-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathjson2coco_crop_flip_region_classify1.py
413 lines (381 loc) · 14.6 KB
/
json2coco_crop_flip_region_classify1.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
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
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
import os
import json
import fnmatch
from pprint import pprint
import cv2
from region2mask import Polygon2Mask,Circle2Mask,CaculateArea,getPolygonCorners,getBoundingBox,circle2Polygon,corners2xy,rect2PolygonCorners
from time import gmtime, strftime
from random import randint
from utils import getRegionLabelDic,getJsonFiles
import numpy as np
from skimage.draw import polygon
import PIL.Image as Image
def cutMainROI(img):
x=img[img.shape[0]/2,:,:].sum(1)
r=0
print img.shape
print x
for v in x:
if v>20:
r+=1
print r
s_x = 0
s_y = 0
if img.shape[0]<r:
return img,s_x,s_y
s_x = (img.shape[1]-r)/2
s_y = (img.shape[0]-r)/2
print s_x
print s_y
cut_img = img[int(s_y):int(s_y+r),int(s_x):int(s_x+r)]
print cut_img.shape
return cut_img,s_x,s_y
def cutMainROI1(img):
x=img[img.shape[0]/2,:,:].sum(1)
xx = img[img.shape[0]/2,:,:]
yy = img[:,img.shape[1]/2,:]
x_s = 0
x_e = 0
threshold = 10
for i in range(len(xx)):
if not (xx[i][0]<10 and xx[i][1]<10 and xx[i][2]<10):
x_s = i
break
for i in range(len(xx)):
if not (xx[len(xx)-i-1][0]<10 and xx[len(xx)-i-1][1]<10 and xx[len(xx)-i-1][2]<10):
x_e = len(xx)-i
break
y_s = 0
y_e = 0
for i in range(len(yy)):
if not (yy[i][0]<10 and yy[i][1]<10 and yy[i][2]<10):
y_s = i
break
for i in range(len(yy)):
if not (yy[len(yy)-i-1][0]<10 and yy[len(yy)-i-1][1]<10 and yy[len(yy)-i-1][2]<10):
y_e = len(yy)-i
break
print [y_s,y_e,x_s,x_e]
cut_img = img[int(y_s):int(y_e),int(x_s):int(x_e)]
return cut_img,x_s,y_s
#-----------get dictionary of region label
def getRegionLabelDic(dic_file_path,split_key,index):
region_labels = []
region_label_file = open(dic_file_path)
line = region_label_file.readline()
while line:
word_list = line.split(split_key)
region_labels.append(word_list[index])
line = region_label_file.readline()
region_label_file.close()
return region_labels
def getJsonFiles(dataset_path = '../DR_COCO3'):
matches = []
paths = []
for root, dirnames, filenames in os.walk(dataset_path):
for filename in fnmatch.filter(filenames, 'via_region_data*.json'):
matches.append(os.path.join(root, filename))
paths.append(root)
return matches, paths
def segToMask( s, h, w, mask, value):
N = len(s)
rr, cc = polygon(np.array(s[1:N:2]).clip(max=h-1), \
np.array(s[0:N:2]).clip(max=w-1)) # (y, x)
mask[rr, cc] = value
return mask
print('searching json...')
matches, paths = getJsonFiles()
print('transfer formate...')
#coco style data
coco_data = {}
coco_data_val = {}
dic_file_path = 'region_label_dic.txt'
region_label_dic = getRegionLabelDic(dic_file_path,'-',1)
region_label_dic_id = 0
coco_data['categories'] = []
coco_data['images'] = []
coco_data['annotations'] = []
coco_data_val['categories'] = []
coco_data_val['images'] = []
coco_data_val['annotations'] = []
region_label_dic_id = 1
for region_label in region_label_dic:
coco_data['categories'].append({'id':region_label_dic_id,
'name':region_label,
'supercategory':region_label})
coco_data_val['categories'].append({'id':region_label_dic_id,
'name':region_label,
'supercategory':region_label})
region_label_dic_id+=1
image_id_c = 0
region_id = 0
dir_count = 0
circle_num = 32
log_file = open('log.txt','w')
category_id_todo = [0,1,2,3,4,5,6,7,8,9,10]
category_train_st = [0,0,0,0,0,0,0,0,0,0,0]
category_val_st = [0,0,0,0,0,0,0,0,0,0,0]
trainlst = open('./annotations/train.lst','w')
vallst = open('./annotations/val.lst','w')
random_v = -1
flag_v = -1
mode_v = 5
flip_img_base_number = 100000
flip_region_base_number = 1000000
flip = True
train_file_list = open('train.txt','w')
val_file_list = open('val.txt','w')
def create_classification(timg,bbox,fi,folder,id,label):
img = timg.copy()
if not os.path.exists(folder):
os.mkdir(folder)
fi.write(str(id)+'.jpg '+str(label)+'\n')
region_img = img[int(bbox[1]):int(bbox[1]+bbox[3]),int(bbox[0]):int(bbox[0]+bbox[2])]
cv2.imwrite(folder+'/'+str(id)+'.jpg',region_img)
cv2.rectangle(img,(bbox[0],bbox[1]),(bbox[0]+bbox[2],bbox[1]+bbox[3]),(255,255,0),5)
cv2.imwrite(folder+'/'+str(id)+'_o.jpg',img)
for file_dir in matches:
with open(file_dir) as orig_data_file:
data = json.load(orig_data_file)
for image_id in data:
#print(paths[dir_count])
img_path = 'images/'+data[image_id]['filename']
img_o = cv2.imread(img_path)
if not os.path.isfile(img_path):
log_file.write(img_path+'---load error!\n')
else:
filelst = ''
#print(img_path)
img,s_x,s_y = cutMainROI1(img_o)
height, width= img.shape[:2]
date_captured = strftime("%Y-%m-%d %H:%M:%S", gmtime())
flag_v = image_id_c - image_id_c%mode_v
if image_id_c%mode_v==0:
random_v = randint(0,mode_v-1)
