def imagenet_input(is_training):
"""Data reader for imagenet.
Reads in imagenet data and performs pre-processing on the images.
Args:
is_training: bool specifying if train or validation dataset is needed.
Returns:
A batch of images and labels.
"""
if is_training:
dataset = dataset_factory.get_dataset('imagenet', 'train',
FLAGS.dataset_dir)
else:
dataset = dataset_factory.get_dataset('imagenet', 'validation',
FLAGS.dataset_dir)
provider = slim.dataset_data_provider.DatasetDataProvider(
dataset,
shuffle=is_training,
common_queue_capacity=2 * FLAGS.batch_size,
common_queue_min=FLAGS.batch_size)
[image, label] = provider.get(['image', 'label'])
image_preprocessing_fn = preprocessing_factory.get_preprocessing(
'mobilenet_v1', is_training=is_training)
image = image_preprocessing_fn(image, FLAGS.image_size, FLAGS.image_size)
images, labels = tf.train.batch([image, label],
batch_size=FLAGS.batch_size,
num_threads=4,
capacity=5 * FLAGS.batch_size)
labels = slim.one_hot_encoding(labels, FLAGS.num_classes)
return images, labels
def provide_data(batch_size,
dataset_dir,
dataset_name='cifar10',
split_name='train',
one_hot=True):
"""Provides batches of CIFAR data.
Args:
batch_size: The number of images in each batch.
dataset_dir: The directory where the CIFAR10 data can be found. If `None`,
use default.
dataset_name: Name of the dataset.
split_name: Should be either 'train' or 'test'.
one_hot: Output one hot vector instead of int32 label.
Returns:
images: A `Tensor` of size [batch_size, 32, 32, 3]. Output pixel values are
in [-1, 1].
labels: Either (1) one_hot_labels if `one_hot` is `True`
A `Tensor` of size [batch_size, num_classes], where each row has a
single element set to one and the rest set to zeros.
Or (2) labels if `one_hot` is `False`
A `Tensor` of size [batch_size], holding the labels as integers.
num_samples: The number of total samples in the dataset.
num_classes: The number of classes in the dataset.
Raises:
ValueError: if the split_name is not either 'train' or 'test'.
"""
dataset = datasets.get_dataset(dataset_name,
split_name,
dataset_dir=dataset_dir)
provider = slim.dataset_data_provider.DatasetDataProvider(
dataset,
common_queue_capacity=5 * batch_size,
common_queue_min=batch_size,
shuffle=(split_name == 'train'))
[image, label] = provider.get(['image', 'label'])
# Preprocess the images.
image = (tf.to_float(image) - 128.0) / 128.0
# Creates a QueueRunner for the pre-fetching operation.
images, labels = tf.train.batch([image, label],
batch_size=batch_size,
num_threads=32,
capacity=5 * batch_size)
labels = tf.reshape(labels, [-1])
if one_hot:
labels = tf.one_hot(labels, dataset.num_classes)
return images, labels, dataset.num_samples, dataset.num_classes
def provide_data(dataset_name='cifar10',
split_name='train',
dataset_dir='',
batch_size=32,
shuffle=True,
num_threads=1,
patch_height=32,
patch_width=32,
colors=3):
"""Provides a batch of image data from predefined dataset.
Args:
dataset_name: A string of dataset name. Defaults to 'cifar10'.
split_name: Either 'train' or 'validation'. Defaults to 'train'.
dataset_dir: The directory where the data can be found. If `None`, use
default.
batch_size: The number of images in each minibatch. Defaults to 32.
shuffle: Whether to shuffle the read images. Defaults to True.
num_threads: Number of prefetching threads. Defaults to 1.
patch_height: A Python integer. The read images height. Defaults to 32.
patch_width: A Python integer. The read images width. Defaults to 32.
colors: Number of channels. Defaults to 3.
Returns:
A float `Tensor`s with shape [batch_size, patch_height, patch_width, colors]
representing a batch of images.
"""
dataset = datasets.get_dataset(dataset_name,
split_name,
dataset_dir=dataset_dir)
provider = tf.contrib.slim.dataset_data_provider.DatasetDataProvider(
dataset,
num_readers=1,
common_queue_capacity=5 * batch_size,
common_queue_min=batch_size,
shuffle=shuffle)
return batch_images(image=normalize_image(provider.get(['image'])[0]),
patch_height=patch_height,
patch_width=patch_width,
colors=colors,
batch_size=batch_size,
shuffle=shuffle,
num_threads=num_threads)
def provide_data(split_name, batch_size, dataset_dir, num_readers=1,
num_threads=1):
"""Provides batches of MNIST digits.
