def main(args):
#此处导入的是：models.inception_resnet_v1模型，以后再看怎么更改模型
network = importlib.import_module(args.model_def)
#用当前日期来命名模型
subdir = datetime.strftime(datetime.now(), '%Y%m%d-%H%M%S')
#日志保存在c:\\users\\Administrator\logs\facenet\ 文件夹里
log_dir = os.path.join(os.path.expanduser(args.logs_base_dir), subdir)
if not os.path.isdir(
log_dir): # Create the log directory if it doesn't exist
os.makedirs(log_dir) #没有日志文件就创建一个
model_dir = os.path.join(os.path.expanduser(args.models_base_dir), subdir)
if not os.path.isdir(
model_dir): # Create the model directory if it doesn't exist
os.makedirs(model_dir)
# 把参数写在日志文件中
facenet.write_arguments_to_file(args, os.path.join(log_dir,
'arguments.txt'))
# Store some git revision info in a text file in the log directory
src_path, _ = os.path.split(os.path.realpath(__file__))
#arg_string:'E:/facenet/train_tripletloss.py' output_dir:'C:\\Users\\Administrator/logs/facenet\\20180314-181556'
facenet.store_revision_info(src_path, log_dir, ' '.join(sys.argv))
np.random.seed(seed=args.seed)
#获取数据集，train_set是包含文件路径与标签的集合
#先输入一个父路径 path：'E:/facenet/data/lfw_160'，接着输入每个子路径
# 输出：一个list，每个元素是一个ImageClass，里边包含图片地址的list（image_paths）以及对应的人名(name)[以后可能会直接调用这几个属性]
train_set = facenet.get_dataset(args.data_dir)
print('Model directory: %s' % model_dir)
print('Log directory: %s' % log_dir)
if args.pretrained_model: #用在判断是否有预训练模型，但是如果有，怎么加载呢？
print('Pre-trained model: %s' %
os.path.expanduser(args.pretrained_model))
if args.lfw_dir:
print('LFW directory: %s' % args.lfw_dir)
# Read the file containing the pairs used for testing
pairs = lfw.read_pairs(os.path.expanduser(args.lfw_pairs))
# Get the paths for the corresponding images
lfw_paths, actual_issame = lfw.get_paths(
os.path.expanduser(args.lfw_dir), pairs, args.lfw_file_ext)
#建立图
#with语句适用于对资源进行访问的场合，确保使用过程中是否发生异常都会执行必要嘚瑟“清理”操作
with tf.Graph().as_default():
tf.set_random_seed(args.seed)
global_step = tf.Variable(0, trainable=False)
# 学习率 Placeholder for the learning rate
learning_rate_placeholder = tf.placeholder(tf.float32,
name='learning_rate')
#批大小
batch_size_placeholder = tf.placeholder(tf.int32, name='batch_size')
#用于判断是训练还是测试
phase_train_placeholder = tf.placeholder(tf.bool, name='phase_train')
#图像路径
image_paths_placeholder = tf.placeholder(tf.string,
shape=(None, 3),
name='image_paths')
# 图像标签
labels_placeholder = tf.placeholder(tf.int64,
shape=(None, 3),
name='labels')
#新建一个队列，数据流操作，fifo先入先出
input_queue = data_flow_ops.FIFOQueue(capacity=100000,
dtypes=[tf.string, tf.int64],
shapes=[(3, ), (3, )],
shared_name=None,
name=None)
#enqueue_many返回的是一个操作
enqueue_op = input_queue.enqueue_many(
[image_paths_placeholder, labels_placeholder])
nrof_preprocess_threads = 4
images_and_labels = []
for _ in range(nrof_preprocess_threads):
filenames, label = input_queue.dequeue()
images = []
for filename in tf.unstack(filenames):
file_contents = tf.read_file(filename)
image = tf.image.decode_image(file_contents, channels=3)
if args.random_crop:
image = tf.random_crop(
image, [args.image_size, args.image_size, 3])
else:
image = tf.image.resize_image_with_crop_or_pad(
image, args.image_size, args.image_size)
if args.random_flip:
image = tf.image.random_flip_left_right(image)
#pylint: disable=no-member
image.set_shape((args.image_size, args.image_size, 3))
images.append(tf.image.per_image_standardization(image))
images_and_labels.append([images, label])
image_batch, labels_batch = tf.train.batch_join(
images_and_labels,
batch_size=batch_size_placeholder,
shapes=[(args.image_size, args.image_size, 3), ()],
enqueue_many=True,
capacity=4 * nrof_preprocess_threads * args.batch_size,
allow_smaller_final_batch=True)
image_batch = tf.identity(image_batch, 'image_batch')
image_batch = tf.identity(image_batch, 'input')
labels_batch = tf.identity(labels_batch, 'label_batch')
# Build the inference (构造计算图)
#其中prelogits是最后一层的输出
prelogits, _ = network.inference(
image_batch,
args.keep_probability,
phase_train=phase_train_placeholder,
bottleneck_layer_size=args.embedding_size,
weight_decay=args.weight_decay)
#L2正则化（范化）函数
