def network_inception_v1():
input_shape = [1, 224, 224, 3]
input_ = tf.placeholder(dtype=tf.float32, name='input', shape=input_shape)
net, _end_points = inception_v1(input_,
num_classes=1000,
is_training=False)
return net
def test_network(img_path, label_path):
x = tf.placeholder("float", shape=[None, 224, 224, 3], name='input')
with slim.arg_scope(inception_v1.inception_v1_arg_scope()):
logits, end_points = inception_v1.inception_v1(x,
num_classes=1001,
dropout_keep_prob=1.0,
is_training=False)
predictions = end_points["Predictions"]
saver = tf.train.Saver()
init = tf.global_variables_initializer()
with tf.Session() as sess:
sess.run(init)
saver.restore(sess, "inception_v1.ckpt")
imgfloat = tf.cast(tf.image.decode_jpeg(tf.read_file(img_path),
channels=3),
dtype=tf.float32)
img = tf.subtract(
tf.multiply(
tf.div(
tf.image.resize_images(tf.expand_dims(imgfloat, 0),
(224, 224),
method=0), 255.0), 2), 1.0)
predictions_val = predictions.eval(feed_dict={x: img.eval()})
predicted_classes = np.argmax(predictions_val, axis=1)
file = open(label_path)
labels = file.readlines()
print(predicted_classes, labels[predicted_classes[0]])
def test_network(img_path, label_path):
x = tf.placeholder("float", shape=[None, 224, 224, 3], name='input')
xscale = tf.subtract(tf.multiply(tf.div(x, 255.0), 2), 1.0)
with slim.arg_scope(inception_v1.inception_v1_arg_scope()):
logits, end_points = inception_v1.inception_v1(xscale,
num_classes=1001,
dropout_keep_prob=1.0,
is_training=False)
predictions = tf.nn.softmax(logits, name="output")
saver = tf.train.Saver()
with tf.Session() as sess:
saver.restore(sess, "inception_v1.ckpt")
#ckpt_info("inception_v1.ckpt")
#var_list = tf.global_variables()
#print(var_list)
constant_graph = tf.get_default_graph().as_graph_def()
output_graph_def = graph_util.convert_variables_to_constants(
sess, constant_graph, ['output'])
with tf.gfile.GFile("inception_v1.pb", "wb") as f:
f.write(output_graph_def.SerializeToString())
imgfloat = tf.cast(tf.image.decode_jpeg(tf.read_file(img_path),
channels=3),
dtype=tf.float32)
img = tf.image.resize_images(tf.expand_dims(imgfloat, 0), (224, 224),
method=0)
predictions_val = predictions.eval(feed_dict={x: img.eval()})
predicted_classes = np.argmax(predictions_val, axis=1)
file = open(label_path)
labels = file.readlines()
print(predicted_classes, labels[predicted_classes[0]])
def testTrainEvalWithReuse(self):
train_batch_size = 5
eval_batch_size = 2
height, width = 224, 224
num_classes = 1000
train_inputs = random_ops.random_uniform(
(train_batch_size, height, width, 3))
inception_v1.inception_v1(train_inputs, num_classes)
eval_inputs = random_ops.random_uniform((eval_batch_size, height, width, 3))
logits, _ = inception_v1.inception_v1(eval_inputs, num_classes, reuse=True)
predictions = math_ops.argmax(logits, 1)
with self.cached_session() as sess:
sess.run(variables.global_variables_initializer())
output = sess.run(predictions)
self.assertEquals(output.shape, (eval_batch_size,))
def create_inception(image_input,
is_training,
scope="",
inception_out="Mixed_5c",
resnet_version=50,
cbn=None):
"""
Create a resnet by overidding the classic batchnorm with conditional batchnorm
:param image_input: placeholder with image
:param is_training: are you using the resnet at training_time or test_time
:param scope: tensorflow scope
:param resnet_version: 50/101/152
:param cbn: the cbn factory
:return: the resnet output
"""
# assert False, "\n" \
# "There is a bug with classic batchnorm with slim networks (https://github.com/tensorflow/tensorflow/issues/4887). \n" \
# "Please use the following config -> 'cbn': {'use_cbn':true, 'excluded_scope_names': ['*']}"
# arg_sc = slim_utils.resnet_arg_scope(is_training=is_training)
# print("--- 1")
arg_sc = inception_v1.inception_v1_arg_scope()