Args:
split_name: Either 'train' or 'test'.
batch_size: The number of images in each batch.
dataset_dir: The directory where the MNIST data can be found.
num_readers: Number of dataset readers.
num_threads: Number of prefetching threads.
Returns:
images: A `Tensor` of size [batch_size, 28, 28, 1]
one_hot_labels: A `Tensor` of size [batch_size, mnist.NUM_CLASSES], where
each row has a single element set to one and the rest set to zeros.
num_samples: The number of total samples in the dataset.
Raises:
ValueError: If `split_name` is not either 'train' or 'test'.
"""
dataset = datasets.get_dataset('mnist', split_name, dataset_dir=dataset_dir)
provider = slim.dataset_data_provider.DatasetDataProvider(
dataset,
num_readers=num_readers,
common_queue_capacity=2 * batch_size,
common_queue_min=batch_size,
shuffle=(split_name == 'train'))
[image, label] = provider.get(['image', 'label'])
# Preprocess the images.
image = (tf.to_float(image) - 128.0) / 128.0
# Creates a QueueRunner for the pre-fetching operation.
images, labels = tf.train.batch(
[image, label],
batch_size=batch_size,
num_threads=num_threads,
capacity=5 * batch_size)
one_hot_labels = tf.one_hot(labels, dataset.num_classes)
return images, one_hot_labels, dataset.num_samples
def main(_):
if not FLAGS.output_file:
raise ValueError('You must supply the path to save to with --output_file')
tf.logging.set_verbosity(tf.logging.INFO)
with tf.Graph().as_default() as graph:
dataset = dataset_factory.get_dataset(FLAGS.dataset_name, 'train',
FLAGS.dataset_dir)
network_fn = nets_factory.get_network_fn(
FLAGS.model_name,
num_classes=(dataset.num_classes - FLAGS.labels_offset),
is_training=FLAGS.is_training)
image_size = FLAGS.image_size or network_fn.default_image_size
placeholder = tf.placeholder(name='input', dtype=tf.float32,
shape=[FLAGS.batch_size, image_size,
image_size, 3])
network_fn(placeholder)
if FLAGS.quantize:
tf.contrib.quantize.create_eval_graph()
graph_def = graph.as_graph_def()
with gfile.GFile(FLAGS.output_file, 'wb') as f:
f.write(graph_def.SerializeToString())
def main(_):
if not FLAGS.dataset_dir:
raise ValueError(
'You must supply the dataset directory with --dataset_dir')
tf.logging.set_verbosity(tf.logging.INFO)
with tf.Graph().as_default():
#######################
# Config model_deploy #
#######################
deploy_config = model_deploy.DeploymentConfig(
num_clones=FLAGS.num_clones,
clone_on_cpu=FLAGS.clone_on_cpu,
replica_id=FLAGS.task,
num_replicas=FLAGS.worker_replicas,
num_ps_tasks=FLAGS.num_ps_tasks)
# Create global_step
with tf.device(deploy_config.variables_device()):
global_step = slim.create_global_step()
######################
# Select the dataset #
######################
dataset = dataset_factory.get_dataset(FLAGS.dataset_name,
FLAGS.dataset_split_name,
FLAGS.dataset_dir)
######################
# Select the network #
######################
network_fn = nets_factory.get_network_fn(
FLAGS.model_name,
num_classes=(dataset.num_classes - FLAGS.labels_offset),
weight_decay=FLAGS.weight_decay,
is_training=True)
#####################################
# Select the preprocessing function #
#####################################
preprocessing_name = FLAGS.preprocessing_name or FLAGS.model_name
image_preprocessing_fn = preprocessing_factory.get_preprocessing(
preprocessing_name, is_training=True)
##############################################################
# Create a dataset provider that loads data from the dataset #
##############################################################
with tf.device(deploy_config.inputs_device()):
provider = slim.dataset_data_provider.DatasetDataProvider(
dataset,
num_readers=FLAGS.num_readers,
common_queue_capacity=20 * FLAGS.batch_size,
common_queue_min=10 * FLAGS.batch_size)
[image, label] = provider.get(['image', 'label'])
label -= FLAGS.labels_offset
train_image_size = FLAGS.train_image_size or network_fn.default_image_size
image = image_preprocessing_fn(image, train_image_size,
train_image_size)
images, labels = tf.train.batch(
[image, label],
batch_size=FLAGS.batch_size,
num_threads=FLAGS.num_preprocessing_threads,
capacity=5 * FLAGS.batch_size)
labels = slim.one_hot_encoding(
labels, dataset.num_classes - FLAGS.labels_offset)
batch_queue = slim.prefetch_queue.prefetch_queue(
[images, labels], capacity=2 * deploy_config.num_clones)
####################
# Define the model #
####################
def clone_fn(batch_queue):
"""Allows data parallelism by creating multiple clones of network_fn."""