# embeddings = tf.nn.l2_normalize(输入向量, L2范化的维数（取0（列L2范化）或1（行L2范化））, 泛化的最小值边界, name='embeddings')
embeddings = tf.nn.l2_normalize(prelogits, 1, 1e-10, name='embeddings')
# Split embeddings into anchor, positive and negative and calculate triplet loss
anchor, positive, negative = tf.unstack(
tf.reshape(embeddings, [-1, 3, args.embedding_size]), 3, 1)
triplet_loss = facenet.triplet_loss(anchor, positive, negative,
args.alpha)
#将指数衰减应用在学习率上
learning_rate = tf.train.exponential_decay(
learning_rate_placeholder,
global_step,
args.learning_rate_decay_epochs * args.epoch_size,
args.learning_rate_decay_factor,
staircase=True)
tf.summary.scalar('learning_rate', learning_rate)
# 计算损失
regularization_losses = tf.get_collection(
tf.GraphKeys.REGULARIZATION_LOSSES)
#构建L2正则化
total_loss = tf.add_n([triplet_loss] + regularization_losses,
name='total_loss')
# Build a Graph that trains the model with one batch of examples and updates the model parameters
# 确定优化方法并根据损失函数求梯度，在这里，每更行一次参数，global_step会加1
train_op = facenet.train(total_loss, global_step, args.optimizer,
learning_rate, args.moving_average_decay,
tf.global_variables())
# Create a saver创建一个saver用来保存或者从内存中回复一个模型参数
saver = tf.train.Saver(tf.trainable_variables(), max_to_keep=3)
# Build the summary operation based on the TF collection of Summaries.
summary_op = tf.summary.merge_all()
# Start running operations on the Graph.能够在GPU上分配的最大内存
gpu_options = tf.GPUOptions(
per_process_gpu_memory_fraction=args.gpu_memory_fraction)
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
# Initialize variables
sess.run(tf.global_variables_initializer(),
feed_dict={phase_train_placeholder: True})
sess.run(tf.local_variables_initializer(),
feed_dict={phase_train_placeholder: True})
#写log文件
summary_writer = tf.summary.FileWriter(log_dir, sess.graph)
#获取线程坐标
coord = tf.train.Coordinator()
#将队列中的多用sunner开始执行
tf.train.start_queue_runners(coord=coord, sess=sess)
with sess.as_default():
if args.pretrained_model:
print('Restoring pretrained model: %s' % args.pretrained_model)
saver.restore(sess, os.path.expanduser(args.pretrained_model))
# Training and validation loop
epoch = 0
#将所有数据过一遍的次数 默认500
while epoch < args.max_nrof_epochs:
step = sess.run(global_step, feed_dict=None)
#epoch_size是一个epoch中批的个数，这个epoch是全局的批处理个数以一个epoch中。。。这个epoch将用于求学习率
epoch = step // args.epoch_size
# Train for one epoch
train(args, sess, train_set, epoch, image_paths_placeholder,
labels_placeholder, labels_batch, batch_size_placeholder,
learning_rate_placeholder, phase_train_placeholder,
enqueue_op, input_queue, global_step, embeddings,
total_loss, train_op, summary_op, summary_writer,
args.learning_rate_schedule_file, args.embedding_size,
anchor, positive, negative, triplet_loss)
# Save variables and the metagraph if it doesn't exist already
save_variables_and_metagraph(sess, saver, summary_writer,
model_dir, subdir, step)
# Evaluate on LFW
if args.lfw_dir:
evaluate(sess, lfw_paths, embeddings, labels_batch,
image_paths_placeholder, labels_placeholder,
batch_size_placeholder, learning_rate_placeholder,
phase_train_placeholder, enqueue_op,
actual_issame, args.batch_size,
args.lfw_nrof_folds, log_dir, step,
summary_writer, args.embedding_size)
return model_dir
def main(args):
network = importlib.import_module(args.model_def, 'inference')
subdir = datetime.strftime(datetime.now(), '%Y%m%d-%H%M%S')
log_dir = os.path.join(os.path.expanduser(args.logs_base_dir), subdir)
if not os.path.isdir(log_dir): # Create the log directory if it doesn't exist
os.makedirs(log_dir)
model_dir = os.path.join(os.path.expanduser(args.models_base_dir), subdir)
if not os.path.isdir(model_dir): # Create the model directory if it doesn't exist
os.makedirs(model_dir)
subdirmaxlin= subdir+'_lin_max' #fbtian_max
maxlin_dir = os.path.join(os.path.expanduser(args.models_base_dir), subdirmaxlin)#fbtian_max
if not os.path.exists(maxlin_dir):#fbtian_max
os.makedirs(maxlin_dir)#fbtian_max
subdirmax= subdir+'_max' #fbtian_max
modelmax_dir = os.path.join(os.path.expanduser(args.models_base_dir), subdirmax)#fbtian_max
if not os.path.exists(modelmax_dir):#fbtian_max
os.makedirs(modelmax_dir)#fbtian_max
