def __init__(self):
self.__test_input_size = cfg.TEST_INPUT_SIZE
self.__anchor_per_scale = cfg.ANCHOR_PER_SCALE
self.__classes = cfg.CLASSES
self.__num_classes = len(self.__classes)
self.__class_to_ind = dict(
zip(self.__classes, range(self.__num_classes)))
self.__anchors = np.array(cfg.ANCHORS)
self.__score_threshold = cfg.SCORE_THRESHOLD
self.__iou_threshold = cfg.IOU_THRESHOLD
self.__log_dir = os.path.join(cfg.LOG_DIR, 'test')
self.__annot_dir_path = cfg.ANNOT_DIR_PATH
self.__moving_ave_decay = cfg.MOVING_AVE_DECAY
self.__dataset_path = cfg.DATASET_PATH
self.__valid_scales = cfg.VALID_SCALES
self.__training = False
with tf.name_scope('input'):
self.__input_data = tf.placeholder(dtype=tf.float32,
name='input_data')
# self.__training = tf.placeholder(dtype=tf.bool, name='training')
_, _, _, self.__pred_sbbox, self.__pred_mbbox, self.__pred_lbbox = \
YOLO_V3(self.__training).build_nework(self.__input_data)
with tf.name_scope('summary'):
tf.summary.FileWriter(self.__log_dir).add_graph(
tf.get_default_graph())
with tf.name_scope('ema'):
ema_obj = tf.train.ExponentialMovingAverage(
self.__moving_ave_decay)
self.__sess = tf.Session(config=config)
self.__saver = tf.train.Saver(ema_obj.variables_to_restore())
self.__saver.restore(self.__sess,
os.path.join(cfg.WEIGHTS_DIR, cfg.WEIGHTS_FILE))
def freeze_graph(input_checkpoint, output_graph):
# 指定输出的节点名称,该节点名称必须是原模型中存在的节点
output_node_names = [
"input/input_data", "pred_lbbox/pred_bbox", "pred_sbbox/pred_bbox",
"pred_mbbox/pred_bbox"
]
import config as cfg
from model.yolo_v3 import YOLO_V3
__training = False
with tf.name_scope('input'):
__input_data = tf.placeholder(dtype=tf.float32, name='input_data')
#__training = tf.placeholder(dtype=tf.bool, name='training')
_, _, _, __pred_sbbox, __pred_mbbox, __pred_lbbox = YOLO_V3(
__training).build_nework(__input_data)
__moving_ave_decay = cfg.MOVING_AVE_DECAY
with tf.name_scope('ema'):
ema_obj = tf.train.ExponentialMovingAverage(__moving_ave_decay)
saver = tf.train.Saver(ema_obj.variables_to_restore())
# saver = tf.train.import_meta_graph(input_checkpoint + '.meta', clear_devices=True) #得到图、clear_devices ：Whether or not to clear the device field for an `Operation` or `Tensor` during import.
graph = tf.get_default_graph() #获得默认的图
input_graph_def = graph.as_graph_def() #返回一个序列化的图代表当前的图
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
saver.restore(sess, input_checkpoint) #恢复图并得到数据
print("loading model ...")
for node in input_graph_def.node:
if node.op == 'RefSwitch':
node.op = 'Switch'
for index in xrange(len(node.input)):
if 'moving_' in node.input[index]:
node.input[index] = node.input[index] + '/read'
elif node.op == 'AssignSub':
node.op = 'Sub'
if 'use_locking' in node.attr: del node.attr['use_locking']
elif node.op == 'AssignAdd':
node.op = 'Add'
if 'use_locking' in node.attr: del node.attr['use_locking']
elif node.op == 'Assign':
node.op = 'Identity'
if 'use_locking' in node.attr: del node.attr['use_locking']
if 'validate_shape' in node.attr:
del node.attr['validate_shape']
if len(node.input) == 2:
# input0: ref: Should be from a Variable node. May be uninitialized.
# input1: value: The value to be assigned to the variable.
node.input[0] = node.input[1]
del node.input[1]
#print ("predictions : ", sess.run("predictions:0", feed_dict={"input_holder:0": [10.0]})) # 测试读出来的模型是否正确，注意这里传入的是输出 和输入 节点的 tensor的名字，不是操作节点的名字
output_graph_def = graph_util.convert_variables_to_constants( # 模型持久化，将变量值固定
sess=sess,
input_graph_def=sess.graph.as_graph_def(), # 等于:sess.graph_def
output_node_names=output_node_names)
with tf.gfile.GFile(output_graph, "wb") as f: #保存模型
f.write(output_graph_def.SerializeToString()) #序列化输出
print("%d ops in the final graph." %
len(output_graph_def.node)) #得到当前图有几个操作节点
if (var_name_mess[-1] not in ['kernel', 'gamma', 'beta', 'moving_mean', 'moving_variance']) or \
(var_name_mess[0] in preserve_org_names):
continue
org_weights_mess.append([var_name, var_shape])
print str(var_name).ljust(50), var_shape
print
tf.reset_default_graph()
cur_weights_mess = []
tf.Graph().as_default()
with tf.name_scope('input'):
input_data = tf.placeholder(dtype=tf.float32,
shape=(1, 416, 416, 3),
name='input_data')
training = tf.placeholder(dtype=tf.bool, name='training')
YOLO_V3(training).build_nework(input_data)
for var in tf.global_variables():
var_name = var.op.name
var_name_mess = str(var_name).split('/')
var_shape = var.shape
if var_name_mess[0] in preserve_cur_names:
continue
cur_weights_mess.append([var_name, var_shape])
print str(var_name).ljust(50), var_shape
org_weights_num = len(org_weights_mess)
cur_weights_num = len(cur_weights_mess)
if cur_weights_num != org_weights_num:
raise RuntimeError
print 'Number of weights that will rename:\t%d' % cur_weights_num
def __init__(self):
self.__anchor_per_scale = cfg.ANCHOR_PER_SCALE
self.__classes = cfg.CLASSES
self.__num_classes = len(self.__classes)
self.__learn_rate_init = cfg.LEARN_RATE_INIT
self.__max_periods = cfg.MAX_PERIODS
self.__max_wave_time = cfg.MAX_WAVE_TIME
self.__max_learn_rate_decay_time = cfg.MAX_LEARN_RATE_DECAY_TIME
self.__weights_dir = cfg.WEIGHTS_DIR
self.__weights_file = cfg.WEIGHTS_FILE
self.__log_dir = os.path.join(cfg.LOG_DIR, 'train')
self.__moving_ave_decay = cfg.MOVING_AVE_DECAY
self.__save_iter = cfg.SAVE_ITER
self.__max_bbox_per_scale = cfg.MAX_BBOX_PER_SCALE
self.__train_data = Data('train')
self.__test_data = Data('test')
with tf.name_scope('input'):
self.__input_data = tf.placeholder(dtype=tf.float32, name='input_data')
self.__label_sbbox = tf.placeholder(dtype=tf.float32, name='label_sbbox')
self.__label_mbbox = tf.placeholder(dtype=tf.float32, name='label_mbbox')
self.__label_lbbox = tf.placeholder(dtype=tf.float32, name='label_lbbox')
self.__sbboxes = tf.placeholder(dtype=tf.float32, name='sbboxes')
self.__mbboxes = tf.placeholder(dtype=tf.float32, name='mbboxes')
self.__lbboxes = tf.placeholder(dtype=tf.float32, name='lbboxes')
self.__training = tf.placeholder(dtype=tf.bool, name='training')
self.__yolo = YOLO_V3(self.__training)
self.__conv_sbbox, self.__conv_mbbox, self.__conv_lbbox, \
self.__pred_sbbox, self.__pred_mbbox, self.__pred_lbbox = self.__yolo.build_nework(self.__input_data)
self.__net_var = tf.global_variables()
print 'Load weights:'
for var in self.__net_var:
print var.op.name
self.__loss = self.__yolo.loss(self.__conv_sbbox, self.__conv_mbbox, self.__conv_lbbox,
self.__pred_sbbox, self.__pred_mbbox, self.__pred_lbbox,
self.__label_sbbox, self.__label_mbbox, self.__label_lbbox,
self.__sbboxes, self.__mbboxes, self.__lbboxes)
with tf.name_scope('learn'):