# Pick the correct version of the resnet
# if resnet_version == 50:
# current_resnet = resnet_v1.resnet_v1_50
# elif resnet_version == 101:
# current_resnet = resnet_v1.resnet_v1_101
# elif resnet_version == 152:
# current_resnet = resnet_v1.resnet_v1_152
# else:
# raise ValueError("Unsupported resnet version")
# inception_scope = os.path.join('InceptionV1/InceptionV1', inception_out)
# print("--- 2")
inception_scope = inception_out
# print(" resnet_out = {} , resnet_scope = {}".format(resnet_out,resnet_scope))
# print("--- 3")
with slim.arg_scope(arg_sc):
net, end_points = inception_v1.inception_v1(
image_input, 1001) # 1000 is the number of softmax class
print("Net = ", net)
# print("--- 4")
if len(scope) > 0 and not scope.endswith("/"):
scope += "/"
# print("--- 5")
# print(end_points)
print(" Batch ", inception_scope)
out = end_points[scope + inception_scope]
print("-- out Use: {},output = {}".format(inception_scope, out))
return out, end_points
def testLogitsNotSqueezed(self):
num_classes = 25
images = random_ops.random_uniform([1, 224, 224, 3])
logits, _ = inception_v1.inception_v1(
images, num_classes=num_classes, spatial_squeeze=False)
with self.cached_session() as sess:
variables.global_variables_initializer().run()
logits_out = sess.run(logits)
self.assertListEqual(list(logits_out.shape), [1, 1, 1, num_classes])
def testTrainEvalWithReuse(self):
train_batch_size = 5
eval_batch_size = 2
height, width = 224, 224
num_classes = 1000
train_inputs = random_ops.random_uniform(
(train_batch_size, height, width, 3))
inception_v1.inception_v1(train_inputs, num_classes)
eval_inputs = random_ops.random_uniform(
(eval_batch_size, height, width, 3))
logits, _ = inception_v1.inception_v1(eval_inputs,
num_classes,
reuse=True)
predictions = math_ops.argmax(logits, 1)
with self.test_session() as sess:
sess.run(variables.global_variables_initializer())
output = sess.run(predictions)
self.assertEquals(output.shape, (eval_batch_size, ))
def testLogitsNotSqueezed(self):
num_classes = 25
images = random_ops.random_uniform([1, 224, 224, 3])
logits, _ = inception_v1.inception_v1(images,
num_classes=num_classes,
spatial_squeeze=False)
with self.test_session() as sess:
variables.global_variables_initializer().run()
logits_out = sess.run(logits)
self.assertListEqual(list(logits_out.shape),
[1, 1, 1, num_classes])
def testBuildClassificationNetwork(self):
batch_size = 5
height, width = 224, 224
num_classes = 1000
inputs = random_ops.random_uniform((batch_size, height, width, 3))
logits, end_points = inception_v1.inception_v1(inputs, num_classes)
self.assertTrue(logits.op.name.startswith('InceptionV1/Logits'))
self.assertListEqual(logits.get_shape().as_list(),
[batch_size, num_classes])
self.assertTrue('Predictions' in end_points)
self.assertListEqual(end_points['Predictions'].get_shape().as_list(),
[batch_size, num_classes])
def testBuildClassificationNetwork(self):
batch_size = 5
height, width = 224, 224
num_classes = 1000
inputs = random_ops.random_uniform((batch_size, height, width, 3))
logits, end_points = inception_v1.inception_v1(inputs, num_classes)
self.assertTrue(logits.op.name.startswith('InceptionV1/Logits'))
self.assertListEqual(logits.get_shape().as_list(),
[batch_size, num_classes])
self.assertTrue('Predictions' in end_points)
self.assertListEqual(end_points['Predictions'].get_shape().as_list(),
[batch_size, num_classes])
def testEvaluation(self):
batch_size = 2
height, width = 224, 224
num_classes = 1000
eval_inputs = random_ops.random_uniform((batch_size, height, width, 3))
logits, _ = inception_v1.inception_v1(