# Store some git revision info in a text file in the log directory
if not args.no_store_revision_info:
src_path,_ = os.path.split(os.path.realpath(__file__))
facenet.store_revision_info(src_path, log_dir, ' '.join(sys.argv))
np.random.seed(seed=args.seed)
train_set = facenet.get_dataset(args.data_dir)
#for i in range(args.send2):
# np.random.shuffle(np.arange(10))
print('Model directory: %s' % model_dir)
print('Log directory: %s' % log_dir)
if args.pretrained_model:
print('Pre-trained model: %s' % os.path.expanduser(args.pretrained_model))
if args.lfw_dir:
print('LFW directory: %s' % args.lfw_dir)
# Read the file containing the pairs used for testing
pairs = lfw.read_pairs(os.path.expanduser(args.lfw_pairs))
# Get the paths for the corresponding images
lfw_paths, actual_issame = lfw.get_paths(os.path.expanduser(args.lfw_dir), pairs, args.lfw_file_ext)
with tf.Graph().as_default():
tf.set_random_seed(args.seed)
global_step = tf.Variable(0, trainable=False)
# Placeholder for the learning rate
learning_rate_placeholder = tf.placeholder(tf.float32, name='learning_rate')
batch_size_placeholder = tf.placeholder(tf.int32, name='batch_size')
phase_train_placeholder = tf.placeholder(tf.bool, name='phase_train')
image_paths_placeholder = tf.placeholder(tf.string, shape=(None,3), name='image_paths')
labels_placeholder = tf.placeholder(tf.int64, shape=(None,3), name='labels')
input_queue = data_flow_ops.FIFOQueue(capacity=100000,
dtypes=[tf.string, tf.int64],
shapes=[(3,), (3,)],
shared_name=None, name=None)
enqueue_op = input_queue.enqueue_many([image_paths_placeholder, labels_placeholder])
nrof_preprocess_threads = 4
images_and_labels = []
for _ in range(nrof_preprocess_threads):
filenames, label = input_queue.dequeue()
images = []
fb_count=0
for filename in tf.unstack(filenames):
file_contents = tf.read_file(filename)
print('filename:%s'%filename )
image = tf.image.decode_png(file_contents)
if args.random_crop:
print('args.random_crop') #fbtian_add
image = tf.random_crop(image, [args.image_size, args.image_size, 3])
else:
#print('else not args.random_crop') #come in fbtian_add
image = tf.image.resize_image_with_crop_or_pad(image, args.image_size, args.image_size)
if args.random_flip:
print('args.random_flip')
image = tf.image.random_flip_left_right(image)
if 1 :
image = tf.image.random_brightness(image, max_delta=0.2) #Random brightness transformation
image = tf.image.random_contrast(image, lower=0.2, upper=1.0)#Random contrast transformation
fb_count+=1
#pylint: disable=no-member# fbtian_add
image.set_shape((args.image_size, args.image_size, 3))
images.append(tf.image.per_image_standardization(image))
images_and_labels.append([images, label])
print('fb_count:%d'%fb_count)
image_batch, labels_batch = tf.train.batch_join(
images_and_labels, batch_size=batch_size_placeholder,
shapes=[(args.image_size, args.image_size, 3), ()], enqueue_many=True,
capacity=4 * nrof_preprocess_threads * args.batch_size,
allow_smaller_final_batch=True)
image_batch = tf.identity(image_batch, 'input') ##fbtian
batch_norm_params = {
# Decay for the moving averages
'decay': 0.995,
# epsilon to prevent 0s in variance
'epsilon': 0.001,
# force in-place updates of mean and variance estimates
'updates_collections': None,
# Moving averages ends up in the trainable variables collection
'variables_collections': [ tf.GraphKeys.TRAINABLE_VARIABLES ],
# Only update statistics during training mode
'is_training': phase_train_placeholder
}
# Build the inference graph
prelogits, _ = network.inference(image_batch, args.keep_probability,
phase_train=phase_train_placeholder, weight_decay=args.weight_decay)
pre_embeddings = slim.fully_connected(prelogits, args.embedding_size, activation_fn=None,
weights_initializer=tf.truncated_normal_initializer(stddev=0.1),
weights_regularizer=slim.l2_regularizer(args.weight_decay),
normalizer_fn=slim.batch_norm,
normalizer_params=batch_norm_params,
scope='Bottleneck', reuse=False)
embeddings = tf.nn.l2_normalize(pre_embeddings, 1, 1e-10, name='embeddings')