imgs, ls = batch_queue.dequeue()
logits, end_points = network_fn(imgs)
#############################
# Specify the loss function #
#############################
if 'AuxLogits' in end_points:
slim.losses.softmax_cross_entropy(
end_points['AuxLogits'],
ls,
label_smoothing=FLAGS.label_smoothing,
weights=0.4,
scope='aux_loss')
slim.losses.softmax_cross_entropy(
logits, ls, label_smoothing=FLAGS.label_smoothing, weights=1.0)
return end_points
# Gather initial summaries.
summaries = set(tf.get_collection(tf.GraphKeys.SUMMARIES))
clones = model_deploy.create_clones(deploy_config, clone_fn,
[batch_queue])
first_clone_scope = deploy_config.clone_scope(0)
# Gather update_ops from the first clone. These contain, for example,
# the updates for the batch_norm variables created by network_fn.
update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS,
first_clone_scope)
# Add summaries for end_points.
end_points = clones[0].outputs
for end_point in end_points:
x = end_points[end_point]
summaries.add(tf.summary.histogram('activations/' + end_point, x))
summaries.add(
tf.summary.scalar('sparsity/' + end_point,
tf.nn.zero_fraction(x)))
# Add summaries for losses.
for loss in tf.get_collection(tf.GraphKeys.LOSSES, first_clone_scope):
summaries.add(tf.summary.scalar('losses/%s' % loss.op.name, loss))
# Add summaries for variables.
for variable in slim.get_model_variables():
summaries.add(tf.summary.histogram(variable.op.name, variable))
#################################
# Configure the moving averages #
#################################
if FLAGS.moving_average_decay:
moving_average_variables = slim.get_model_variables()
variable_averages = tf.train.ExponentialMovingAverage(
FLAGS.moving_average_decay, global_step)
else:
moving_average_variables, variable_averages = None, None
if FLAGS.quantize_delay >= 0:
tf.contrib.quantize.create_training_graph(
quant_delay=FLAGS.quantize_delay)
#########################################
# Configure the optimization procedure. #
#########################################
with tf.device(deploy_config.optimizer_device()):
learning_rate = _configure_learning_rate(dataset.num_samples,
global_step)
optimizer = _configure_optimizer(learning_rate)
summaries.add(tf.summary.scalar('learning_rate', learning_rate))
if FLAGS.sync_replicas:
# If sync_replicas is enabled, the averaging will be done in the chief
# queue runner.
optimizer = tf.train.SyncReplicasOptimizer(
opt=optimizer,
replicas_to_aggregate=FLAGS.replicas_to_aggregate,
total_num_replicas=FLAGS.worker_replicas,
variable_averages=variable_averages,
variables_to_average=moving_average_variables)
elif FLAGS.moving_average_decay:
# Update ops executed locally by trainer.
update_ops.append(
variable_averages.apply(moving_average_variables))
# Variables to train.
variables_to_train = _get_variables_to_train()
# and returns a train_tensor and summary_op
total_loss, clones_gradients = model_deploy.optimize_clones(
clones, optimizer, var_list=variables_to_train)
# Add total_loss to summary.
summaries.add(tf.summary.scalar('total_loss', total_loss))
# Create gradient updates.
grad_updates = optimizer.apply_gradients(clones_gradients,
global_step=global_step)
update_ops.append(grad_updates)
update_op = tf.group(*update_ops)
with tf.control_dependencies([update_op]):
train_tensor = tf.identity(total_loss, name='train_op')
