def test_variable_scope(self):
with self.test_session():
inputs = tf.constant(1)
with alexnet.variable_scope(inputs) as variable_scope:
self.assertEqual(variable_scope.name, 'alexnet_v2')
v = tf.get_variable('test', [1])
self.assertIs(v.graph, inputs.graph)
def test_fc_layers_end_points(self):
layers_name_prefix = "length"
batch_size, height, width, channels = 5, 224, 224, 3
with self.test_session():
inputs = tf.random_uniform((batch_size, height, width, channels))
with alexnet.variable_scope([inputs]) as variable_scope:
end_points_collection = alexnet.end_points_collection_name(
variable_scope)
net, _ = alexnet.cnn_layers(inputs, variable_scope,
end_points_collection)
_, end_points_collection = alexnet.fc_layers(
net,
variable_scope,
end_points_collection,
num_classes=10,
name_prefix=layers_name_prefix)
end_points = slim.utils.convert_collection_to_dict(
end_points_collection)
expected_names = [
'alexnet_v2/%s_fc8' % layers_name_prefix,
'alexnet_v2/%s_fc8/squeezed' % layers_name_prefix
]
self.assertSetEqual(
set(filter(lambda x: x.find('_fc') != -1, end_points.keys())),
set(expected_names))
def test_cnn_layers_end_points(self):
batch_size, height, width, channels = 5, 224, 224, 3
with self.test_session():
inputs = tf.random_uniform((batch_size, height, width, channels))
with alexnet.variable_scope([inputs]) as variable_scope:
end_points_collection = alexnet.end_points_collection_name(
variable_scope)
_, end_points_collection = alexnet.cnn_layers(
inputs, variable_scope, end_points_collection)
end_points = slim.utils.convert_collection_to_dict(
end_points_collection)
expected_names = [
'alexnet_v2/conv1',
'alexnet_v2/pool1',
'alexnet_v2/conv2',
'alexnet_v2/pool2',
'alexnet_v2/conv3',
'alexnet_v2/conv4',
'alexnet_v2/conv5',
'alexnet_v2/pool5',
'alexnet_v2/fc6',
'alexnet_v2/fc7',
]
self.assertSetEqual(set(end_points.keys()), set(expected_names))
def test_network_output(self):
batch_size, height, width, channels = 5, 224, 224, 3
with self.test_session():
inputs = tf.random_uniform((batch_size, height, width, channels))
with alexnet.variable_scope([inputs]) as variable_scope:
end_points_collection = alexnet.end_points_collection_name(
variable_scope)
net, _ = alexnet.cnn_layers(inputs, variable_scope,
end_points_collection)
output, _ = alexnet.fc_layers(net,
variable_scope,
end_points_collection,
num_classes=10)
self.checkOutputs({
output: (5, 10),
})