# Split embeddings into anchor, positive and negative and calculate triplet loss
anchor, positive, negative = tf.unstack(tf.reshape(embeddings, [-1,3,args.embedding_size]), 3, 1)
triplet_loss = facenet.triplet_loss(anchor, positive, negative, args.alpha)
learning_rate = tf.train.exponential_decay(learning_rate_placeholder, global_step,
args.learning_rate_decay_epochs*args.epoch_size, args.learning_rate_decay_factor, staircase=True)
tf.summary.scalar('learning_rate', learning_rate)
# Calculate the total losses
regularization_losses = tf.get_collection(tf.GraphKeys.REGULARIZATION_LOSSES)
total_loss = tf.add_n([triplet_loss] + regularization_losses, name='total_loss')
# Build a Graph that trains the model with one batch of examples and updates the model parameters
train_op = facenet.train(total_loss, global_step, args.optimizer,
learning_rate, args.moving_average_decay, tf.global_variables())
# Create a saver
saver = tf.train.Saver(tf.trainable_variables(), max_to_keep=3)
# Build the summary operation based on the TF collection of Summaries.
summary_op = tf.summary.merge_all()
# Start running operations on the Graph.
#gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=args.gpu_memory_fraction)###
#sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
config = tf.ConfigProto(allow_soft_placement=True) ###########################################
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.7)###################################
config.gpu_options.allow_growth = True###########################################
#sess = tf.Session(config=config)###########################################
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options,intra_op_parallelism_threads=8))
# Initialize variables
sess.run(tf.global_variables_initializer(), feed_dict={phase_train_placeholder:True})
sess.run(tf.local_variables_initializer(), feed_dict={phase_train_placeholder:True})
summary_writer = tf.summary.FileWriter(log_dir, sess.graph)
coord = tf.train.Coordinator()
tf.train.start_queue_runners(coord=coord, sess=sess)
with sess.as_default():
if args.pretrained_model:
print('Restoring pretrained model: %s' % args.pretrained_model)
saver.restore(sess, os.path.expanduser(args.pretrained_model))
# Training and validation loop
epoch = 0
acc_tmp=0
val_tmp=0
while epoch < args.max_nrof_epochs:
step = sess.run(global_step, feed_dict=None)
print(global_step )
epoch = step // args.epoch_size
# Train for one epoch
train(args, sess, train_set, epoch, image_paths_placeholder, labels_placeholder, labels_batch,
batch_size_placeholder, learning_rate_placeholder, phase_train_placeholder, enqueue_op, input_queue, global_step,
embeddings, total_loss, train_op, summary_op, summary_writer, args.learning_rate_schedule_file,
args.embedding_size, anchor, positive, negative, triplet_loss)
# Save variables and the metagraph if it doesn't exist already
save_variables_and_metagraph(sess, saver, summary_writer, model_dir, subdir, step)
# Evaluate on LFW
if args.lfw_dir:
acc,val=evaluate(sess, lfw_paths, embeddings, labels_batch, image_paths_placeholder, labels_placeholder,
batch_size_placeholder, learning_rate_placeholder, phase_train_placeholder, enqueue_op, actual_issame, args.batch_size,
args.lfw_nrof_folds, log_dir, step, summary_writer, args.embedding_size)
print('starting to save the maxacc and maxval ') #fbtian_max
if acc>acc_tmp: #fbtian_max
maxmodel_path = os.path.join(maxlin_dir, 'model-%s.ckpt_accmax'%subdir) #fbtian_max
saver.save(sess, maxmodel_path, write_meta_graph=False)#fbtian_max
shutil.copy( maxmodel_path+'.data-00000-of-00001', modelmax_dir)
shutil.copy( maxmodel_path+'.index', modelmax_dir)
acc_tmp=acc #fbtian_max
if val>val_tmp: #fbtian_max
maxmodel_path = os.path.join(maxlin_dir, 'model-%s.ckpt_valmax'%subdir) #fbtian_max
saver.save(sess, maxmodel_path, write_meta_graph=False)#fbtian_max
shutil.copy( maxmodel_path+'.data-00000-of-00001', modelmax_dir)
shutil.copy( maxmodel_path+'.index', modelmax_dir)
val_tmp=val #fbtian
print('end to save the maxacc and maxval ') #fbtian
sess.close()
return model_dir
def transform(args, pretrained_ckpt, image_size, output_dir, bottleneck_size):
