forked from RuoyuChen10/Image_weighing
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrain_cov_R_restore.py
121 lines (101 loc) · 5.06 KB
/
train_cov_R_restore.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
import tensorflow as tf
import numpy as np
import os
from data_reload import *
# REGULARIZER = 0.01
BATCH_SIZE = 10
def compute_accuracy(v_xs, v_ys):
global prediction
y_pre = sess.run(prediction, feed_dict = {xs: v_xs, keep_prob: 1})
correct_prediction = tf.equal(tf.argmax(y_pre, 1), tf.argmax(v_ys, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
result = sess.run(accuracy, feed_dict = {xs: v_xs, ys: v_ys, keep_prob: 1})
return result
def weight_variable(shape, weight_name):
initial = tf.truncated_normal(shape, stddev=0.1)
return tf.Variable(initial, name = weight_name)
def bias_variable(shape, bias_name):
initial = tf.constant(0.1, shape = shape)
return tf.Variable(initial, name = bias_name)
def conv2d(x, W):
# stride[1, x_movement, y_movement, 1]
# Must have strides[0] = strides[3] =1
return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding="VALID") # padding="SAME"用零填充边界，VALID是不填充
def conv2d_2(x, W):
# stride[1, x_movement, y_movement, 1]
return tf.nn.conv2d(x, W, strides=[1, 2, 2, 1], padding="VALID") # padding="SAME"用零填充边界，VALID是不填充
def max_pool_2x2(x):
return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding="VALID")
def forward(x, keep_prob, sess):
## convl layer ##
x_image = tf.reshape(x, [-1, 324, 80, 1])
# W_conv1 = weight_variable([3, 3, 1, 16], weight_name = "conv1") # kernel 3*3, channel is 3
W_conv1 = sess.graph.get_tensor_by_name("conv1:0")
# b_conv1 = bias_variable([16], bias_name = "cb1")
b_conv1 = sess.graph.get_tensor_by_name("cb1:0")
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1) # output size 322*78*32
h_pool1 = max_pool_2x2(h_conv1) # output size 161*39*32
## conv2 layer ##
#W_conv2 = weight_variable([3, 3, 16, 32], weight_name = "conv2") # kernel 3*3, in size 3, out size 5
W_conv2 = sess.graph.get_tensor_by_name("conv2:0")
# b_conv2 = bias_variable([32], bias_name = "cb2")
b_conv2 = sess.graph.get_tensor_by_name("cb2:0")
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2) # output size 159*37*5
h_pool2 = max_pool_2x2(h_conv2) # output size 79*18*32
## funcl layer ##
# W_fc1 = weight_variable([45504, 1024], weight_name = "fc1")
W_fc1 = sess.graph.get_tensor_by_name("fc1:0")
#b_fc1 = bias_variable([1024], bias_name = "fb1")
b_fc1 = sess.graph.get_tensor_by_name("fb1:0")
# [n_samples,7,7,64]->>[n_samples, 7*7*64]
h_pool2_flat = tf.reshape(h_pool2, [-1, 45504])
h_fc1 = tf.nn.tanh(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
## func2 layer ##
#W_fc2 = weight_variable([1024, 512], weight_name = "fc2")
W_fc2 = sess.graph.get_tensor_by_name("fc2:0")
#b_fc2 = bias_variable([512], bias_name = "fb2")
b_fc2 = sess.graph.get_tensor_by_name("fb2:0")
h_fc2 = tf.nn.relu(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)
h_fc2_drop = tf.nn.dropout(h_fc2, keep_prob)
## func3 layer ##
# W_fc3 = weight_variable([512, 1], weight_name = "fc3")
W_fc3 = sess.graph.get_tensor_by_name("fc3:0")
#b_fc3 = bias_variable([1], bias_name = "fb3")
b_fc3 = sess.graph.get_tensor_by_name("fb3:0")
fc3 = tf.matmul(h_fc2_drop, W_fc3)
y = tf.add(fc3, b_fc3, name = "y")
return y
def backward(datasets, label, test_data, test_label):
saver = tf.train.import_meta_graph('checkpoint/variable-1011000.meta')
# X = tf.placeholder(tf.float32, [None, 324, 80], name = "X")
Y_ = tf.placeholder(tf.float32, [None, 1])
with tf.Session() as sess:
saver.restore(sess, tf.train.latest_checkpoint(r'./checkpoint'))
# keep_prob = tf.placeholder(tf.float32, name="keep_prob")
keep_prob = sess.graph.get_tensor_by_name("keep_prob:0")
X = sess.graph.get_tensor_by_name("X:0")
y = sess.graph.get_tensor_by_name("y:0")
loss_mse = tf.reduce_mean(tf.square(y - Y_))
train_step = tf.train.GradientDescentOptimizer(0.00001).minimize(loss_mse)
global_step = tf.Variable(0, trainable=False)
init_op = tf.global_variables_initializer()
sess.run(init_op)
# 训练模型。
STEPS = 50000000001
for i in range(STEPS):
start = (i*BATCH_SIZE) % 1720
end = start + BATCH_SIZE
sess.run(train_step, feed_dict={X: datasets[start:end], Y_: label[start:end], keep_prob:1})
if i % 200 == 0:
train_loss = sess.run(loss_mse, feed_dict={X: datasets, Y_: label, keep_prob:1})
total_loss = sess.run(loss_mse, feed_dict={X: test_data, Y_: test_label, keep_prob:1})
print("After %d training step(s), loss_mse on train data is %g, loss_mse on val data is %g" % (i, train_loss, total_loss))
if i % 10000 == 0:
saver.save(sess, './checkpoint4/variable', global_step = i)
def main():
tf.reset_default_graph()
datasets, label, test_data, test_label = reload_all_data_R()
backward(datasets, label, test_data, test_label)
if __name__ == '__main__':
main()