-
Notifications
You must be signed in to change notification settings - Fork 101
/
new_train.py
206 lines (165 loc) · 7.42 KB
/
new_train.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
200
201
202
203
204
205
# coding=utf-8
import argparse
import glob
import os
import keras
import tensorflow as tf
from keras.utils import multi_gpu_model
from keras.callbacks import (CSVLogger, EarlyStopping, ModelCheckpoint,
ReduceLROnPlateau)
import keras as K
import data
import Models
from Models import build_model
from utils.utils import *
from metrics import metrics
from losses import LOSS_FACTORY
config = tf.ConfigProto()
config.gpu_options.allow_growth = True
session = tf.Session(config=config)
parser = argparse.ArgumentParser()
parser.add_argument("--model_name", type=str, default="unet")
parser.add_argument("--exp_name", type=str, default='exp1')
parser.add_argument(
"--dataset_name", type=str,
default="bbufdataset") # camvid(32)(720x960), helen_small(11)(512x512), bbufdataset(2)(224x224)
parser.add_argument("--n_classes", type=int, default=2)
parser.add_argument("--epochs", type=int, default=50)
parser.add_argument("--input_height", type=int, default=224)
parser.add_argument("--input_width", type=int, default=224)
parser.add_argument('--validate', type=bool, default=True)
parser.add_argument("--resize_op", type=int, default=1)
parser.add_argument("--train_batch_size", type=int, default=4)
parser.add_argument("--val_batch_size", type=int, default=4)
parser.add_argument("--train_save_path", type=str, default="weights/")
parser.add_argument("--resume", type=str, default="")
parser.add_argument("--optimizer_name", type=str, default="sgd")
parser.add_argument("--image_init", type=str, default="divide")
parser.add_argument("--multi_gpus", type=bool, default=False)
parser.add_argument("--gpu_count", type=int, default=1)
parser.add_argument("--loss", type=str, default='ce')
args = parser.parse_args()
# 再定义一些keras回调函数需要的参数
# 权重保存
train_save_path = os.path.join(args.train_save_path, args.exp_name, args.model_name)
epochs = args.epochs
load_weights = args.resume
mk_if_not_exits(train_save_path)
# patience：没有提升的轮次，即训练过程中最多容忍多少次没有提升
patience = 50
# log_file_path：日志保存的路径
log_file_path = 'weights/' + args.exp_name + '/%s/log.csv' % args.model_name
# 模型参数
model_name = args.model_name
optimizer_name = args.optimizer_name
image_init = args.image_init
multi_gpus = args.multi_gpus
gpu_count = args.gpu_count
# 数据存储位置
data_root = os.path.join("data", args.dataset_name)
train_images = os.path.join(data_root, "train_image")
train_segs = os.path.join(data_root, "train_label")
train_batch_size = args.train_batch_size
validate = args.validate
if validate:
val_images = os.path.join(data_root, "test_image")
val_segs = os.path.join(data_root, "test_label")
val_batch_size = args.val_batch_size
# 数据参数
n_classes = args.n_classes
input_height = args.input_height
input_width = args.input_width
resize_op = args.resize_op
model = build_model(model_name,
n_classes,
input_height=input_height,
input_width=input_width)
print(get_flops(model))
parallel_model = multi_gpu_model(model, gpus=gpu_count)
# 需要保证脚本开头指定的gpu个数和现在要使用的gpu数量相等
# if multi_gpus == True:
# model = multi_gpu_model(model, gpus=2)
# 统计一下训练集/验证集样本数，确定每一个epoch需要训练的iter
images = glob.glob(os.path.join(train_images, "*.jpg")) + \
glob.glob(os.path.join(train_images, "*.png")) + \
glob.glob(os.path.join(train_images, "*.jpeg"))
num_train = len(images)
images = glob.glob(os.path.join(val_images, "*.jpg")) + \
glob.glob(os.path.join(val_images, "*.png")) + \
glob.glob(os.path.join(val_images, "*.jpeg"))
num_val = len(images)
from keras.callbacks import History
from keras.callbacks import ModelCheckpoint
history = History()
# 设置log的存储位置，将网络权值以图片格式保持在tensorboard中显示，设置每一个周期计算一次网络的
tb_cb = keras.callbacks.TensorBoard(log_dir='weights/'+ args.exp_name +'/%s/log' % args.model_name, write_images=1, histogram_freq=0)
# 模型回调函数
early_stop = EarlyStopping('loss', min_delta=0.1, patience=patience, verbose=1)
reduce_lr = ReduceLROnPlateau('loss',
factor=0.01,
patience=int(patience / 2),
verbose=1)
csv_logger = CSVLogger(log_file_path, append=False)
model_names = os.path.join(train_save_path, '%s.{epoch:02d}-{acc:2f}.hdf5' % (
args.model_name))
# model_checkpoint = ModelCheckpoint(model_names,
# monitor='val_iou_score',
# save_best_only=True,
# save_weights_only=True,
# mode='max')
model_checkpoint = ParallelModelCheckpoint(model, filepath=model_names,
monitor='val_iou_score',
save_best_only=True,
save_weights_only=True,
mode='max')
# if multi_gpus == True:
# model_checkpoint = ParallelModelCheckpoint(model, filepath=model_names,
# monitor='val_iou_score',
# save_best_only=True,
# save_weights_only=True,
# mode='max')
call_backs = [model_checkpoint, csv_logger, early_stop, reduce_lr, tb_cb]
loss_func = LOSS_FACTORY[args.loss]
# compile
parallel_model.compile(loss=loss_func,
optimizer=optimizer_name,
metrics=['accuracy', 'iou_score', 'dice_score', 'f1_score', 'f2_score'])
if len(load_weights) > 0:
parallel_model.load_weights(load_weights)
print("Model output shape : ", model.output_shape)
model.summary()
output_height = model.outputHeight
output_width = model.outputWidth
# data generator
train_ge = data.imageSegmentationGenerator(train_images, train_segs,
train_batch_size, n_classes,
input_height, input_width, resize_op,
output_height, output_width,
image_init)
if validate:
val_ge = data.imageSegmentationGenerator(val_images, val_segs,
val_batch_size, n_classes,
input_height, input_width, resize_op,
output_height, output_width,
image_init)
# 开始训练
if not validate:
history = parallel_model.fit_generator(train_ge,
epochs=epochs,
callbacks=call_backs,
steps_per_epoch=int(num_train / train_batch_size),
verbose=1,
shuffle=True,
max_q_size=10,
workers=1)
else:
history = parallel_model.fit_generator(train_ge,
validation_data=val_ge,
epochs=epochs,
callbacks=call_backs,
verbose=1,
steps_per_epoch=int(num_train / train_batch_size),
shuffle=True,
validation_steps=int(num_val / val_batch_size),
max_q_size=10,
workers=1)