-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrainroynet.py
176 lines (144 loc) · 6.55 KB
/
trainroynet.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
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
#import modules
import numpy as np
import tensorflow as tf
import cv2
import time
import sys
sys.path.insert(0, './dataset/')
import importBakeryDataCSV
from tensorflow.python.framework import ops
#import dataset
data = importBakeryDataCSV.BakeryDataset()
total_images = data.total_images
print "Total images in dataset: %i" % total_images
#constants
disp_console = True
alpha = 0.1
grid_size = 7
classes = ["bananabread", "cinnamonroll", "croissant", "hotcross"]
n_classes = len(classes)
#define layers
def conv_layer(idx,inputs,filters,size,stride):
channels = inputs.get_shape()[3]
with tf.variable_scope("conv"):
#weight = tf.get_variable('w',[size,size,int(channels),filters])
weight = tf.Variable(tf.truncated_normal([size,size,int(channels),filters], stddev=0.1),name='w')
biases = tf.Variable(tf.constant(0.1, shape=[filters]),name='b')
pad_size = size//2
pad_mat = np.array([[0,0],[pad_size,pad_size],[pad_size,pad_size],[0,0]])
inputs_pad = tf.pad(inputs,pad_mat)
conv = tf.nn.conv2d(inputs_pad, weight, strides=[1, stride, stride, 1], padding='VALID',name=str(idx)+'_conv')
conv_biased = tf.add(conv,biases,name=str(idx)+'_conv_biased')
if disp_console : print ' Layer %d : Type = Conv, Size = %d * %d, Stride = %d, Filters = %d, Input channels = %d' % (idx,size,size,stride,filters,int(channels))
#return tf.maximum(alpha*conv_biased,conv_biased,name=str(idx)+'_leaky_relu')
return tf.maximum(conv_biased,conv_biased,name=str(idx)+'_leaky_relu')
def pooling_layer(idx,inputs,size,stride):
if disp_console : print ' Layer %d : Type = Pool, Size = %d * %d, Stride = %d' % (idx,size,size,stride)
return tf.nn.max_pool(inputs, ksize=[1, size, size, 1],strides=[1, stride, stride, 1], padding='SAME',name=str(idx)+'_pool')
def fc_layer(idx,inputs,hiddens, flat = False,linear = False):
input_shape = inputs.get_shape().as_list()
if flat:
dim = input_shape[1]*input_shape[2]*input_shape[3]
inputs_transposed = tf.transpose(inputs,(0,3,1,2))
inputs_processed = tf.reshape(inputs_transposed, [-1,dim])
else:
dim = input_shape[1]
inputs_processed = inputs
with tf.variable_scope("fc"):
weight = tf.Variable(tf.truncated_normal([dim,hiddens], stddev=0.1),name='w')
biases = tf.Variable(tf.constant(0.1, shape=[hiddens]),name='b')
if disp_console : print ' Layer %d : Type = Full, Hidden = %d, Input dimension = %d, Flat = %d, Activation = %d' % (idx,hiddens,int(dim),int(flat),1-int(linear))
if linear : return tf.add(tf.matmul(inputs_processed,weight),biases,name=str(idx)+'_fc')
ip = tf.add(tf.matmul(inputs_processed,weight),biases)
return tf.maximum(alpha*ip,ip,name=str(idx)+'_fc')
#build layers
print "Building graph..."