self.__learn_rate = tf.Variable(self.__learn_rate_init, trainable=False, name='learn_rate_init')
moving_ave = tf.train.ExponentialMovingAverage(self.__moving_ave_decay).apply(tf.trainable_variables())
self.__trainable_var_list = []
for var in tf.trainable_variables():
var_name = var.op.name
var_name_mess = str(var_name).split('/')
if var_name_mess[0] in ['conv_sbbox', 'conv_mbbox', 'conv_lbbox']:
self.__trainable_var_list.append(var)
optimize0 = tf.train.AdamOptimizer(self.__learn_rate).\
minimize(self.__loss, var_list=self.__trainable_var_list)
with tf.control_dependencies([optimize0]):
with tf.control_dependencies([moving_ave]):
self.__train_op_with_frozen_variables = tf.no_op()
optimize1 = tf.train.AdamOptimizer(self.__learn_rate).\
minimize(self.__loss, var_list=tf.trainable_variables())
with tf.control_dependencies([optimize1]):
with tf.control_dependencies([moving_ave]):
self.__train_op_with_all_variables = tf.no_op()
self.__train_op = self.__train_op_with_frozen_variables
print 'Default trian step0 is freeze the weight of darknet'
for var in self.__trainable_var_list:
print '\t' + str(var.op.name).ljust(50) + str(var.shape)
with tf.name_scope('load_save'):
self.__load = tf.train.Saver(self.__net_var)
self.__save = tf.train.Saver(tf.global_variables(), max_to_keep=50)
with tf.name_scope('summary'):
tf.summary.scalar('loss', self.__loss)
self.__summary_op = tf.summary.merge_all()
self.__summary_writer = tf.summary.FileWriter(self.__log_dir)
self.__summary_writer.add_graph(tf.get_default_graph())
self.__sess = tf.Session(config=tf.ConfigProto(allow_soft_placement=True))
if (var_name_mess[-1] not in ['weights', 'gamma', 'beta', 'moving_mean', 'moving_variance']) or \
(var_name_mess[1] == 'yolo-v3' and (var_name_mess[-2] in preserve_org_names)):
continue
org_weights_mess.append([var_name, var_shape])
print str(var_name).ljust(50), var_shape
print
tf.reset_default_graph()
cur_weights_mess = []
tf.Graph().as_default()
with tf.name_scope('input'):
input_data = tf.placeholder(dtype=tf.float32,
shape=(1, 416, 416, 3),
name='input_data')
training = tf.placeholder(dtype=tf.bool, name='training')
YOLO_V3(True).build_nework(input_data)
for var in tf.global_variables():
var_name = var.op.name
var_name_mess = str(var_name).split('/')
var_shape = var.shape
if var_name_mess[0] in preserve_cur_names:
continue
cur_weights_mess.append([var_name, var_shape])
print str(var_name).ljust(50), var_shape
org_weights_num = len(org_weights_mess)
cur_weights_num = len(cur_weights_mess)
if cur_weights_num != org_weights_num:
raise RuntimeError
print 'Number of weights that will rename:\t%d' % cur_weights_num
def __init__(self):
self.__anchor_per_scale = cfg.ANCHOR_PER_SCALE
self.__classes = cfg.CLASSES
self.__num_classes = len(self.__classes)
self.__learn_rate_init = cfg.LEARN_RATE_INIT
self.__learn_rate_end = cfg.LEARN_RATE_END
self.__max_periods = cfg.MAX_PERIODS
self.__periods_for_step0 = cfg.PERIODS_FOR_STEP0
self.__warmup_periods = cfg.WARMUP_PERIODS
self.__weights_dir = cfg.WEIGHTS_DIR
self.__weights_file = cfg.WEIGHTS_FILE
self.__time = time.strftime('%Y-%m-%d-%H-%M-%S',
time.localtime(time.time()))
self.__log_dir = os.path.join(cfg.LOG_DIR, 'train', self.__time)
self.__moving_ave_decay = cfg.MOVING_AVE_DECAY
self.__max_bbox_per_scale = cfg.MAX_BBOX_PER_SCALE
self.__train_data = Data('train')
self.__test_data = Data('test')
self.__steps_per_period = len(self.__train_data)
with tf.name_scope('input'):
self.__input_data = tf.placeholder(dtype=tf.float32,
name='input_data')
