-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathtrain_v3.py
executable file
·199 lines (176 loc) · 9.37 KB
/
train_v3.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
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
from keras.preprocessing.image import ImageDataGenerator
from keras.applications.inception_v3 import InceptionV3, preprocess_input
from keras.layers import GlobalAveragePooling2D, Dense
from keras.models import Model
from keras.utils.vis_utils import plot_model
from keras.optimizers import Adagrad
from keras.callbacks import TensorBoard
import os
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
# 仅使用GPU0 GeForce GTX 1080Ti
os.environ["CUDA_VISIBLE_DEVICES"]="0"
# 迁移训练模型保存
save_file_name_TransferTraining = "D:\\Inception-Camelyon17\\modules\\transfer_learning_model.h5"
# 微调模型保存
save_file_name_fileTune = "D:\\Inception-Camelyon17\\modules\\fine_tune_model.h5"
# 训练数据集的路径
train_data_dir = "F:\\ai_data\\camelyon17\\research_data\\train"
# 训练测试数据集的路径
val_data_dir = "F:\\ai_data\\camelyon17\\research_data\\validation"
# 测试数据集的路径
test_data_dir = "F:\\ai_data\\camelyon17\\research_data\\test"
# 类别数目
class_number = 3
# 每一个epoch内的训练次数 default : 4000, 20000, 40000, 80000
# 官方建议的值 = 训练用图像总数量 / batch_size
# 本程序：训练用图像总数量=400000， batch_size=64, 所以，step_per_epoch = 6250
step_per_epoch = 6250
# epoch数目
epoch = 10
# 运行测试的间隔点
val_step = 1
# 绘图函数
def plot_training(history):
acc = history.history["acc"]
val_acc = history.history["val_acc"]
loss = history.history["loss"]
val_loss = history.history["val_loss"]
epochs = range(len(acc))
plt.plot(epochs, acc, "r.")
plt.plot(epochs, val_acc, "r")
plt.title("Training and validation accuracy")
plt.figure()
plt.plot(epochs, loss, "r.")
plt.plot(epochs, val_loss, "r-")
plt.title("Training and validation loss")
plt.show()
# 转移训练设置函数
# Transfer Learning：将骨架模型的所有层都设置为不可训练
def setup_to_transfer_learning(model, base_model):
for layer in base_model.layers:
layer.trainable = False
model.compile(optimizer="adam", loss="categorical_crossentropy", metrics=["accuracy"])
# Fine Tune：将骨架模型中的前几层设置为不可训练，后面的所有Inception模块都设置为可训练
def setup_to_fine_tune(model, base_model):
GAP_LAYER = class_number
for layer in base_model.layers[:GAP_LAYER+1]:
layer.trainable = False
for layer in base_model.layers[GAP_LAYER+1:]:
layer.trainable = True
model.compile(optimizer=Adagrad(lr=0.0001), loss="categorical_crossentropy", metrics=["accuracy"])
def main():
# 准备训练数据
Gen_Train_Data = ImageDataGenerator(
preprocessing_function=preprocess_input, # 将原始数据归一化到-1.0~+1.0
rotation_range = 45.0, # 设置随机化的旋转随机
width_shift_range = 0.0, # 设置随机化的平移随机
height_shift_range = 0.0, # 设置随机化的平移垂直随机
shear_range = 0.0, # 逆时针方向的剪切变换角度
zoom_range = 0.0, # 图像缩放比例
horizontal_flip = True, # 进行随机水平旋转
vertical_flip = True, # 进行随机垂直旋转
)
# 准备训练测试数据
Gen_Val_Data = ImageDataGenerator(
preprocessing_function=preprocess_input,
rotation_range = 45.0,
width_shift_range = 0.0,
height_shift_range = 0.0,
shear_range = 0.0,
zoom_range = 0.0,
horizontal_flip = True,
vertical_flip = True
)
# 读取训练数据
train_data_generator = Gen_Train_Data.flow_from_directory(directory = train_data_dir,
target_size = (299, 299),
batch_size = 64)
# 读取训练测试数据
val_data_generator = Gen_Val_Data.flow_from_directory(directory = val_data_dir,
target_size = (299, 299),
batch_size = 64)
