def testBuildAndCheckAllEndPointsUptoPreAuxLogitsWithOutputStrideEight(
self):
batch_size = 5
height, width = 299, 299
inputs = tf.random_uniform((batch_size, height, width, 3))
_, end_points = inception_resnet_v2_base(inputs,
final_endpoint='PreAuxLogits',
output_stride=8)
endpoints_shapes = {
'Conv2d_1a_3x3': [5, 149, 149, 32],
'Conv2d_2a_3x3': [5, 147, 147, 32],
'Conv2d_2b_3x3': [5, 147, 147, 64],
'MaxPool_3a_3x3': [5, 73, 73, 64],
'Conv2d_3b_1x1': [5, 73, 73, 80],
'Conv2d_4a_3x3': [5, 71, 71, 192],
'MaxPool_5a_3x3': [5, 35, 35, 192],
'Mixed_5b': [5, 35, 35, 320],
'Mixed_6a': [5, 33, 33, 1088],
'PreAuxLogits': [5, 33, 33, 1088]
}
self.assertItemsEqual(endpoints_shapes.keys(), end_points.keys())
for endpoint_name in endpoints_shapes:
expected_shape = endpoints_shapes[endpoint_name]
self.assertTrue(endpoint_name in end_points)
self.assertListEqual(
end_points[endpoint_name].get_shape().as_list(),
expected_shape)
def testBuildAndCheckAllEndPointsUptoPreAuxLogitsWithAlignedFeatureMaps(
self):
batch_size = 5
height, width = 299, 299
inputs = tf.random_uniform((batch_size, height, width, 3))
_, end_points = inception_resnet_v2_base(inputs,
final_endpoint='PreAuxLogits',
align_feature_maps=True)
endpoints_shapes = {
'Conv2d_1a_3x3': [5, 150, 150, 32],
'Conv2d_2a_3x3': [5, 150, 150, 32],
'Conv2d_2b_3x3': [5, 150, 150, 64],
'MaxPool_3a_3x3': [5, 75, 75, 64],
'Conv2d_3b_1x1': [5, 75, 75, 80],
'Conv2d_4a_3x3': [5, 75, 75, 192],
'MaxPool_5a_3x3': [5, 38, 38, 192],
'Mixed_5b': [5, 38, 38, 320],
'Mixed_6a': [5, 19, 19, 1088],
'PreAuxLogits': [5, 19, 19, 1088]
}
self.assertItemsEqual(endpoints_shapes.keys(), end_points.keys())
for endpoint_name in endpoints_shapes:
expected_shape = endpoints_shapes[endpoint_name]
self.assertTrue(endpoint_name in end_points)
self.assertListEqual(
end_points[endpoint_name].get_shape().as_list(),
expected_shape)
def conv_tower_fn(self, images, is_training=True, reuse=None):
"""Computes convolutional features using the InceptionV3 model.
Args:
images: A tensor of shape [batch_size, height, width, channels].
is_training: whether is training or not.
reuse: whether or not the network and its variables should be reused. To
be able to reuse 'scope' must be given.
Returns:
A tensor of shape [batch_size, OH, OW, N], where OWxOH is resolution of
output feature map and N is number of output features (depends on the
network architecture).
"""
mparams = self._mparams['conv_tower_fn']
logging.debug('Using final_endpoint=%s', mparams.final_endpoint)
with tf.variable_scope('conv_tower_fn/INCE'):
if reuse:
tf.get_variable_scope().reuse_variables()
# with slim.arg_scope(inception.inception_v3_arg_scope()):
with slim.arg_scope(inception_resnet_v2.inception_resnet_v2_arg_scope()):
with slim.arg_scope([slim.batch_norm, slim.dropout],
is_training=is_training):
net, _ = inception_resnet_v2.inception_resnet_v2_base(
images, final_endpoint=mparams.final_endpoint)
# net, _ = inception.inception_v3_base(
# images, final_endpoint=mparams.final_endpoint)
return net
def testBuildBaseNetwork(self):
batch_size = 5
height, width = 299, 299
inputs = tf.random_uniform((batch_size, height, width, 3))
net, end_points = inception_resnet_v2_base(inputs)
self.assertTrue(
net.op.name.startswith('InceptionResnetV2/Conv2d_7b_1x1'))
self.assertListEqual(net.get_shape().as_list(),
[batch_size, 8, 8, 1536])
expected_endpoints = [
'Conv2d_1a_3x3', 'Conv2d_2a_3x3', 'Conv2d_2b_3x3',
'MaxPool_3a_3x3', 'Conv2d_3b_1x1', 'Conv2d_4a_3x3',
'MaxPool_5a_3x3', 'Mixed_5b', 'Mixed_6a', 'PreAuxLogits',
'Mixed_7a', 'Conv2d_7b_1x1'
]
self.assertItemsEqual(end_points.keys(), expected_endpoints)
def testBuildOnlyUptoFinalEndpoint(self):
batch_size = 5
height, width = 299, 299
endpoints = [
'Conv2d_1a_3x3', 'Conv2d_2a_3x3', 'Conv2d_2b_3x3',
'MaxPool_3a_3x3', 'Conv2d_3b_1x1', 'Conv2d_4a_3x3',
'MaxPool_5a_3x3', 'Mixed_5b', 'Mixed_6a', 'PreAuxLogits',
'Mixed_7a', 'Conv2d_7b_1x1'
]
for index, endpoint in enumerate(endpoints):
with tf.Graph().as_default():
inputs = tf.random_uniform((batch_size, height, width, 3))
out_tensor, end_points = inception_resnet_v2_base(
inputs, final_endpoint=endpoint)
if endpoint != 'PreAuxLogits':
self.assertTrue(
out_tensor.op.name.startswith('InceptionResnetV2/' +