self.__label_sbbox = tf.placeholder(dtype=tf.float32,
name='label_sbbox')
self.__label_mbbox = tf.placeholder(dtype=tf.float32,
name='label_mbbox')
self.__label_lbbox = tf.placeholder(dtype=tf.float32,
name='label_lbbox')
self.__sbboxes = tf.placeholder(dtype=tf.float32, name='sbboxes')
self.__mbboxes = tf.placeholder(dtype=tf.float32, name='mbboxes')
self.__lbboxes = tf.placeholder(dtype=tf.float32, name='lbboxes')
self.__training = tf.placeholder(dtype=tf.bool, name='training')
self.__yolo = YOLO_V3(self.__training)
self.__conv_sbbox, self.__conv_mbbox, self.__conv_lbbox, \
self.__pred_sbbox, self.__pred_mbbox, self.__pred_lbbox = self.__yolo.build_nework(self.__input_data)
self.__net_var = tf.global_variables()
logging.info('Load weights:')
for var in self.__net_var:
logging.info(var.op.name)
self.__loss = self.__yolo.loss(self.__conv_sbbox, self.__conv_mbbox,
self.__conv_lbbox, self.__pred_sbbox,
self.__pred_mbbox, self.__pred_lbbox,
self.__label_sbbox, self.__label_mbbox,
self.__label_lbbox, self.__sbboxes,
self.__mbboxes, self.__lbboxes)
with tf.name_scope('optimize'):
with tf.name_scope('learn_rate'):
self.__global_step = tf.Variable(1.0,
dtype=tf.float64,
trainable=False,
name='global_step')
warmup_steps = tf.constant(self.__warmup_periods *
self.__steps_per_period,
dtype=tf.float64,
name='warmup_steps')
train_steps = tf.constant(self.__max_periods *
self.__steps_per_period,
dtype=tf.float64,
name='train_steps')
self.__learn_rate = tf.cond(
pred=self.__global_step < warmup_steps,
true_fn=lambda: self.__global_step / warmup_steps * self.
__learn_rate_init,
false_fn=lambda: self.__learn_rate_end + 0.5 *
(self.__learn_rate_init - self.__learn_rate_end) *
(1 + tf.cos((self.__global_step - warmup_steps) /
(train_steps - warmup_steps) * np.pi)))
global_step_update = tf.assign_add(self.__global_step, 1.0)
moving_ave = tf.train.ExponentialMovingAverage(
self.__moving_ave_decay).apply(tf.trainable_variables())
self.__trainable_var_list = []
for var in tf.trainable_variables():
var_name = var.op.name
var_name_mess = str(var_name).split('/')
if var_name_mess[0] in [
'conv_sbbox', 'conv_mbbox', 'conv_lbbox'
]:
self.__trainable_var_list.append(var)
optimize0 = tf.train.AdamOptimizer(self.__learn_rate).\
minimize(self.__loss, var_list=self.__trainable_var_list)
with tf.control_dependencies([optimize0, global_step_update]):
with tf.control_dependencies([moving_ave]):
self.__train_op_with_frozen_variables = tf.no_op()
optimize1 = tf.train.AdamOptimizer(self.__learn_rate).\
minimize(self.__loss, var_list=tf.trainable_variables())
with tf.control_dependencies([optimize1, global_step_update]):
with tf.control_dependencies([moving_ave]):
self.__train_op_with_all_variables = tf.no_op()
self.__train_op = self.__train_op_with_frozen_variables
logging.info('Default trian step0 is freeze the weight of darknet')
for var in self.__trainable_var_list:
logging.info('\t' + str(var.op.name).ljust(50) +
str(var.shape))
with tf.name_scope('load_save'):
self.__load = tf.train.Saver(self.__net_var)
self.__save = tf.train.Saver(tf.global_variables(), max_to_keep=50)
with tf.name_scope('summary'):
self.__loss_ave = tf.Variable(0, dtype=tf.float32, trainable=False)
tf.summary.scalar('loss_ave', self.__loss_ave)
tf.summary.scalar('learn_rate', self.__learn_rate)
self.__summary_op = tf.summary.merge_all()
self.__summary_writer = tf.summary.FileWriter(self.__log_dir)
self.__summary_writer.add_graph(tf.get_default_graph())
self.__sess = tf.Session(config=tf.ConfigProto(
allow_soft_placement=True))
def __init__(self):