x = tf.placeholder(tf.float32,[None,448,448,3])
y = tf.placeholder(tf.float32, [None, n_classes])
conv_1 = conv_layer(1,x,16,3,1)
pool_2 = pooling_layer(2,conv_1,2,2)
conv_3 = conv_layer(3,pool_2,32,3,1)
pool_4 = pooling_layer(4,conv_3,2,2)
conv_5 = conv_layer(5,pool_4,64,3,1)
pool_6 = pooling_layer(6,conv_5,2,2)
conv_7 = conv_layer(7,pool_6,128,3,1)
pool_8 = pooling_layer(8,conv_7,2,2)
conv_9 = conv_layer(9,pool_8,256,3,1)
pool_10 = pooling_layer(10,conv_9,2,2)
conv_11 = conv_layer(11,pool_10,512,3,1)
pool_12 = pooling_layer(12,conv_11,2,2)
conv_13 = conv_layer(13,pool_12,1024,3,1)
conv_14 = conv_layer(14,conv_13,1024,3,1)
conv_15 = conv_layer(15,conv_14,1024,3,1)
fc_16 = fc_layer(16,conv_15,256,flat=True,linear=False)
fc_17 = fc_layer(17,fc_16,4096,flat=False,linear=False)
fc_19 = fc_layer(19,fc_17,4,flat=False,linear=False)
varlist = []
finallayer = []
print ""
for v in tf.all_variables():
varlist.append(v)
if v.name == 'fc_2/w:0' or v.name =='fc_2/b:0' or v.name == 'fc_1/w:0' or v.name == 'fc_1/b:0' or v.name == 'fc/w:0' or v.name == 'fc/b:0' or v.name == 'conv_15/w:0' or v.name == 'conv_15/b:0':
finallayer.append(v)
#Start session
NUM_THREADS = 8
cfg = tf.ConfigProto(inter_op_parallelism_threads=NUM_THREADS,
intra_op_parallelism_threads=NUM_THREADS,
log_device_placement=True)
sess = tf.Session(config=cfg)
#Hyperparameter placeholder
learning_rate = tf.placeholder(tf.float32, shape=[])
#loss and gradient descent functions
loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(fc_19, y))
#optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(loss,var_list=finallayer)
optimizer = tf.train.GradientDescentOptimizer(learning_rate=learning_rate).minimize(loss)
# Evaluate model
correct_pred = tf.equal(tf.argmax(fc_19,1), tf.argmax(y,1))
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
#initialize all variables
sess.run(tf.initialize_all_variables())
#restore select variables from checkpoint
saver = tf.train.Saver(varlist)
saver.restore(sess,"tmp/saved/modelRoy7.ckpt")
print '/nWeights succesfully loaded from checkpoint'
#tensorboard
tf.scalar_summary("cross_entropy", loss)
tf.scalar_summary("training_accuracy", accuracy)
tf.image_summary('layer3', sess.run(varlist[0])[:,:,:,0:3])
summary_op = tf.merge_all_summaries()
summary_writer = tf.train.SummaryWriter('./tmp/roynet_log', graph_def=sess.graph_def)
#set learning rate
lr = 0.000000001
training_iters = 20000
batch_size = 40
display_step = 1
try:
while True:
for i in range(training_iters):
epoch = (batch_size*i)/float(total_images)
batch_x , batch_y = data.pickSample(batch_size)
if i % display_step == 0:
# Calculate batch accuracy and loss
summary_str, _, acc, batch_loss = sess.run([summary_op,optimizer, accuracy,loss], feed_dict={x: batch_x, y: batch_y,learning_rate: lr})
print "Step " + str(i) + ", Epoch "+ str(epoch)+ ", Minibatch Loss= " + str(batch_loss) + ", Training Accuracy= " + str(acc)
if batch_loss != batch_loss:
#check for loss = NaN
break
else:
summary_str, _ = sess.run([summary_op,optimizer], feed_dict={x: batch_x, y: batch_y,learning_rate: lr})
#save at specific steps:
if i % 100 == 0:
saver = tf.train.Saver(varlist)
save_path = saver.save(sess, "./tmp/modelRoy.ckpt")
print("Model saved in file: %s" % save_path)
#saving tensorboard summary
if i%10 ==0 :
summary_writer.add_summary(summary_str, i)
save_path = saver.save(sess, "./tmp/modelRoy.ckpt")
sess.close()
print 'done'
except KeyboardInterrupt:
print "Stopping training... saving checkpoint first.."
save_path = saver.save(sess, "./tmp/modelRoy.ckpt")
sess.close()
print 'done'