def Train(graph, Batch_size, Iteration):
filenames = [("NL测试数据/train_%d" % m + ".tfrecords") for m in range(2, 7)]
print(filenames)
filenames_cross = ["NL测试数据/train_7.tfrecords"]
filename_queue = tf.train.string_input_producer(filenames)
# Tensorflow需要队列流输入数据
reader = tf.TFRecordReader()
_, serialized_example = reader.read(filename_queue)
# tf.parse_single_example为解析器
features = tf.parse_single_example(serialized_example,
features={
"Mat":
tf.FixedLenFeature([], tf.string),
"label":
tf.FixedLenFeature([], tf.int64),
})
Mat_16 = tf.decode_raw(features["Mat"], tf.float64)
Mat_16 = tf.reshape(Mat_16, [16, 16, 1])
Label_16 = tf.cast(features["label"], tf.float32)
# Mat_batch,Label64_batch,Label32_batch,Label16_Batch = tf.train.batch\
# ([Mat_64, Label_64,Label_32,Label_16],batch_size= Batch_size,capacity=300,num_threads= 1)
# tf.train.shuffle_batch（）这个函数对读取到的数据进行了batch处理，这样更有利于后续的训练。
Mat_batch, Label16_batch = tf.train.shuffle_batch \
([Mat_16, Label_16], batch_size=Batch_size, capacity=3 * Batch_size, min_after_dequeue=5,
num_threads=32)
print("\n--------- finish reading train data-----------")
filename_queue_cross = tf.train.string_input_producer(filenames_cross)
reader_cross = tf.TFRecordReader()
_, serialized_example_cross = reader_cross.read(filename_queue_cross)
features_cross = tf.parse_single_example(serialized_example_cross,
features={
"Mat":
tf.FixedLenFeature([],
tf.string),
"label":
tf.FixedLenFeature([],
tf.int64),
})
Mat_16_cross = tf.decode_raw(features_cross["Mat"], tf.float64)
Mat_16_cross = tf.reshape(Mat_16_cross, [16, 16, 1])
Label_16_cross = tf.cast(features_cross["label"], tf.float32)
Mat_batch_cross, Label16_batch_cross = tf.train.shuffle_batch \
([Mat_16_cross, Label_16_cross], batch_size=Batch_size,
capacity=Batch_size * 3, min_after_dequeue=5,
num_threads=32)
# saver = tf.train.Saver()
# print("Mat_Batttt",Mat_Batch)
with tf.Session() as sess:
# 开启协调器
coord = tf.train.Coordinator()
# 使用start_queue_runners 启动队列填充
# 一定要开启队列，要不然是空的
threads = tf.train.start_queue_runners(sess, coord)
initglo = tf.global_variables_initializer()
initloc = tf.local_variables_initializer()
sess.run(initglo)
sess.run(initloc)
tf.local_variables_initializer().run()
epoch = 1
learning_rate = 0.01
for j in range(epoch):
if j % 3 == 0:
learning_rate = learning_rate * 0.1
print(learning_rate)
acc = 0
loss = 0
acc_cross = 0
loss_cross = 0
for i in range(Iteration):
# 这句话的意思是从队列里拿出指定批次的数据
is_training = True
mat_16, label_16 = sess.run([Mat_batch, Label16_batch])
mat_16 = mat_16.reshape(Batch_size, 16, 16, 1)
# print("Mat_64 is ",Mat_64)
# print("Label_64 is ",Label_64)
mat_16 = mat_16.astype('float32')
label_16 = label_16.astype("float32")
label_16 = label_16.reshape(Batch_size, 1)
for b in range(Batch_size):
mat_16[b] = LoadData.Normalize(mat_16[b])
# LOSS_64, acc_rate64, Pre_lable_64, _ = sess.run(
# [graph["LOSS"], graph["acc"], graph["Pre"], graph["Train"]],
# feed_dict={
# graph["inputs_Mat"]: mat_64,
# graph["ys_label"]: label_64,
# graph['is_training']: is_training,
# graph["is_64"]: 1,
# graph["is_32"]: 0,
# graph["kp"]: 0.5,
# graph["ln"]:learning_rate,
# })
# LOSS_32, acc_rate32, Pre_lable_32, _ = sess.run(
# [graph["LOSS"], graph["acc"], graph["Pre"], graph["Train"]],
# feed_dict={
# graph["inputs_Mat"]: mat_32,
# graph["ys_label"]: label_32,
# graph['is_training']: is_training,
# graph["is_64"]: 0,
# graph["is_32"]: 1,
# graph["kp"]: 0.5,
# graph["ln"]: learning_rate,
# })
# print(label_16)
LOSS_16, acc_rate16, Pre_lable_16, _ = sess.run(