endpoint))
self.assertItemsEqual(endpoints[:index + 1], end_points.keys())
scaledInputBatchImages = tf.subtract(scaledInputBatchImages, 0.5)
scaledInputBatchImages = tf.multiply(scaledInputBatchImages, 2.0)
# Create model
if options.modelName == "NASNet":
arg_scope = nasnet.nasnet_large_arg_scope()
with slim.arg_scope(arg_scope):
logits, endPoints = nasnet.build_nasnet_large(scaledInputBatchImages, is_training=False, num_classes=options.numClasses)
elif options.modelName == "IncResV2":
arg_scope = inception_resnet_v2.inception_resnet_v2_arg_scope()
with slim.arg_scope(arg_scope):
# logits, endPoints = inception_resnet_v2.inception_resnet_v2(scaledInputBatchImages, is_training=False)
with tf.variable_scope('InceptionResnetV2', 'InceptionResnetV2', [scaledInputBatchImages], reuse=None) as scope:
with slim.arg_scope([slim.batch_norm, slim.dropout], is_training=False):
net, endPoints = inception_resnet_v2.inception_resnet_v2_base(scaledInputBatchImages, scope=scope, activation_fn=tf.nn.relu)
variablesToRestore = slim.get_variables_to_restore(include=["InceptionResnetV2"])
else:
print ("Error: Model not found!")
exit (-1)
# TODO: Attach the decoder to the encoder
print (endPoints.keys())
if options.useSkipConnections:
print ("Adding skip connections from the encoder to the decoder!")
predictedLogits = attachDecoder(net, endPoints, tf.shape(scaledInputBatchImages))
predictedMask = tf.expand_dims(tf.argmax(predictedLogits, axis=-1), -1, name="predictedMasks")
if options.tensorboardVisualization:
def inception_resnet_v2(images,
trainable=True,
is_training=True,
weight_decay=0.00004,
stddev=0.1,
dropout_keep_prob=0.8,
use_batch_norm=True,
batch_norm_params=None,
add_summaries=True,
scope="InceptionResnetV2"):
"""Builds an Inception_resnet_V2 subgraph for image embeddings.
Args:
images: A float32 Tensor of shape [batch, height, width, channels].
trainable: Whether the inception submodel should be trainable or not.
is_training: Boolean indicating training mode or not.
weight_decay: Coefficient for weight regularization.
stddev: The standard deviation of the trunctated normal weight initializer.
dropout_keep_prob: Dropout keep probability.
use_batch_norm: Whether to use batch normalization.
batch_norm_params: Parameters for batch normalization. See
tf.contrib.layers.batch_norm for details.
add_summaries: Whether to add activation summaries.
scope: Optional Variable scope.
Returns:
end_points: A dictionary of activations from inception_Resnet__v2 layers.
"""
# Only consider the inception model to be in training mode if it's trainable.
is_inception_model_training = trainable and is_training
if use_batch_norm:
# Default parameters for batch normalization.
if not batch_norm_params:
batch_norm_params = {
"is_training": is_inception_model_training,
"trainable": trainable,
# Decay for the moving averages.
"decay": 0.9997,
# Epsilon to prevent 0s in variance.
"epsilon": 0.001,
# Collection containing the moving mean and moving variance.
"variables_collections": {
"beta": None,
"gamma": None,
"moving_mean": ["moving_vars"],
"moving_variance": ["moving_vars"],
}
}
else:
batch_norm_params = None
if trainable:
weights_regularizer = tf.contrib.layers.l2_regularizer(weight_decay)
else:
weights_regularizer = None
with tf.variable_scope(scope, "InceptionResnetV2", [images]) as scope:
with slim.arg_scope(
[slim.conv2d, slim.fully_connected],
weights_regularizer=weights_regularizer,
trainable=trainable):
with slim.arg_scope(
[slim.conv2d],
weights_initializer=tf.truncated_normal_initializer(stddev=stddev),
activation_fn=tf.nn.relu,
normalizer_fn=slim.batch_norm,
normalizer_params=batch_norm_params):
net, end_points = inception_resnet_v2_base(images, scope=scope)
# Jay-modification: # Keep the spatial information of 2D images for attention (batch_size, 8, 8, 1536)
'''with tf.variable_scope("logits"):
shape = net.get_shape()
net = slim.avg_pool2d(net, shape[1:3], padding="VALID", scope="pool")
net = slim.dropout(
net,
keep_prob=dropout_keep_prob,
is_training=is_inception_model_training,
scope="dropout")
net = slim.flatten(net, scope="flatten")'''
# Add summaries.
if add_summaries:
for v in end_points.values():
tf.contrib.layers.summaries.summarize_activation(v)
return net