with tf.Graph().as_default():
learning_rate_placeholder = tf.placeholder(tf.float32,
name='learning_rate')
batch_size_placeholder = tf.placeholder(tf.int32, name='batch_size')
phase_train_placeholder = tf.placeholder(tf.bool, name='phase_train')
image_paths_placeholder = tf.placeholder(tf.string,
shape=(None, 3),
name='image_paths')
labels_placeholder = tf.placeholder(tf.int64,
shape=(None, 3),
name='labels')
input_queue = data_flow_ops.FIFOQueue(capacity=100000,
dtypes=[tf.string, tf.int64],
shapes=[(3, ), (3, )],
shared_name=None,
name=None)
enqueue_op = input_queue.enqueue_many(
[image_paths_placeholder, labels_placeholder])
nrof_preprocess_threads = 4
images_and_labels = []
for _ in range(nrof_preprocess_threads):
filenames, label = input_queue.dequeue()
images = []
for filename in tf.unstack(filenames):
file_contents = tf.read_file(filename)
image = tf.image.decode_image(file_contents, channels=3)
processed_image = inception_preprocessing.preprocess_image(
image, args.image_size, args.image_size, is_training=False)
# if args.random_crop:
# image = tf.random_crop(image, [args.image_size, args.image_size, 3])
# else:
# image = tf.image.resize_image_with_crop_or_pad(image, args.image_size, args.image_size)
if args.random_flip:
processed_image = tf.image.random_flip_left_right(
processed_image)
images.append(processed_image)
images_and_labels.append([images, label])
image_batch, labels_batch = tf.train.batch_join(
images_and_labels,
batch_size=batch_size_placeholder,
shapes=[(args.image_size, args.image_size, 3), ()],
enqueue_many=True,
capacity=4 * nrof_preprocess_threads * args.batch_size,
allow_smaller_final_batch=True)
image_batch = tf.identity(image_batch, 'image_batch')
image_batch = tf.identity(image_batch, 'input')
labels_batch = tf.identity(labels_batch, 'label_batch')
with slim.arg_scope(
inception_resnet_v2.inception_resnet_v2_arg_scope(
weight_decay=args.weight_decay)):
prelogits, _ = inception_resnet_v2.inception_resnet_v2(
image_batch,
num_classes=bottleneck_size,
is_training=phase_train_placeholder)
exclude = ['InceptionResnetV2/Logits', 'InceptionResnetV2/AuxLogits']
variables_to_restore = slim.get_variables_to_restore(exclude=exclude)
loader = tf.train.Saver(variables_to_restore)
global_step = tf.train.get_or_create_global_step()
embeddings = tf.nn.l2_normalize(prelogits, 1, 1e-10, name='embeddings')
# Split embeddings into anchor, positive and negative and calculate triplet loss
anchor, positive, negative = tf.unstack(
tf.reshape(embeddings, [-1, 3, args.embedding_size]), 3, 1)
triplet_loss = facenet.triplet_loss(anchor, positive, negative,
args.alpha)
learning_rate = tf.train.exponential_decay(
learning_rate_placeholder,
global_step,
args.learning_rate_decay_epochs * args.epoch_size,
args.learning_rate_decay_factor,
staircase=True)
tf.summary.scalar('learning_rate', learning_rate)
# Calculate the total losses
regularization_losses = tf.get_collection(
tf.GraphKeys.REGULARIZATION_LOSSES)
total_loss = tf.add_n([triplet_loss] + regularization_losses,
name='total_loss')
# Build a Graph that trains the model with one batch of examples and updates the model parameters
train_layers = [
'Logits', 'Conv2d_7b_1x1', 'Block8', 'Repeat_2', 'Mixed_7a'
]
var_list = []
for v in tf.global_variables():
splits = v.name.split("/")
if len(splits) > 2 and splits[1] in train_layers:
var_list.append(v)
facenet.train(total_loss, global_step, args.optimizer, learning_rate,
args.moving_average_decay, var_list)
saver = tf.train.Saver(max_to_keep=3)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
loader.restore(sess, pretrained_ckpt)
checkpoint_path = os.path.join(output_dir,
'model-%s.ckpt' % bottleneck_size)
saver.save(sess, checkpoint_path, write_meta_graph=False)
metagraph_filename = os.path.join(
output_dir, 'model-%s.meta' % bottleneck_size)
saver.export_meta_graph(metagraph_filename)
def main(args):
print(args.model_def)
network = importlib.import_module(args.model_def, 'inference')
print(1111)
subdir = datetime.strftime(datetime.now(), '%Y%m%d-%H%M%S')