# Add the summaries from the first clone. These contain the summaries
# created by model_fn and either optimize_clones() or _gather_clone_loss().
summaries |= set(
tf.get_collection(tf.GraphKeys.SUMMARIES, first_clone_scope))
# Merge all summaries together.
summary_op = tf.summary.merge(list(summaries), name='summary_op')
###########################
# Kicks off the training. #
###########################
slim.learning.train(
train_tensor,
logdir=FLAGS.train_dir,
master=FLAGS.master,
is_chief=(FLAGS.task == 0),
init_fn=_get_init_fn(),
summary_op=summary_op,
number_of_steps=FLAGS.max_number_of_steps,
log_every_n_steps=FLAGS.log_every_n_steps,
save_summaries_secs=FLAGS.save_summaries_secs,
save_interval_secs=FLAGS.save_interval_secs,
sync_optimizer=optimizer if FLAGS.sync_replicas else None)
def main(_):
if not FLAGS.dataset_dir:
raise ValueError(
'You must supply the dataset directory with --dataset_dir')
tf.logging.set_verbosity(tf.logging.INFO)
with tf.Graph().as_default():
tf_global_step = slim.get_or_create_global_step()
######################
# Select the dataset #
######################
dataset = dataset_factory.get_dataset(FLAGS.dataset_name,
FLAGS.dataset_split_name,
FLAGS.dataset_dir)
####################
# Select the model #
####################
network_fn = nets_factory.get_network_fn(
FLAGS.model_name,
num_classes=(dataset.num_classes - FLAGS.labels_offset),
is_training=False)
##############################################################
# Create a dataset provider that loads data from the dataset #
##############################################################
provider = slim.dataset_data_provider.DatasetDataProvider(
dataset,
shuffle=False,
common_queue_capacity=2 * FLAGS.batch_size,
common_queue_min=FLAGS.batch_size)
[image, label] = provider.get(['image', 'label'])
label -= FLAGS.labels_offset
#####################################
# Select the preprocessing function #
#####################################
preprocessing_name = FLAGS.preprocessing_name or FLAGS.model_name
image_preprocessing_fn = preprocessing_factory.get_preprocessing(
preprocessing_name, is_training=False)
eval_image_size = FLAGS.eval_image_size or network_fn.default_image_size
image = image_preprocessing_fn(image, eval_image_size, eval_image_size)
images, labels = tf.train.batch(
[image, label],
batch_size=FLAGS.batch_size,
num_threads=FLAGS.num_preprocessing_threads,
capacity=5 * FLAGS.batch_size)
####################
# Define the model #
####################
logits, _ = network_fn(images)
if FLAGS.quantize:
tf.contrib.quantize.create_eval_graph()
if FLAGS.moving_average_decay:
variable_averages = tf.train.ExponentialMovingAverage(
FLAGS.moving_average_decay, tf_global_step)
variables_to_restore = variable_averages.variables_to_restore(
slim.get_model_variables())
variables_to_restore[tf_global_step.op.name] = tf_global_step
else:
variables_to_restore = slim.get_variables_to_restore()
predictions = tf.argmax(logits, 1)
labels = tf.squeeze(labels)
# 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(logits, labels, 5),
})
# Print the summaries to screen.
for name, value in names_to_values.items():
summary_name = 'eval/%s' % name
op = tf.summary.scalar(summary_name, value, collections=[])
op = tf.Print(op, [value], summary_name)
tf.add_to_collection(tf.GraphKeys.SUMMARIES, op)
# TODO(sguada) use num_epochs=1
if FLAGS.max_num_batches:
num_batches = FLAGS.max_num_batches
else:
# This ensures that we make a single pass over all of the data.
num_batches = math.ceil(dataset.num_samples /
float(FLAGS.batch_size))
if tf.gfile.IsDirectory(FLAGS.checkpoint_path):
checkpoint_path = tf.train.latest_checkpoint(FLAGS.checkpoint_path)
else:
checkpoint_path = FLAGS.checkpoint_path
tf.logging.info('Evaluating %s' % checkpoint_path)
slim.evaluation.evaluate_once(
master=FLAGS.master,
checkpoint_path=checkpoint_path,
logdir=FLAGS.eval_dir,
num_evals=num_batches,
eval_op=list(names_to_updates.values()),
variables_to_restore=variables_to_restore)