forked from qianrenjian/ICDAR2019-ReCTS
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdivide_train_test.py
110 lines (101 loc) · 4.68 KB
/
divide_train_test.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
import tensorflow as tf
import glob
import os.path
import time
from shutil import copy2
# 原始输入数据的目录，这个目录下有五个子目录，每个子目录底下保存属于该类别的所有图片
INPUT_DATA = 'H:/why_workspace/ReCTS/img_dir2/'
TRAIN_OUTPUT = 'H:/why_workspace/ReCTS/img_dir3/train/'
TEST_OUTPUT = 'H:/why_workspace/ReCTS/img_dir3/validation/'
# 读取数据并将数据分割成训练数据测试数据(未处理的图像地址list)
def cpy_image_lists(sess):
sub_dirs = []
sub_file = os.listdir(INPUT_DATA)
img = []
for dirs in sub_file:
dst_train = os.path.join(TRAIN_OUTPUT, dirs)
dst_test = os.path.join(TEST_OUTPUT, dirs)
os.mkdir(dst_train)
os.mkdir(dst_test)
img = os.listdir(os.path.join(INPUT_DATA, dirs))
num_pic = len(img)
num_train = int(0.8*num_pic)
if num_train == 0:
src = os.path.join(os.path.join(INPUT_DATA, dirs), img[0])
copy2(src, dst_train)
src = os.path.join(os.path.join(INPUT_DATA, dirs), img[0])
copy2(src, dst_test)
else:
for i in range(0, num_train):
src = os.path.join(os.path.join(INPUT_DATA, dirs), img[i])
copy2(src, dst_train)
for i in range(num_train, num_pic):
src = os.path.join(os.path.join(INPUT_DATA, dirs), img[i])
copy2(src, dst_test)
img = []
# for dirs in sub_file:
# os.mkdir(os.path.join(TRAIN_OUTPUT, dirs))
# #print(dirs)
# m = os.path.join(INPUT_DATA, dirs)
# if os.path.isdir(m):
# sub_dirs.append(m)
# for sub_dir in sub_dirs:
# next_sub_dirs = os.listdir(sub_dir)
# #print('next_sub_dirs is ', next_sub_dirs)
# # 获取一个子目录中所有的图片文件
# for next_sub_dir in next_sub_dirs:
# file_list = []
# file_glob = os.path.join(sub_dir, next_sub_dir)
# print(file_glob)
# #print(file_glob[32:36])
# file_list.extend(file_glob)
# num_example = len(file_list)
# num_train = int(num_example * 0.8)
# #print(file_list)
# print(num_example)
#print(file_list[1])
# for i in range(0, num_train):
# print(file_list[i])
#copy2(file_list[i])
# #for extension in extensions:
# # print('*****当前图片后缀为：', extension, " *****")
# # os.path.join()合成地址
# file_glob = os.path.join(sub_dir, next_sub_dir)
# #print(file_glob)
# # print("*****当前图片地址格式为：", file_glob, " *****")
# # glob.glob（）搜索所有符合要求的文件
# file_list.extend(glob.glob(file_glob))
# if not file_list: # 如果file_list为空list，就判断为False
# continue
#
# # 处理图片数据
# num_example = len(file_list)
# num_train = int(num_example * 0.8)
# # print("*****当前file_list的图片数量为：", num_example, "*****")
# # 创建目录文件夹
# #os.mkdir(os.path.join(TRAIN_OUTPUT, next_sub_dir))
# #print(os.path.join(TRAIN_OUTPUT, next_sub_dir))
# #os.mkdir(os.path.join(TEST_OUTPUT, next_sub_dir))
# #print(os.path.join(TEST_OUTPUT, next_sub_dir))
# for i in range(0, num_train):
# png_name1 = os.path.basename(file_list[i])
# train_output1 = os.path.join(TRAIN_OUTPUT, next_sub_dir, png_name1)
# #print(file_list[i])
# #print(train_output1)
# #copy2(file_list[i], train_output1)
#
# for n in range(num_example - num_train):
# png_name2 = os.path.basename(file_list[n + num_train])
# train_output2 = os.path.join(TEST_OUTPUT, next_sub_dir, png_name2)
# #copy2(file_list[n + num_train], train_output2)
# end_time = time.time()
# #print("the folder {0} takes time is {1}".format(next_sub_dir, end_time - start_time))
def main(argv=None):
config = tf.ConfigProto(allow_soft_placement=True)
# 设置根据需求增长使用的内存
config.gpu_options.allow_growth = True
with tf.Session(config=config) as sess:
cpy_image_lists(sess)
print("all finished")
if __name__ == '__main__':
tf.app.run()