log_dir = os.path.join(os.path.expanduser(args.logs_base_dir), subdir)
if not os.path.isdir(
log_dir): # Create the log directory if it doesn't exist
os.makedirs(log_dir)
model_dir = os.path.join(os.path.expanduser(args.models_base_dir), subdir)
if not os.path.isdir(
model_dir): # Create the model directory if it doesn't exist
os.makedirs(model_dir)
# Store some git revision info in a text file in the log directory
src_path, _ = os.path.split(os.path.realpath(__file__))
facenet.store_revision_info(src_path, log_dir, ' '.join(sys.argv))
np.random.seed(seed=args.seed)
train_set = facenet.get_dataset(args.data_dir)
print('Model directory: %s' % model_dir)
print('Log directory: %s' % log_dir)
if args.pretrained_model:
print('Pre-trained model: %s' %
os.path.expanduser(args.pretrained_model))
if args.lfw_dir:
print('LFW directory: %s' % args.lfw_dir)
# Read the file containing the pairs used for testing
pairs = lfw.read_pairs(os.path.expanduser(args.lfw_pairs))
# Get the paths for the corresponding images
lfw_paths, actual_issame = lfw.get_paths(
os.path.expanduser(args.lfw_dir), pairs, args.lfw_file_ext)
with tf.Graph().as_default():
tf.set_random_seed(args.seed)
global_step = tf.Variable(0, trainable=False)
# Placeholder for the learning rate
learning_rate_placeholder = tf.placeholder(tf.float32,
name='learning_rate')
batch_size_placeholder = tf.placeholder(tf.int32, name='batch_size')
phase_train_placeholder = tf.placeholder(tf.bool, name='phase_train')
image_paths_placeholder = tf.placeholder(tf.string,
shape=(None, 3),
name='image_paths')
labels_placeholder = tf.placeholder(tf.int64,
shape=(None, 3),
name='labels')
input_queue = data_flow_ops.FIFOQueue(capacity=100000,
dtypes=[tf.string, tf.int64],
shapes=[(3, ), (3, )],
shared_name=None,
name=None)
enqueue_op = input_queue.enqueue_many(
[image_paths_placeholder, labels_placeholder])
nrof_preprocess_threads = 4
images_and_labels = []
for _ in range(nrof_preprocess_threads):
filenames, label = input_queue.dequeue()
images = []
for filename in tf.unstack(filenames):
file_contents = tf.read_file(filename)
image = tf.image.decode_png(file_contents)
if args.random_crop:
image = tf.random_crop(
image, [args.image_size, args.image_size, 3])
else:
image = tf.image.resize_image_with_crop_or_pad(
image, args.image_size, args.image_size)
if args.random_flip:
image = tf.image.random_flip_left_right(image)
#pylint: disable=no-member
image.set_shape((args.image_size, args.image_size, 3))
images.append(tf.image.per_image_standardization(image))
images_and_labels.append([images, label])
image_batch, labels_batch = tf.train.batch_join(
images_and_labels,
batch_size=batch_size_placeholder,
shapes=[(args.image_size, args.image_size, 3), ()],
enqueue_many=True,
capacity=4 * nrof_preprocess_threads * args.batch_size,
allow_smaller_final_batch=True)
# Build the inference graph
prelogits, _ = network.inference(
image_batch,
args.keep_probability,
phase_train=phase_train_placeholder,
bottleneck_layer_size=args.embedding_size,
weight_decay=args.weight_decay)
embeddings = tf.nn.l2_normalize(prelogits, 1, 1e-10, name='embeddings')
# Split embeddings into anchor, positive and negative and calculate triplet loss
anchor, positive, negative = tf.unstack(
tf.reshape(embeddings, [-1, 3, args.embedding_size]), 3, 1)
triplet_loss = facenet.triplet_loss(anchor, positive, negative,
args.alpha)
learning_rate = tf.train.exponential_decay(
learning_rate_placeholder,
global_step,
args.learning_rate_decay_epochs * args.epoch_size,
args.learning_rate_decay_factor,
staircase=True)
tf.summary.scalar('learning_rate', learning_rate)
# Calculate the total losses
regularization_losses = tf.get_collection(
tf.GraphKeys.REGULARIZATION_LOSSES)
total_loss = tf.add_n([triplet_loss] + regularization_losses,
name='total_loss')
# Build a Graph that trains the model with one batch of examples and updates the model parameters
train_op = facenet.train(total_loss, global_step, args.optimizer,
learning_rate, args.moving_average_decay,
tf.global_variables())
# Create a saver
saver = tf.train.Saver(tf.trainable_variables(), max_to_keep=3)