self.__frozen = cfg.FROZEN
self.__classes = cfg.CLASSES
self.__learn_rate_init = cfg.LEARN_RATE_INIT
self.__max_epochs = cfg.MAX_PERIODS
self.__lr_decay_epoch = cfg.lr_decay_epoch
self.__weights_dir = cfg.WEIGHTS_DIR
self.__weights_file = cfg.WEIGHTS_FILE
self.__time = time.strftime('%Y-%m-%d-%H-%M-%S',
time.localtime(time.time()))
self.__log_dir = os.path.join(cfg.LOG_DIR, 'train', self.__time)
self.__moving_ave_decay = cfg.MOVING_AVE_DECAY
self.__save_iter = cfg.SAVE_ITER
self.__train_data = Data('train')
self.__test_data = Data('test')
with tf.name_scope('input'):
self.__input_data = tf.placeholder(dtype=tf.float32,
name='input_data')
self.__label_sbbox = tf.placeholder(dtype=tf.float32,
name='label_sbbox')
self.__label_mbbox = tf.placeholder(dtype=tf.float32,
name='label_mbbox')
self.__label_lbbox = tf.placeholder(dtype=tf.float32,
name='label_lbbox')
self.__sbboxes = tf.placeholder(dtype=tf.float32, name='sbboxes')
self.__mbboxes = tf.placeholder(dtype=tf.float32, name='mbboxes')
self.__lbboxes = tf.placeholder(dtype=tf.float32, name='lbboxes')
self.__training = True
self.__yolo = YOLO_V3(self.__training)
self.__conv_sbbox, self.__conv_mbbox, self.__conv_lbbox, \
self.__pred_sbbox, self.__pred_mbbox, self.__pred_lbbox = self.__yolo.build_nework(self.__input_data)
self.__net_var = tf.global_variables()
logging.info('Load weights:')
for var in self.__net_var:
logging.info(var.op.name)
self.__loss = self.__yolo.loss(self.__conv_sbbox, self.__conv_mbbox,
self.__conv_lbbox, self.__pred_sbbox,
self.__pred_mbbox, self.__pred_lbbox,
self.__label_sbbox, self.__label_mbbox,
self.__label_lbbox, self.__sbboxes,
self.__mbboxes, self.__lbboxes)
with tf.name_scope('learn'):
self.__learn_rate = tf.Variable(self.__learn_rate_init,
trainable=False,
name='learn_rate_init')
moving_ave = tf.train.ExponentialMovingAverage(
self.__moving_ave_decay).apply(tf.trainable_variables())
self.__trainable_var_list = []
for var in tf.trainable_variables():
var_name = var.op.name
var_name_mess = str(var_name).split('/')
if var_name_mess[0] in [
'conv_sbbox', 'conv_mbbox', 'conv_lbbox'
]:
self.__trainable_var_list.append(var)
optimize0 = tf.train.AdamOptimizer(self.__learn_rate).\
minimize(self.__loss, var_list=self.__trainable_var_list)
with tf.control_dependencies([optimize0]):
with tf.control_dependencies([moving_ave]):
self.__train_op_with_frozen_variables = tf.no_op()
optimize1 = tf.train.AdamOptimizer(self.__learn_rate).\
minimize(self.__loss, var_list=tf.trainable_variables())
with tf.control_dependencies([optimize1]):
with tf.control_dependencies([moving_ave]):
self.__train_op_with_all_variables = tf.no_op()
if self.__frozen:
self.__train_op = self.__train_op_with_frozen_variables
logging.info('freeze the weight of darknet')
print('freeze the weight of darknet')
else:
self.__train_op = self.__train_op_with_all_variables
logging.info("train all variables")
print("train all variables")
for var in self.__trainable_var_list:
logging.info("++++++ trainable variables list: ++++++" + "\n")
logging.info('\t' + str(var.op.name).ljust(50) +
str(var.shape))
with tf.name_scope('load_save'):
self.__load = tf.train.Saver(self.__net_var)
self.__save = tf.train.Saver(tf.global_variables(), max_to_keep=50)
with tf.name_scope('summary'):
tf.summary.scalar('loss', self.__loss)
self.__summary_op = tf.summary.merge_all()
self.__summary_writer = tf.summary.FileWriter(self.__log_dir)
self.__summary_writer.add_graph(tf.get_default_graph())
self.__sess = tf.Session(config=tf.ConfigProto(
allow_soft_placement=True))