-
Notifications
You must be signed in to change notification settings - Fork 9
/
Copy pathBH_dataset.py
223 lines (187 loc) · 7.47 KB
/
BH_dataset.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
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
import os
import os.path as osp
from os.path import join
import pandas as pd
from glob import glob
import shutil
from multiprocessing import Pool
import cv2
from functools import partial
import time
from pathlib import Path
def get_file(ipath, respath, dirs=('sen2', 'bh', 'cbra')):
'''
:param ipath: subdir includes: sen2, bh, cbra
:param respath:
:return:
'''
imglist = os.listdir(join(ipath, dirs[0]))
sen2 = [join(ipath, dirs[0], i) for i in imglist]
bh = [join(ipath, dirs[1], i) for i in imglist]
cbra = [join(ipath, dirs[2], i) for i in imglist]
df = pd.DataFrame({dirs[0]:sen2, dirs[1]:bh, dirs[2]:cbra})
df.to_csv(respath, sep=',', index=False, header=False)
def split_df(df, value=0, split_rate=0.9):
df1 = df.loc[df[1]==value].sample(frac=1, random_state=1) # extract and shuffle
num_train = int(len(df1)*split_rate)
return df1[:num_train], df1[num_train:]
def split_data(datalist_path, split_rate=0.9, id='2', n1='train', n2='test'):
data_dir = os.path.dirname(datalist_path)
base_name = os.path.basename(datalist_path)[:-4]
train_path = join(data_dir, base_name+'_'+ n1 + id +'.csv')
test_path = join(data_dir, base_name+'_'+ n2 + id +'.csv')
if os.path.exists(train_path) and os.path.exists(test_path):
print('train and test list exist')
return
else:
df = pd.read_csv(datalist_path, sep=',', header=None)
# random sampling
df1 = df.sample(frac=1, random_state=1) # shuffle
num_train = int(len(df1) * split_rate)
df_train = df1[:num_train]
df_test = df1[num_train:]
df_train.to_csv(train_path, index=False, sep=',', header=None)
df_test.to_csv(test_path, index=False, sep=',', header=None)
print('success')
def generate_allfile(ipath=r'D:\data\Landcover\samples62', subdir='s1_vvvhratio',
invalid='shenzhen', suffix='',
numsample=0):
if not isinstance(subdir, list):
subdir = [subdir,]
# cat all tif images
fplist = []
for i in subdir:
tmp = Path(os.path.join(ipath, i)).rglob('*.tif')
fplist.extend(tmp)
imglist = []
for i in fplist:
iname = i.stem+'.tif'
if invalid is not None:
if invalid not in iname:
imglist.append(iname)
else:
imglist.append(iname)
print(len(imglist))
# if invalid is None:
# invalid=''
# else:
# invalid='_del'+invalid
df = pd.DataFrame({'imglist': imglist})
# random select
if numsample != 0:
df = df.sample(n=numsample, random_state=1)
respath = os.path.join(ipath, 'datalist_'+suffix+'.csv') #'datalist'+invalid+'_'+suffix+'.csv')
if not os.path.exists(respath):
df.to_csv(respath, header=False, index=False)
# Train/Test
# train/test=0.6:0.4
split_data(respath, split_rate=0.7, id='_0.7', n1='train', n2='test')
# val/test = 0.1:0.3
respath = os.path.join(ipath, 'datalist_'+suffix+'_test_0.7.csv') #'datalist'+invalid+'_'+suffix+'_test_0.7.csv')
split_data(respath, split_rate=0.33, id='_0.3', n1='val', n2='test')
# absolute path
def generate_allfile_abspath(ipath=r'D:\data\Landcover\samples62',
flist=('china', 'eu', 'usa'),
suffix='globe',
mergetype='',
numsample=0):
# cat all tif images
imglist = []
citylist = []
for city in flist:
file = os.path.join(ipath, 'datalist_'+city+mergetype+'.csv')