[
graph["LOSS"], graph["acc"], graph["Pre"],
graph["Train"]
],
feed_dict={
graph["inputs_Mat"]: mat_16,
graph["ys_label"]: label_16,
graph['is_training']: is_training,
graph["is_64"]: 0,
graph["is_32"]: 0,
graph["kp"]: 0.5,
graph["ln"]: learning_rate,
})
if i % 100 == 0:
is_training = False
print("\n------------- Iteration %d --------------" % (i))
# # print("LOSS64 is ", LOSS_64)
# # print("LOSS32 is ", LOSS_32)
# print("LOSS16 is ", LOSS_16)
# # print("acc_rate64 is : ", acc_rate64)
# # print("acc_rate32 is : ", acc_rate32)
# print("acc_rate16 is : ", acc_rate16)
# # 把输出写入到文件里
# stdout_backup = sys.stdout
# log_file = open(YUV + "_RunDetail.log", "a")
# sys.stdout = log_file
# print("--------------- Iteration %d ---------------" % (i))
#
# print("LOSS64 is ", LOSS_64)
# print("LOSS32 is ", LOSS_32)
# print("LOSS16 is ", LOSS_16)
# print("acc_rate64 is : ", acc_rate64)
# print("acc_rate32 is : ", acc_rate32)
# print("acc_rate16 is : ", acc_rate16)
mat_cross16, label_cross16 = sess.run(
[Mat_batch_cross, Label16_batch_cross])
mat_cross16 = mat_cross16.reshape(Batch_size, 16, 16, 1)
mat_cross16 = mat_cross16.astype("float32")
for b in range(Batch_size):
mat_cross16[b] = LoadData.Normalize(mat_cross16[b])
label_cross16 = label_cross16.astype("float32")
label_cross16 = label_cross16.reshape(Batch_size, 1)
LOSS_ceoss16, acc_rate_cross16, Pre_lable_cross16 = sess.run(
[graph["LOSS"], graph["acc"], graph["Pre"]],
feed_dict={
graph["inputs_Mat"]: mat_cross16,
graph["ys_label"]: label_cross16,
graph['is_training']: is_training,
graph["is_64"]: 0,
graph["is_32"]: 0,
graph["kp"]: 1.0,
})
acc_cross = acc_cross + acc_rate_cross16
loss_cross = loss_cross + LOSS_ceoss16
loss = loss + LOSS_16
acc = acc + acc_rate16
print("*************train*************")
print("acc:" + str(acc / 60))
print("loss:" + str(loss / 60))
print("*************cross*************")
print("acc:" + str(acc_cross / 60))
print("loss:" + str(loss_cross / 60))
# sum16 = 0
# a16 = 0
# for e in range(Batch_size):
# if label_cross16[e][0] == 1:
# sum16 += 1
# if Pre_lable_cross16[e][0] == 1:
# a16 += 1
# if sum16 != 0:
# print("reCall 16 is : ", a16 / sum16)
# sum16 = 0
# a16 = 0
# for e in range(Batch_size):
# if label_cross16[e][0] == 0:
# sum16 += 1
# if Pre_lable_cross16[e][0] == 0:
# a16 += 1
# if sum16 != 0:
# print("Call 16 is : ", a16 / sum16)
if j == 0:
print("save the model finally:" + str(i))
var_list = [
var for var in tf.global_variables()
if "moving" in var.name
]
var_list += tf.trainable_variables()
saver = tf.train.Saver(var_list=var_list, max_to_keep=20)
Write_List = [
'cond/Merge', 'is_train', 'm_64', 'm_32', 'kp', 'y', 'acc',
'output'
]
constant_graph = graph_util.convert_variables_to_constants \
(sess, sess.graph_def, Write_List)
with tf.gfile.FastGFile(pb_file_path, mode='wb') as f:
f.write(constant_graph.SerializeToString())
saver.save(sess, "model/Demo-model")
# saver.save(sess, "./model/Demo-model-one")
print("save the cross model")
coord.request_stop()
print("program ending")
coord.join(threads)
