84648488
 Chunk
reverted.
|
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
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
|
# -*- coding: utf-8 -*-
__author__ = 'chunk'
from mdata import *
from mfeat import HOG, IntraBlockDiff
from mspark import SC
import os, sys
from PIL import Image
from hashlib import md5
import csv
import shutil
from common import *
import json
import collections
import happybase
class DataCV(DataDumperBase):
def __init__(self, base_dir='/home/hadoop/data/HeadShoulder/', category='Train'):
DataDumperBase.__init__(self, base_dir, category)
self.data_dir = self.base_dir + self.category + '/'
self.dict_data = {}
self.table_name = self.base_dir.split('/')[-2] + '-' + self.category
self.sparkcontex = None
def format(self):
self.extract()
def _hash_copy(self, image, ispos):
if not image.endswith('jpg'):
img = Image.open(image)
img.save('res/tmp.jpg', format='JPEG')
image = 'res/tmp.jpg'
with open(image, 'rb') as f:
index = md5(f.read()).hexdigest()
self.dict_data[index] = ispos
# origion:
# dir = base_dir + 'Img/Train/' + index[:3]
dir = self.img_dir + index[:3] + '/'
if not os.path.exists(dir):
os.makedirs(dir)
image_path = dir + index[3:] + '.jpg'
# print image_path
if not os.path.exists(image_path):
shutil.copy(image, image_path)
else:
pass
def extract(self):
for path, subdirs, files in os.walk(self.data_dir):
for name in files:
imagepath = os.path.join(path, name)
print imagepath
if imagepath.split('/')[-2].startswith('pos'):
self._hash_copy(imagepath, True)
else:
self._hash_copy(imagepath, False)
def build_list(self):
assert self.list_file != None
ordict_img = collections.OrderedDict(sorted(self.dict_data.items(), key=lambda d: d[0]))
with open(self.list_file, 'w') as f:
tsvfile = csv.writer(f, delimiter='\t')
for key, value in ordict_img.items():
tsvfile.writerow([key] + [value])
def get_table(self):
if self.table != None:
return self.table
if self.connection is None:
c = happybase.Connection('HPC-server')
self.connection = c
tables = self.connection.tables()
if self.table_name not in tables:
families = {'cf_pic': dict(),
'cf_info': dict(max_versions=10),
'cf_tag': dict(),
'cf_feat': dict(),
}
self.connection.create_table(name=self.table_name, families=families)
table = self.connection.table(name=self.table_name)
self.table = table
return table
def store_image(self):
if self.table == None:
self.table = self.get_table()
dict_databuf = {}
with open(self.list_file, 'rb') as tsvfile:
tsvfile = csv.reader(tsvfile, delimiter='\t')
for line in tsvfile:
path_img = self.img_dir + line[0][:3] + '/' + line[0][3:] + '.jpg'
if path_img:
with open(path_img, 'rb') as fpic:
dict_databuf[line[0] + '.jpg'] = fpic.read()
try:
with self.table.batch(batch_size=5000) as b:
for imgname, imgdata in dict_databuf.items():
b.put(imgname, {'cf_pic:data': imgdata})
except ValueError:
raise
pass
def store_tag(self, feattype='hog'):
if self.table == None:
self.table = self.get_table()
dict_tagbuf = {}
with open(self.list_file, 'rb') as tsvfile:
tsvfile = csv.reader(tsvfile, delimiter='\t')
for line in tsvfile:
dict_tagbuf[line[0] + '.jpg'] = line[1]
try:
with self.table.batch(batch_size=5000) as b:
for imgname, imgtag in dict_tagbuf.items():
b.put(imgname, {'cf_tag:' + feattype: imgtag})
except ValueError:
raise
pass
def get_feat(self, image, feattype='hog', **kwargs):
size = kwargs.get('size', (48, 48))
if feattype == 'hog':
feater = HOG.FeatHOG(size=size)
elif feattype == 'ibd':
feater = IntraBlockDiff.FeatIntraBlockDiff()
else:
raise Exception("Unknown feature type!")
desc = feater.feat(image)
return desc
def extract_feat(self, feattype='hog'):
if feattype == 'hog':
feater = HOG.FeatHOG(size=(48, 48))
elif feattype == 'ibd':
feater = IntraBlockDiff.FeatIntraBlockDiff()
else:
raise Exception("Unknown feature type!")
list_image = []
with open(self.list_file, 'rb') as tsvfile:
tsvfile = csv.reader(tsvfile, delimiter='\t')
for line in tsvfile:
list_image.append(line[0])
dict_featbuf = {}
for imgname in list_image:
# if imgtag == 'True':
image = self.img_dir + imgname[:3] + '/' + imgname[3:] + '.jpg'
desc = feater.feat(image)
dict_featbuf[imgname] = desc
for imgname, desc in dict_featbuf.items():
# print imgname, desc
dir = self.feat_dir + imgname[:3] + '/'
if not os.path.exists(dir):
os.makedirs(dir)
featpath = dir + imgname[3:].split('.')[0] + '.' + feattype
with open(featpath, 'wb') as featfile:
featfile.write(json.dumps(desc.tolist()))
def store_feat(self, feattype='hog'):
if self.table == None:
self.table = self.get_table()
dict_featbuf = {}
for path, subdirs, files in os.walk(self.feat_dir):
for name in files:
featpath = os.path.join(path, name)
# print featpath
with open(featpath, 'rb') as featfile:
imgname = path.split('/')[-1] + name.replace('.' + feattype, '.jpg')
dict_featbuf[imgname] = featfile.read()
try:
with self.table.batch(batch_size=5000) as b:
for imgname, featdesc in dict_featbuf.items():
b.put(imgname, {'cf_feat:' + feattype: featdesc})
except ValueError:
raise
pass
def load_data(self, mode='local', feattype='hog'):
INDEX = []
X = []
Y = []
if mode == "local":
dict_tagbuf = {}
with open(self.list_file, 'rb') as tsvfile:
tsvfile = csv.reader(tsvfile, delimiter='\t')
for line in tsvfile:
imgname = line[0] + '.jpg'
dict_tagbuf[imgname] = line[1]
dict_dataset = {}
for path, subdirs, files in os.walk(self.feat_dir):
for name in files:
featpath = os.path.join(path, name)
with open(featpath, 'rb') as featfile:
imgname = path.split('/')[-1] + name.replace('.' + feattype, '.jpg')
dict_dataset[imgname] = json.loads(featfile.read())
for imgname, tag in dict_tagbuf.items():
tag = 1 if tag == 'True' else 0
INDEX.append(imgname)
X.append(dict_dataset[imgname])
Y.append(tag)
elif mode == "remote" or mode == "hbase":
if self.table == None:
self.table = self.get_table()
col_feat, col_tag = 'cf_feat:' + feattype, 'cf_tag:' + feattype
for key, data in self.table.scan(columns=[col_feat, col_tag]):
X.append(json.loads(data[col_feat]))
Y.append(1 if data[col_tag] == 'True' else 0)
elif mode == "spark" or mode == "cluster":
if self.sparkcontex == None:
self.sparkcontex = SC.Sparker(host='HPC-server', appname='ImageCV', master='local')
result = self.sparkcontex.read_habase(self.table_name) # result = {key:[feat,tag],...}
for feat, tag in result:
X.append(feat)
Y.append(tag)
else:
raise Exception("Unknown mode!")
return X, Y
|