_squeeze=spatial_squeeze,
+                   reuse=reuse, scope=scope)
+resnet_v2_50.default_image_size = resnet_v2.default_image_size
 
 
 def resnet_v2_101(inputs,
@@ -241,6 +249,7 @@ def resnet_v2_101(inputs,
                   is_training=True,
                   global_pool=True,
                   output_stride=None,
+                  spatial_squeeze=True,
                   reuse=None,
                   scope='resnet_v2_101'):
   """ResNet-101 model of [1]. See resnet_v2() for arg and return description."""
@@ -255,7 +264,9 @@ def resnet_v2_101(inputs,
           'block4', bottleneck, [(2048, 512, 1)] * 3)]
   return resnet_v2(inputs, blocks, num_classes, is_training=is_training,
                    global_pool=global_pool, output_stride=output_stride,
-                   include_root_block=True, reuse=reuse, scope=scope)
+                   include_root_block=True, spatial_squeeze=spatial_squeeze,
+                   reuse=reuse, scope=scope)
+resnet_v2_101.default_image_size = resnet_v2.default_image_size
 
 
 def resnet_v2_152(inputs,
@@ -263,6 +274,7 @@ def resnet_v2_152(inputs,
                   is_training=True,
                   global_pool=True,
                   output_stride=None,
+                  spatial_squeeze=True,
                   reuse=None,
                   scope='resnet_v2_152'):
   """ResNet-152 model of [1]. See resnet_v2() for arg and return description."""
@@ -277,7 +289,9 @@ def resnet_v2_152(inputs,
           'block4', bottleneck, [(2048, 512, 1)] * 3)]
   return resnet_v2(inputs, blocks, num_classes, is_training=is_training,
                    global_pool=global_pool, output_stride=output_stride,
-                   include_root_block=True, reuse=reuse, scope=scope)
+                   include_root_block=True, spatial_squeeze=spatial_squeeze,
+                   reuse=reuse, scope=scope)
+resnet_v2_152.default_image_size = resnet_v2.default_image_size
 
 
 def resnet_v2_200(inputs,
@@ -285,6 +299,7 @@ def resnet_v2_200(inputs,
                   is_training=True,
                   global_pool=True,
                   output_stride=None,
+                  spatial_squeeze=True,
                   reuse=None,
                   scope='resnet_v2_200'):
   """ResNet-200 model of [2]. See resnet_v2() for arg and return description."""
@@ -299,4 +314,6 @@ def resnet_v2_200(inputs,
           'block4', bottleneck, [(2048, 512, 1)] * 3)]
   return resnet_v2(inputs, blocks, num_classes, is_training=is_training,
                    global_pool=global_pool, output_stride=output_stride,
-                   include_root_block=True, reuse=reuse, scope=scope)
+                   include_root_block=True, spatial_squeeze=spatial_squeeze,
+                   reuse=reuse, scope=scope)
+resnet_v2_200.default_image_size = resnet_v2.default_image_size

slim/nets/vgg.py

@@ -58,7 +58,7 @@ def vgg_arg_scope(weight_decay=0.0005):
   with slim.arg_scope([slim.conv2d, slim.fully_connected],
                       activation_fn=tf.nn.relu,
                       weights_regularizer=slim.l2_regularizer(weight_decay),
-                      biases_initializer=tf.zeros_initializer):
+                      biases_initializer=tf.zeros_initializer()):
     with slim.arg_scope([slim.conv2d], padding='SAME') as arg_sc:
       return arg_sc
 
@@ -68,7 +68,8 @@ def vgg_a(inputs,
           is_training=True,
           dropout_keep_prob=0.5,
           spatial_squeeze=True,
-          scope='vgg_a'):
+          scope='vgg_a',
+          fc_conv_padding='VALID'):
   """Oxford Net VGG 11-Layers version A Example.
 
   Note: All the fully_connected layers have been transformed to conv2d layers.
@@ -83,6 +84,11 @@ def vgg_a(inputs,
     spatial_squeeze: whether or not should squeeze the spatial dimensions of the
       outputs. Useful to remove unnecessary dimensions for classification.
     scope: Optional scope for the variables.
+    fc_conv_padding: the type of padding to use for the fully connected layer
+      that is implemented as a convolutional layer. Use 'SAME' padding if you
+      are applying the network in a fully convolutional manner and want to
+      get a prediction map downsampled by a factor of 32 as an output. Otherwise,
+      the output prediction map will be (input / 32) - 6 in case of 'VALID' padding.
 
   Returns:
     the last op containing the log predictions and end_points dict.
@@ -103,7 +109,7 @@ def vgg_a(inputs,
       net = slim.repeat(net, 2, slim.conv2d, 512, [3, 3], scope='conv5')
       net = slim.max_pool2d(net, [2, 2], scope='pool5')
       # Use conv2d instead of fully_connected layers.
-      net = slim.conv2d(net, 4096, [7, 7], padding='VALID', scope='fc6')
+      net = slim.conv2d(net, 4096, [7, 7], padding=fc_conv_padding, scope='fc6')
       net = slim.dropout(net, dropout_keep_prob, is_training=is_training,
                          scope='dropout6')
       net = slim.conv2d(net, 4096, [1, 1], scope='fc7')
@@ -127,7 +133,8 @@ def vgg_16(inputs,
            is_training=True,
            dropout_keep_prob=0.5,
            spatial_squeeze=True,
-           scope='vgg_16'):
+           scope='vgg_16',
+           fc_conv_padding='VALID'):
   """Oxford Net VGG 16-Layers version D Example.
 