# read
df = pd.read_csv(file, header=None)
dflist = df[0].values.tolist()
imglist.extend(dflist)
num = len(dflist)
citylist.extend([city]*num)
print(len(imglist))
s1list = ['s1'+i+'_check' for i in citylist]
s2list = ['s2'+i+ '_check' for i in citylist]
bhlist = ['bh'+i for i in citylist]
df = pd.DataFrame({'imglist': imglist,
's1dir':s1list, 's2dir': s2list,
'bhdir': bhlist })
# random select
if numsample != 0:
df = df.sample(n=numsample, random_state=1)
respath = os.path.join(ipath, 'datalist_'+suffix+mergetype+'.csv') #'datalist'+invalid+'_'+suffix+'.csv')
if not os.path.exists(respath):
df.to_csv(respath, header=False, index=False)
# Train/Test
# train/test=0.6:0.4
# split_data(respath, split_rate=0.7, id='_0.7', n1='train', n2='test')
# # val/test = 0.1:0.3
# respath = os.path.join(ipath, 'datalist_'+suffix+'_test_0.7.csv') #'datalist'+invalid+'_'+suffix+'_test_0.7.csv')
# split_data(respath, split_rate=0.33, id='_0.3', n1='val', n2='test')
def concat_allfile(ipath=r'D:\data\Landcover\samples62',
flist=('china', 'eu', 'usa'),
suffix='globe',
mergetype='',
):
# cat all tif images
df = []
for city in flist:
file = os.path.join(ipath, 'datalist_'+city+mergetype+'.csv')
# read
tmp = pd.read_csv(file, header=None)
df.append(tmp)
df = pd.concat(df)
respath = os.path.join(ipath, 'datalist_'+suffix+mergetype+'.csv') #'datalist'+invalid+'_'+suffix+'.csv')
if not os.path.exists(respath):
df.to_csv(respath, header=False, index=False)
# Train/Test
# train/test=0.6:0.4
# split_data(respath, split_rate=0.7, id='_0.7', n1='train', n2='test')
# # val/test = 0.1:0.3
# respath = os.path.join(ipath, 'datalist_'+suffix+'_test_0.7.csv') #'datalist'+invalid+'_'+suffix+'_test_0.7.csv')
# split_data(respath, split_rate=0.33, id='_0.3', n1='val', n2='test')
def addabspath(ipath=r'D:\data\buildheight\samples',
city='china', flist=None):
if flist is None:
fpath = os.path.join(ipath, 'datalistcopy')
flist = [i for i in Path(fpath).glob('*'+city+'*.csv')]
print(len(flist))
subdir = {'s1': f's1{city}_check',
's2': f's2{city}_check',
'bh': f'bh{city}',
'ge': f'ge{city}_check',
'dem': f'dem{city}',
'dsm': f'dsm{city}',
}
for file in flist:
fname = file.name
file = str(file)
df = pd.read_csv(file, header=None)
for k, v in subdir.items():
df[k] = v
resfile = os.path.join(ipath, fname)
df.to_csv(resfile, header=False, index=False)
if __name__=="__main__":
ipath = r'.\data'
# 2023.11.07: generate samples for usa
generate_allfile(ipath=ipath,
subdir='s1usa_check', invalid=None, suffix='usa')
# 2023.11.9: generate samples for China,
# delete shenzhen and randomly select 15000
generate_allfile(ipath=ipath, subdir='s1china_check',
invalid='shenzhen', suffix='china',
numsample=15000)
# 2023.11.9: generate samples for EU
generate_allfile(ipath=ipath, subdir='s1eu_check',
invalid=None, suffix='eu',
)
# 2023.11.9: generate samples for China, EU, USA
# should merge the training and test data directly from the three countries
# add s1,s2,bh dir to datalist
addabspath(ipath, 'china')
addabspath(ipath, 'eu')
addabspath(ipath, 'usa')
mergelist =['', '_test_0.7', '_train_0.7', '_test_0.7_test_0.3', '_test_0.7_val_0.3']
for i in mergelist:
concat_allfile(ipath=ipath,
flist=('china', 'eu', 'usa'),
suffix = 'globe',
mergetype=i,
)