# 构建基础模型
# weights 使用 imagenet 模型进行迁移学习
# include_top False表示去掉原始的全连接层，输出一个8X8X2048的张量；True则表示使用原有的全连接层，1000个分类输出
# 一般做迁移训练，去掉顶层，后面街上各种自定义的新层，这已经成为了训练任务的惯用套路
# Inception基础骨架
base_model = InceptionV3(weights='imagenet', include_top=False);
# 输出层先用GlobalAveragePooling2D函数将8 * 8 * 2048的输出转换成1 * 2048的张量
# 后面接了一个1024个节点的全连接层，最后是一个17个节点的输出层，用softmax激活函数
# 增加新的输出层
x = base_model.output
# GlobalAveragePooling2D 将 MxNxC 的张量转换成 1xC 张量，C是通道数
x = GlobalAveragePooling2D()(x)
x = Dense(1024, activation="relu")(x)
predictions = Dense(class_number, activation="softmax")(x)
# 迁移学习模型骨架
model = Model(inputs=base_model.input, outputs=predictions)
# 输出迁移学习框架图
plot_model(model, to_file="model.png", show_shapes = False, show_layer_names = False)
plot_model(model, to_file="model_with_Shapes.png", show_shapes = True, show_layer_names = False)
plot_model(model, to_file="model_with_LayerNames.png", show_shapes = False, show_layer_names = True)
plot_model(model, to_file="model_with_Shape&LayerNames.png", show_shapes = True, show_layer_names = True)
# 学习过程TensorBoard数据准备
tb_log = TensorBoard(log_dir="tmp/log", batch_size=64, histogram_freq=1, write_graph=True, write_images=False, write_grads=True)
# 设置并执行Transfer Learning迁移训练
print("Start to Transfer Learning...")
setup_to_transfer_learning(model, base_model)
history_transferTraining = model.fit_generator(generator = train_data_generator, # 训练数据生成器
steps_per_epoch = step_per_epoch, # 一个epoch中的训练次数，这个训练次数可以理解为训练数据的个数，也即遍历学习
epochs = epoch, # 迭代的轮数
validation_data = val_data_generator, # 训练测试数据生成器
validation_steps = val_step, # 指定训练测试数据集的生成器返回次数
class_weight = "auto")
model.save(save_file_name_TransferTraining)
print("Saved transfer training model : {0}".format(save_file_name_TransferTraining))
# 设置并执行Fine Tune迁移训练
print("Start to Fine Tune...")
setup_to_fine_tune(model, base_model)
history_fineTune = model.fit_generator(generator = train_data_generator,
steps_per_epoch = step_per_epoch,
epochs = epoch,
validation_data = val_data_generator,
validation_steps = val_step,
class_weight = "auto")
model.save(save_file_name_fileTune)
print("Saved fine tune model : {0}".format(save_file_name_fileTune))
plot_training(history_transferTraining)
plot_training(history_fineTune)
#print("Output transfer learning history...")
# 输出迁移训练的精度的分布图
#plt.plot(history_transferTraining.history["acc"], "o-", lable="accuracy")
#plt.plot(history_transferTraining.history["val_acc"], "o-", label="val_acc")
#plt.title("Transfer learning model accuracy")
#plt.xlabel("epoch")
#plt.ylabel("accuracy")
#plt.legend(loc="lower right")
#plt.show()
# 输出迁移训练的损失的分布图
#plt.plot(history_transferTraining.history["loss"], "o-", label="loss")
#plt.plot(history_transferTraining.history["val_loss"], "o-", label="val_loss")
#plt.title("Transfer learning model loss")
#plt.xlabel("epoch")
#plt.ylabel("loss")
#plt.legend(loc="lower right")
#plt.show()
#print("Output fine tune history...")
# 输出迁移训练微调的精度的分布图
#plt.plot(history_ft.history["acc"], "o-", lable="accuracy")
#plt.plot(history_ft.history["val_acc"], "o-", label="val_acc")
#plt.title("Fine time model accuracy")
#plt.xlabel("epoch")
#plt.ylabel("accuracy")
#plt.legend(loc="lower right")
#plt.show()
# 输出迁移训练微调的损失的分布图
#plt.plot(history_ft.history["loss"], "o-", label="loss")
#plt.plot(history_ft.history["val_loss"], "o-", label="val_loss")
#plt.title("Fine tune model loss")
#plt.xlabel("epoch")
#plt.ylabel("loss")
#plt.legend(loc="lower right")
#plt.show()
if __name__ == '__main__':
main()