   Note: All the fully_connected layers have been transformed to conv2d layers.
@@ -142,6 +149,11 @@ def vgg_16(inputs,
     spatial_squeeze: whether or not should squeeze the spatial dimensions of the
       outputs. Useful to remove unnecessary dimensions for classification.
     scope: Optional scope for the variables.
+    fc_conv_padding: the type of padding to use for the fully connected layer
+      that is implemented as a convolutional layer. Use 'SAME' padding if you
+      are applying the network in a fully convolutional manner and want to
+      get a prediction map downsampled by a factor of 32 as an output. Otherwise,
+      the output prediction map will be (input / 32) - 6 in case of 'VALID' padding.
 
   Returns:
     the last op containing the log predictions and end_points dict.
@@ -162,7 +174,7 @@ def vgg_16(inputs,
       net = slim.repeat(net, 3, slim.conv2d, 512, [3, 3], scope='conv5')
       net = slim.max_pool2d(net, [2, 2], scope='pool5')
       # Use conv2d instead of fully_connected layers.
-      net = slim.conv2d(net, 4096, [7, 7], padding='VALID', scope='fc6')
+      net = slim.conv2d(net, 4096, [7, 7], padding=fc_conv_padding, scope='fc6')
       net = slim.dropout(net, dropout_keep_prob, is_training=is_training,
                          scope='dropout6')
       net = slim.conv2d(net, 4096, [1, 1], scope='fc7')
@@ -186,7 +198,8 @@ def vgg_19(inputs,
            is_training=True,
            dropout_keep_prob=0.5,
            spatial_squeeze=True,
-           scope='vgg_19'):
+           scope='vgg_19',
+           fc_conv_padding='VALID'):
   """Oxford Net VGG 19-Layers version E Example.
 
   Note: All the fully_connected layers have been transformed to conv2d layers.
@@ -201,6 +214,11 @@ def vgg_19(inputs,
     spatial_squeeze: whether or not should squeeze the spatial dimensions of the
       outputs. Useful to remove unnecessary dimensions for classification.
     scope: Optional scope for the variables.
+    fc_conv_padding: the type of padding to use for the fully connected layer
+      that is implemented as a convolutional layer. Use 'SAME' padding if you
+      are applying the network in a fully convolutional manner and want to
+      get a prediction map downsampled by a factor of 32 as an output. Otherwise,
+      the output prediction map will be (input / 32) - 6 in case of 'VALID' padding.
 
   Returns:
     the last op containing the log predictions and end_points dict.
@@ -221,7 +239,7 @@ def vgg_19(inputs,
       net = slim.repeat(net, 4, slim.conv2d, 512, [3, 3], scope='conv5')
       net = slim.max_pool2d(net, [2, 2], scope='pool5')
       # Use conv2d instead of fully_connected layers.
-      net = slim.conv2d(net, 4096, [7, 7], padding='VALID', scope='fc6')
+      net = slim.conv2d(net, 4096, [7, 7], padding=fc_conv_padding, scope='fc6')
       net = slim.dropout(net, dropout_keep_prob, is_training=is_training,
                          scope='dropout6')
       net = slim.conv2d(net, 4096, [1, 1], scope='fc7')

slim/preprocessing/cifarnet_preprocessing.py

@@ -45,7 +45,7 @@ def preprocess_for_train(image,
   Returns:
     A preprocessed image.
   """
-  tf.image_summary('image', tf.expand_dims(image, 0))
+  tf.summary.image('image', tf.expand_dims(image, 0))
 
   # Transform the image to floats.
   image = tf.to_float(image)
@@ -58,7 +58,7 @@ def preprocess_for_train(image,
   # Randomly flip the image horizontally.
   distorted_image = tf.image.random_flip_left_right(distorted_image)
 
-  tf.image_summary('distorted_image', tf.expand_dims(distorted_image, 0))
+  tf.summary.image('distorted_image', tf.expand_dims(distorted_image, 0))
 
   # Because these operations are not commutative, consider randomizing
   # the order their operation.
@@ -67,7 +67,7 @@ def preprocess_for_train(image,
   distorted_image = tf.image.random_contrast(distorted_image,
                                              lower=0.2, upper=1.8)
   # Subtract off the mean and divide by the variance of the pixels.
-  return tf.image.per_image_whitening(distorted_image)
+  return tf.image.per_image_standardization(distorted_image)
 
 
 def preprocess_for_eval(image, output_height, output_width):
@@ -81,7 +81,7 @@ def preprocess_for_eval(image, output_height, output_width):
   Returns:
     A preprocessed image.
   """
-  tf.image_summary('image', tf.expand_dims(image, 0))
+  tf.summary.image('image', tf.expand_dims(image, 0))
   # Transform the image to floats.
   image = tf.to_float(image)
 
@@ -89,10 +89,10 @@ def preprocess_for_eval(image, output_height, output_width):
   resized_image = tf.image.resize_image_with_crop_or_pad(image,
                                                          output_width,
                                                          output_height)
-  tf.image_summary('resized_image', tf.expand_dims(resized_image, 0))
+  tf.summary.image('resized_image', tf.expand_dims(resized_image, 0))
 
   # Subtract off the mean and divide by the variance of the pixels.
-  return tf.image.per_image_whitening(resized_image)
+  return tf.image.per_image_standardization(resized_image)
 
 
 def preprocess_image(image, output_height, output_width, is_training=False):