-
Notifications
You must be signed in to change notification settings - Fork 0
/
resfasterattention.py
125 lines (95 loc) · 6.06 KB
/
resfasterattention.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
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Fri Aug 31 02:13:40 2018
@author: ck807
"""
import os
import glob
import numpy as np
#import cv2
import tensorflow as tf
from keras.models import Model
import matplotlib.pyplot as plt
import keras
from keras import layers
from keras.layers.convolutional import Conv2D, UpSampling2D, SeparableConv2D
from keras.layers.pooling import MaxPooling2D, GlobalAveragePooling2D, AveragePooling2D
from keras.layers import Input, Dropout, Dense, BatchNormalization, Flatten
from keras.regularizers import l2
from keras.optimizers import SGD, Adam
from keras.metrics import mean_absolute_error
from keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau
from keras.preprocessing.image import img_to_array
from frozenresidualblockwithlayernames import ResidualR
from frozenresidualblockwithlayernames import initial_conv_block1, Residual2, Residual3, Residual4, Residual5, Residual6
from frozenresidualblockwithlayernames import Residual7, Residual8, Residual9, Residual10, Residual11, Residual12
from frozenresidualblockwithlayernames import Residual13, Residual14, Residual15, Residual16, Residual17, Residual18, Residual19
from se import squeeze_excite_block
from layers import initial_conv_block, bottleneck_block_with_se
from resnet import _conv_bn_relu, _residual_block, basic_block, _bn_relu
import keras.backend as K
CHANNEL_AXIS = 3
print('Loading the data..')
trainData = np.load('trainData.npy')
trainMask = np.load('trainMask.npy')
valData = np.load('valData.npy')
valMask = np.load('valMask.npy')
print('Building and compiling the model..')
smooth=1
def dice_coef(y_true, y_pred, smooth=1):
intersection = K.sum(y_true * y_pred, axis=[1,2,3])
union = K.sum(y_true, axis=[1,2,3]) + K.sum(y_pred, axis=[1,2,3])
return K.mean( (2. * intersection + smooth) / (union + smooth), axis=0)
def dice_coef_loss(y_true, y_pred):
return 1-dice_coef(y_true, y_pred)
def jaccard_coef(y_true, y_pred):
intersection = K.sum(y_true * y_pred, axis=[0, -1, -2])
sum_ = K.sum(y_true + y_pred, axis=[0, -1, -2])
jac = (intersection + smooth) / (sum_ - intersection + smooth)
return K.mean(jac)
with tf.device('/device:GPU:0'):
input = Input((192, 192, 3), name='Input')
conv1 = _conv_bn_relu(filters=32, kernel_size=(7, 7), strides=(1, 1))(input)
conv1 = _residual_block(basic_block, filters=32, repetitions=1, is_first_layer=True)(conv1)
pool1 = MaxPooling2D(pool_size=(2, 2), name='MaxPool1')(conv1)
conv2 = _residual_block(basic_block, filters=64, repetitions=1, is_first_layer=True)(pool1)
conv2 = _residual_block(basic_block, filters=64, repetitions=1, is_first_layer=True)(conv2)
pool2 = MaxPooling2D(pool_size=(2, 2), name='MaxPool2')(conv2)
conv3 = _residual_block(basic_block, filters=128, repetitions=1, is_first_layer=True)(pool2)
conv3 = _residual_block(basic_block, filters=128, repetitions=1, is_first_layer=True)(conv3)
pool3 = MaxPooling2D(pool_size=(2, 2), name='MaxPool3')(conv3)
conv4 = _residual_block(basic_block, filters=256, repetitions=1, is_first_layer=True)(pool3)
conv4 = _residual_block(basic_block, filters=256, repetitions=1, is_first_layer=True)(conv4)
drop4 = Dropout(0.2, name='Dropout1')(conv4)
pool4 = MaxPooling2D(pool_size=(2, 2), name='MaxPool4')(drop4)
conv5 = _residual_block(basic_block, filters=256, repetitions=1, is_first_layer=True)(pool4)
conv5 = _residual_block(basic_block, filters=256, repetitions=1, is_first_layer=True)(conv5)
drop5 = Dropout(0.2, name='Dropout2')(conv5)
up6 = Conv2D(128, 2, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal', name='UpConv1')(UpSampling2D(size = (2,2), name='Up1')(drop5))
merge6 = keras.layers.Concatenate(name='Concat1')([drop4,up6])
conv6 = _residual_block(basic_block, filters=128, repetitions=1, is_first_layer=True)(merge6)
conv6_1 = _residual_block(basic_block, filters=128, repetitions=1, is_first_layer=True)(conv6)
up7 = Conv2D(64, 2, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal', name='UpConv2')(UpSampling2D(size = (2,2), name='Up2')(conv6_1))
merge7 = keras.layers.Concatenate(name='Concat2')([conv3,up7])
conv7 = _residual_block(basic_block, filters=64, repetitions=1, is_first_layer=True)(merge7)
conv7_1 = _residual_block(basic_block, filters=64, repetitions=1, is_first_layer=True)(conv7)
up8 = Conv2D(32, 2, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal', name='UpConv3')(UpSampling2D(size = (2,2), name='Up3')(conv7_1))
merge8 = keras.layers.Concatenate(name='Concat3')([conv2,up8])
conv8 = _residual_block(basic_block, filters=32, repetitions=1, is_first_layer=True)(merge8)
conv8_1 = _residual_block(basic_block, filters=32, repetitions=1, is_first_layer=True)(conv8)
up9 = Conv2D(16, 2, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal', name='UpConv4')(UpSampling2D(size = (2,2), name='Up4')(conv8_1))
merge9 = keras.layers.Concatenate(name='Concat4')([conv1,up9])
conv9 = _residual_block(basic_block, filters=16, repetitions=1, is_first_layer=True)(merge9)
conv10 = _residual_block(basic_block, filters=4, repetitions=1, is_first_layer=True)(conv9)
conv10 = _residual_block(basic_block, filters=1, repetitions=1, is_first_layer=True)(conv10)
conv11 = Conv2D(1, 1, activation = 'sigmoid', name='Output')(conv10)
model = Model(input, conv11)
model.compile(loss=dice_coef_loss, optimizer=SGD(lr=0.03, momentum=0.9, nesterov=True), metrics=[dice_coef,jaccard_coef])
model.summary()
callbacks = [
EarlyStopping(monitor='val_loss', patience=10, verbose=1, min_delta=0.001),
ModelCheckpoint("resnet_faster_val_loss_ldmk_checkpoint.h5", monitor='val_loss', verbose=1, save_best_only=True, mode='auto'),
ReduceLROnPlateau(monitor='val_loss', patience=5, verbose=1, mode='auto', epsilon=0.001, cooldown=0, min_lr=0.5e-7)
]
history = model.fit(trainData, trainMask, validation_data=(valData, valMask), batch_size=32, epochs=200, verbose=1, shuffle=True, callbacks=callbacks)