eval_inputs, num_classes, is_training=False)
predictions = math_ops.argmax(logits, 1)
with self.cached_session() as sess:
sess.run(variables.global_variables_initializer())
output = sess.run(predictions)
self.assertEquals(output.shape, (batch_size,))
def trainmodel(train_batch, train_label_batch, train_label, num_epochs):
with slim.arg_scope(inception_v1_arg_scope()):
train_logits, end_points = inception_v1.inception_v1(train_batch,
num_classes=2,
is_training=True)
tf.losses.sparse_softmax_cross_entropy(labels=train_label,
logits=train_logits)
total_loss = tf.losses.get_total_loss()
global_step = tf.Variable(0, name='global_step', trainable=False)
update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
with tf.control_dependencies(update_ops):
optimizer = tf.train.AdamOptimizer(learning_rate=0.001)
train_op = optimizer.minimize(total_loss, global_step=global_step)
prediction_labels = tf.argmax(end_points['Predictions'], 1)
read_labels = tf.argmax(train_label_batch, 1)
correct_prediction = tf.equal(prediction_labels, read_labels)
train_accuracy_batch = tf.reduce_mean(tf.cast(correct_prediction, "float"))
saver = tf.train.Saver(tf.trainable_variables() +
tf.get_collection_ref("moving_vars"))
with tf.Session() as sess:
sess.run(
tf.group(tf.global_variables_initializer(),
tf.local_variables_initializer()))
print("Initialized")
step = 0
start_time = time.time()
for epoch_index in range(num_epochs):
_, l, end_points2, logits2, train_acc2_batch = sess.run([
train_op, total_loss, end_points, train_logits,
train_accuracy_batch
])
duration = time.time() - start_time
print("Minibatch loss at step %d: %.6f (%.3f sec)" %
(step, l, duration))
print(end_points2['Predictions'])
print("Minibatch accuracy: %.6f" % train_acc2_batch)
#print("lr: %.6f" % optimizer._lr)
step += 1
saver.save(sess, './train.ckpt')
def testUnknownBatchSize(self):
batch_size = 1
height, width = 224, 224
num_classes = 1000
inputs = array_ops.placeholder(dtypes.float32, (None, height, width, 3))
logits, _ = inception_v1.inception_v1(inputs, num_classes)
self.assertTrue(logits.op.name.startswith('InceptionV1/Logits'))
self.assertListEqual(logits.get_shape().as_list(), [None, num_classes])
images = random_ops.random_uniform((batch_size, height, width, 3))
with self.cached_session() as sess:
sess.run(variables.global_variables_initializer())
output = sess.run(logits, {inputs: images.eval()})
self.assertEquals(output.shape, (batch_size, num_classes))
def testEvaluation(self):
batch_size = 2
height, width = 224, 224
num_classes = 1000
eval_inputs = random_ops.random_uniform((batch_size, height, width, 3))
logits, _ = inception_v1.inception_v1(eval_inputs,
num_classes,
is_training=False)
predictions = math_ops.argmax(logits, 1)
with self.test_session() as sess:
sess.run(variables.global_variables_initializer())
output = sess.run(predictions)
self.assertEquals(output.shape, (batch_size, ))
def testUnknownBatchSize(self):
batch_size = 1
height, width = 224, 224
num_classes = 1000
inputs = array_ops.placeholder(dtypes.float32,
(None, height, width, 3))
logits, _ = inception_v1.inception_v1(inputs, num_classes)
self.assertTrue(logits.op.name.startswith('InceptionV1/Logits'))
self.assertListEqual(logits.get_shape().as_list(), [None, num_classes])
images = random_ops.random_uniform((batch_size, height, width, 3))
with self.test_session() as sess:
sess.run(variables.global_variables_initializer())
output = sess.run(logits, {inputs: images.eval()})
self.assertEquals(output.shape, (batch_size, num_classes))
def testUnknownImageShape(self):
ops.reset_default_graph()
batch_size = 2
height, width = 224, 224
num_classes = 1000
input_np = np.random.uniform(0, 1, (batch_size, height, width, 3))