# Build the summary operation based on the TF collection of Summaries.
summary_op = tf.summary.merge_all()
# Start running operations on the Graph.
gpu_options = tf.GPUOptions(
per_process_gpu_memory_fraction=args.gpu_memory_fraction)
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
# Initialize variables
sess.run(tf.global_variables_initializer(),
feed_dict={phase_train_placeholder: True})
sess.run(tf.local_variables_initializer(),
feed_dict={phase_train_placeholder: True})
summary_writer = tf.summary.FileWriter(log_dir, sess.graph)
coord = tf.train.Coordinator()
tf.train.start_queue_runners(coord=coord, sess=sess)
with sess.as_default():
if args.pretrained_model:
print('Restoring pretrained model: %s' % args.pretrained_model)
saver.restore(sess, os.path.expanduser(args.pretrained_model))
# Training and validation loop
epoch = 0
while epoch < args.max_nrof_epochs:
step = sess.run(global_step, feed_dict=None)
epoch = step // args.epoch_size
# Train for one epoch
train(args, sess, train_set, epoch, image_paths_placeholder,
labels_placeholder, labels_batch, batch_size_placeholder,
learning_rate_placeholder, phase_train_placeholder,
enqueue_op, input_queue, global_step, embeddings,
total_loss, train_op, summary_op, summary_writer,
args.learning_rate_schedule_file, args.embedding_size,
anchor, positive, negative, triplet_loss)
# Save variables and the metagraph if it doesn't exist already
save_variables_and_metagraph(sess, saver, summary_writer,
model_dir, subdir, step)
# Evaluate on LFW
if args.lfw_dir:
evaluate(sess, lfw_paths, embeddings, labels_batch,
image_paths_placeholder, labels_placeholder,
batch_size_placeholder, learning_rate_placeholder,
phase_train_placeholder, enqueue_op,
actual_issame, args.batch_size,
args.lfw_nrof_folds, log_dir, step,
summary_writer, args.embedding_size)
sess.close()
return model_dir
def train(server, cluster_spec, args, ctx):
task_index = ctx.task_index
num_workers = len(cluster_spec['worker'])
is_chief = task_index == 0
local_data_path = args.local_data_path
data_dir = os.path.join(local_data_path, "train")
val_dir = os.path.join(local_data_path, "val")
val_pairs = os.path.join(local_data_path, "pairs.txt")
subdir = datetime.strftime(datetime.now(), '%Y%m%d-%H%M%S')
log_dir = args.workspace + "/logs/" + subdir
checkpoint_dir = args.checkpoint_dir if args.checkpoint_dir else (
args.workspace + "/models/" + subdir)
if task_index == 0:
if not tf.gfile.Exists(args.workspace):
tf.gfile.MakeDirs(args.workspace)
if not tf.gfile.Exists(checkpoint_dir):
tf.gfile.MakeDirs(checkpoint_dir)
if not tf.gfile.Exists(log_dir):
tf.gfile.MakeDirs(log_dir)
if is_chief and args.transfer_learning:
print("Transforming the pretrained inception model...")
transform_pretrained.transform(args, args.pretrained_ckpt,
args.image_size, checkpoint_dir,
args.embedding_size)
print("Transform finished")
tf.reset_default_graph()
seed = random.SystemRandom().randint(0, 10240)
print("Random seed: " + str(seed))
np.random.seed(seed=seed)
train_set = facenet.get_dataset(data_dir)
print('Model directory: %s' % checkpoint_dir)
print('Log directory: %s' % log_dir)
# Read the file containing the pairs used for testing
pairs = read_pairs(val_pairs)
# Get the paths for the corresponding images
val_image_paths, actual_issame = get_paths(val_dir, pairs)
with tf.device(
tf.train.replica_device_setter(
worker_device="/job:worker/task:%d" % task_index,
cluster=cluster_spec)):
tf.set_random_seed(seed)
# Placeholder for the learning rate
learning_rate_placeholder = tf.placeholder(tf.float32,
name='learning_rate')
batch_size_placeholder = tf.placeholder(tf.int32, name='batch_size')
phase_train_placeholder = tf.placeholder(tf.bool, name='phase_train')
image_paths_placeholder = tf.placeholder(tf.string,
shape=(None, 3),
name='image_paths')
labels_placeholder = tf.placeholder(tf.int64,
shape=(None, 3),
name='labels')
input_queue = data_flow_ops.FIFOQueue(capacity=10000,
dtypes=[tf.string, tf.int64],
shapes=[(3, ), (3, )],
shared_name=None,
name=None)
enqueue_op = input_queue.enqueue_many(
[image_paths_placeholder, labels_placeholder])
nrof_preprocess_threads = 4
images_and_labels = []