with self.cached_session() as sess:
inputs = array_ops.placeholder(
dtypes.float32, shape=(batch_size, None, None, 3))
logits, end_points = inception_v1.inception_v1(inputs, num_classes)
self.assertTrue(logits.op.name.startswith('InceptionV1/Logits'))
self.assertListEqual(logits.get_shape().as_list(),
[batch_size, num_classes])
pre_pool = end_points['Mixed_5c']
feed_dict = {inputs: input_np}
variables.global_variables_initializer().run()
pre_pool_out = sess.run(pre_pool, feed_dict=feed_dict)
self.assertListEqual(list(pre_pool_out.shape), [batch_size, 7, 7, 1024])
def testUnknownImageShape(self):
ops.reset_default_graph()
batch_size = 2
height, width = 224, 224
num_classes = 1000
input_np = np.random.uniform(0, 1, (batch_size, height, width, 3))
with self.test_session() as sess:
inputs = array_ops.placeholder(dtypes.float32,
shape=(batch_size, None, None, 3))
logits, end_points = inception_v1.inception_v1(inputs, num_classes)
self.assertTrue(logits.op.name.startswith('InceptionV1/Logits'))
self.assertListEqual(logits.get_shape().as_list(),
[batch_size, num_classes])
pre_pool = end_points['Mixed_5c']
feed_dict = {inputs: input_np}
variables.global_variables_initializer().run()
pre_pool_out = sess.run(pre_pool, feed_dict=feed_dict)
self.assertListEqual(list(pre_pool_out.shape),
[batch_size, 7, 7, 1024])
def test_network():
x = tf.placeholder("float", shape=[None, 224, 224, 3], name='input')
with slim.arg_scope(inception_v1.inception_v1_arg_scope()):
logits, end_points = inception_v1.inception_v1(x,
num_classes=2,
dropout_keep_prob=1.0,
is_training=False)
predictions = end_points["Predictions"]
saver = tf.train.Saver()
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
saver.restore(sess, "train.ckpt")
path = './picture/'
w = 224
h = 224
c = 3
cate = [path + x for x in os.listdir(path) if os.path.isdir(path + x)]
imgs = []
labels = []
for idx, folder in enumerate(cate):
for im in glob.glob(folder + '/*.jpg'):
print('reading the image: %s' % (im))
img = io.imread(im)
img = transform.resize(img, (w, h, c))
imgs.append(img)
labels.append([1, 0] if idx == 0 else [0, 1])
break
break
data = np.asarray(imgs, np.float32)
label = np.asarray(labels, np.int32)
predictions_val = predictions.eval(feed_dict={x: data})
print(predictions_val)
#sys.path
#slim_path = "hdfs:///slim/models/research/slim"
# Please set this to the directory where you clone the tensorflow models repository
#sys.path.append(slim_path)
#from tensorflow.contrib.slim.python.slim.nets import inception
from tensorflow.contrib.slim.python.slim.nets.inception_v1 import inception_v1
from tensorflow.contrib.slim.python.slim.nets.inception_v1 import inception_v1_arg_scope
from tensorflow.contrib.slim.python.slim.nets.inception_v1 import inception_v1_base
slim = tf.contrib.slim
tf.reset_default_graph()
images = tf.placeholder(dtype=tf.float32, shape=(None, 224, 224, 3))
with slim.arg_scope(inception_v1_arg_scope()):
logits, end_points = inception_v1(images, num_classes=1001)
sess = tf.Session()
saver = tf.train.Saver()
saver.restore(sess, "file:///home/hduser/slim/checkpoint/inception_v1.ckpt")
#saver.restore(sess, "hdfs:///slim/checkpoint/inception_v1.ckpt")
# You need to edit this path to the checkpoint you downloaded
from zoo.util.tf import export_tf
avg_pool = end_points['Mixed_3c']
export_tf(sess,
"file:///home/hduser/slim/tfnet/",
inputs=[images],
outputs=[avg_pool])
from zoo.pipeline.api.net import TFNet
amodel = TFNet.from_export_folder("file:///home/hduser/slim/tfnet/")
def trainmodel(train_batch, train_label_batch, train_label, num_epochs):
ps_hosts = FLAGS.ps_hosts.split(",")
worker_hosts = FLAGS.worker_hosts.split(",")