for _ in range(nrof_preprocess_threads):
filenames, label = input_queue.dequeue()
images = []
for filename in tf.unstack(filenames):
file_contents = tf.read_file(filename)
image = tf.image.decode_image(file_contents, channels=3)
processed_image = inception_preprocessing.preprocess_image(
image, args.image_size, args.image_size, is_training=False)
if args.random_flip:
processed_image = tf.image.random_flip_left_right(
processed_image)
images.append(processed_image)
images_and_labels.append([images, label])
image_batch, labels_batch = tf.train.batch_join(
images_and_labels,
batch_size=batch_size_placeholder,
shapes=[(args.image_size, args.image_size, 3), ()],
enqueue_many=True,
capacity=4 * nrof_preprocess_threads * args.batch_size,
allow_smaller_final_batch=True)
image_batch = tf.identity(image_batch, 'image_batch')
image_batch = tf.identity(image_batch, 'input')
labels_batch = tf.identity(labels_batch, 'label_batch')
with slim.arg_scope(
inception_resnet_v2_arg_scope(weight_decay=args.weight_decay)):
prelogits, _ = inception_resnet_v2(
image_batch,
num_classes=args.embedding_size,
is_training=phase_train_placeholder)
global_step = tf.train.get_or_create_global_step()
embeddings = tf.nn.l2_normalize(prelogits, 1, 1e-10, name='embeddings')
# Split embeddings into anchor, positive and negative and calculate triplet loss
anchor, positive, negative = tf.unstack(
tf.reshape(embeddings, [-1, 3, args.embedding_size]), 3, 1)
triplet_loss = facenet.triplet_loss(anchor, positive, negative,
args.alpha)
learning_rate = tf.train.exponential_decay(
learning_rate_placeholder,
global_step,
args.learning_rate_decay_epochs * args.epoch_size,
args.learning_rate_decay_factor,
staircase=True)
# Calculate the total losses
regularization_losses = tf.get_collection(
tf.GraphKeys.REGULARIZATION_LOSSES)
total_loss = tf.add_n([triplet_loss] + regularization_losses,
name='total_loss')
tf.summary.scalar('learning_rate', learning_rate)
tf.summary.scalar('triplet_loss', triplet_loss)
tf.summary.scalar('total_losses', total_loss)
# Build a Graph that trains the model with one batch of examples and updates the model parameters
train_layers = [
'Logits', 'Conv2d_7b_1x1', 'Block8', 'Repeat_2', 'Mixed_7a'
]
var_list = []
for v in tf.global_variables():
splits = v.name.split("/")
if len(splits) > 2 and splits[1] in train_layers:
var_list.append(v)
train_op, opt = facenet.train(total_loss,
global_step,
args.optimizer,
learning_rate,
args.moving_average_decay,
var_list,
sync_replicas=args.sync_replicas,
replicas_to_aggregate=num_workers)
summary_op = tf.summary.merge_all()
hooks = [
tf.train.StopAtStepHook(int(args.epochs) * int(args.epoch_size))
]
if args.sync_replicas:
hooks += [opt.make_session_run_hook(is_chief)]
sess_config = tf.ConfigProto(
allow_soft_placement=True,
log_device_placement=False,
device_filters=['/job:ps',
'/job:worker/task:%d' % task_index])
#sess_config.operation_timeout_in_ms=80000
with tf.train.MonitoredTrainingSession(
master=server.target,
is_chief=is_chief,
config=sess_config,
checkpoint_dir=checkpoint_dir,
hooks=hooks,
save_summaries_steps=None,
save_summaries_secs=None,
stop_grace_period_secs=30,
save_checkpoint_secs=600) as sess:
# Training and validation loop
summary_writer = tf.summary.FileWriter(
log_dir, sess.graph) if is_chief else None
while not sess.should_stop():
# Train for one epoch
step = _train(args, sess, train_set, image_paths_placeholder,
labels_placeholder, labels_batch,
batch_size_placeholder,
learning_rate_placeholder,
phase_train_placeholder, enqueue_op, input_queue,
global_step, embeddings, total_loss, train_op,
args.embedding_size, triplet_loss, summary_op,
summary_writer)
if is_chief:
# checkpoint_path = os.path.join(checkpoint_dir, 'model-%s.ckpt' % "test")
# saver.save(sess._sess._sess._sess._sess, checkpoint_path, global_step=step, write_meta_graph=False)
evaluate(sess, val_image_paths, embeddings, labels_batch,
image_paths_placeholder, labels_placeholder,
batch_size_placeholder, learning_rate_placeholder,
phase_train_placeholder, enqueue_op,
actual_issame, args.batch_size,
args.lfw_nrof_folds, step, summary_writer,
args.embedding_size)
return checkpoint_dir