# Create a cluster from the parameter server and worker hosts.
cluster = tf.train.ClusterSpec({"ps": ps_hosts, "worker": worker_hosts})
# Create and start a server for the local task.
config = tf.ConfigProto()
config.gpu_options.per_process_gpu_memory_fraction = 0.4
server = tf.train.Server(cluster,
job_name=FLAGS.job_name,
task_index=FLAGS.task_index,
config=config)
print("Cluster job: %s, task_index: %d, target: %s" %
(FLAGS.job_name, FLAGS.task_index, server.target))
if FLAGS.job_name == "ps":
server.join()
elif FLAGS.job_name == "worker":
# Assigns ops to the local worker by default.
with tf.device(
tf.train.replica_device_setter(
worker_device="/job:worker/task:%d" % FLAGS.task_index,
cluster=cluster)):
with slim.arg_scope(inception_v1_arg_scope()):
train_logits, end_points = inception_v1.inception_v1(
train_batch, num_classes=2, is_training=True)
tf.losses.sparse_softmax_cross_entropy(labels=train_label,
logits=train_logits)
total_loss = tf.losses.get_total_loss()
global_step = tf.Variable(0, name='global_step', trainable=False)
update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
with tf.control_dependencies(update_ops):
optimizer = tf.train.AdamOptimizer(learning_rate=0.001)
train_op = optimizer.minimize(total_loss,
global_step=global_step)
prediction_labels = tf.argmax(end_points['Predictions'], 1)
read_labels = tf.argmax(train_label_batch, 1)
correct_prediction = tf.equal(prediction_labels, read_labels)
train_accuracy_batch = tf.reduce_mean(
tf.cast(correct_prediction, "float"))
saver = tf.train.Saver(tf.trainable_variables() +
tf.get_collection_ref("moving_vars"))
#init_op = tf.global_variables_initializer()
local_init_op = tf.global_variables_initializer()
# Create a "Supervisor", which oversees the training process.
sv = tf.train.Supervisor(
is_chief=(FLAGS.task_index == 0),
logdir="./tflog",
#init_op=init_op,
local_init_op=local_init_op,
saver=saver,
global_step=global_step,
save_model_secs=600)
# The supervisor takes care of session initialization and restoring from
# a checkpoint.
config = tf.ConfigProto()
config.gpu_options.per_process_gpu_memory_fraction = 0.4
#session = tf.Session(config=config, ...)
sess = sv.prepare_or_wait_for_session(server.target, config=config)
# Start queue runners for the input pipelines (if ang).
sv.start_queue_runners(sess)
print("Initialized")
step = 0
start_time = time.time()
for epoch_index in range(num_epochs):
_, l, end_points2, logits2, train_acc2_batch = sess.run([
train_op, total_loss, end_points, train_logits,
train_accuracy_batch
])
duration = time.time() - start_time
print("Minibatch loss at step %d: %.6f (%.3f sec)" %
(step, l, duration))
print(end_points2['Predictions'])
print("Minibatch accuracy: %.6f" % train_acc2_batch)
#print("lr: %.6f" % optimizer._lr)
step += 1
sv.stop()