if flag_v+random_v == image_id_c:
file_name = 'val_'+str(image_id_c)+'_'+data[image_id]['filename']
filelst = './val2014/'+file_name
cv2.imwrite('./val2014/'+file_name,img)
coco_data_val['images'].append({'coco_url':img_path,
'date_captured':date_captured,
'flickr_url':img_path,
'file_name':file_name,
'id':image_id_c,
'height':height,
'width':width,
'license':0})
if flip:
flip_img = cv2.flip(img,1)
flip_file_name = 'flip_'+file_name
cv2.imwrite('./val2014/'+flip_file_name,flip_img)
flipfilelst = './val2014/'+flip_file_name
coco_data_val['images'].append({'coco_url':img_path,
'date_captured':date_captured,
'flickr_url':img_path,
'file_name':flip_file_name,
'id':image_id_c+flip_img_base_number,
'height':height,
'width':width,
'license':0})
else:
file_name = 'train_'+str(image_id_c)+'_'+data[image_id]['filename']
filelst = './train2014/'+file_name
cv2.imwrite('./train2014/'+file_name,img)
coco_data['images'].append({'coco_url':img_path,
'date_captured':date_captured,
'flickr_url':img_path,
'file_name':file_name,
'id':image_id_c,
'height':height,
'width':width,
'license':0})
if flip:
flip_img = cv2.flip(img,1)
flip_file_name = 'flip_'+file_name
cv2.imwrite('./train2014/'+flip_file_name,flip_img)
flipfilelst = './train2014/'+flip_file_name
coco_data['images'].append({'coco_url':img_path,
'date_captured':date_captured,
'flickr_url':img_path,
'file_name':flip_file_name,
'id':image_id_c+flip_img_base_number,
'height':height,
'width':width,
'license':0})
labels_mask = np.zeros((height,width),dtype = np.uint32)
instances_mask = np.zeros((height,width),dtype = np.uint32)
instances_count = np.zeros(len(category_id_todo))
for region in data[image_id]['regions']:
for region_attribute in data[image_id]['regions'][region]['region_attributes']:
if data[image_id]['regions'][region]['region_attributes'][region_attribute]=='default' or data[image_id]['regions'][region]['region_attributes'][region_attribute]=='Fovea':
continue
#category_id = region_label_dic.index(data[image_id]['regions'][region]['region_attributes'][region_attribute])+1
id_count=0
category_id=0
for region_label in region_label_dic:
id_count+=1
if region_label in data[image_id]['regions'][region]['region_attributes'][region_attribute]:
category_id=id_count
break
if category_id==0:
print 'error:category_id is 0'
print data[image_id]['regions'][region]['region_attributes'][region_attribute]
continue
if category_id in category_id_todo:
bad_region = 0
circle_ca = ['big_circle','small_circle','circle','ellipse']
if data[image_id]['regions'][region]['shape_attributes']['name'] in circle_ca:
#print(data[image_id]['regions'][region]['shape_attributes']['name'])
#print(data[image_id]['regions'][region]['shape_attributes'])
r =0
if data[image_id]['regions'][region]['shape_attributes']['name'] =='big_circle':
if width<1000:
r = int(80 *float(width)/1600+20);
else:
r = int(80 *float(width)/1600);
elif data[image_id]['regions'][region]['shape_attributes']['name'] =='small_circle':
if width<1000:
r = int(60*float(width)/1600+15);
else:
r = int(60*float(width)/1600);
elif data[image_id]['regions'][region]['shape_attributes']['name'] =='ellipse':
r = (data[image_id]['regions'][region]['shape_attributes']['ry']+data[image_id]['regions'][region]['shape_attributes']['rx'])/2
else:
r = data[image_id]['regions'][region]['shape_attributes']['r']
'''
if 'Microaneurysms' in data[image_id]['regions'][region]['region_attributes'][region_attribute]:
print 'r:'+str(r)
if r>35:
r = r/2
print 'fixed r'
if '(Hemorrhages)' in data[image_id]['regions'][region]['region_attributes'][region_attribute]:
print 'r:'+str(r)
r = r*2/3
print 'fixed r:'+str(r)
'''
circle = (data[image_id]['regions'][region]['shape_attributes']['cx'],
data[image_id]['regions'][region]['shape_attributes']['cy'],
r)
corners = circle2Polygon(circle,circle_num)
xy,x,y = corners2xy(corners)
bbox = getBoundingBox(x,y)
area = CaculateArea(Polygon2Mask([width,height],corners,category_id+1))
if area ==0:
print('stoped:'+str(data[image_id]['regions'][region]['shape_attributes']))
print('---:'+str(circle))
if bbox[2]==0 or bbox[3]==0:
bad_region=1
elif data[image_id]['regions'][region]['shape_attributes']['name']=='rect':
rect = [data[image_id]['regions'][region]['shape_attributes']['x'],
data[image_id]['regions'][region]['shape_attributes']['y'],
data[image_id]['regions'][region]['shape_attributes']['width'],
data[image_id]['regions'][region]['shape_attributes']['height']]
bbox = rect
corners = rect2PolygonCorners(rect)
xy,x,y = corners2xy(corners)
area = CaculateArea(Polygon2Mask([width,height],corners,category_id+1))
if bbox[2]==0 or bbox[3]==0:
bad_region=1
elif data[image_id]['regions'][region]['shape_attributes']['name']=='polygon':
x = data[image_id]['regions'][region]['shape_attributes']['all_points_x']
y = data[image_id]['regions'][region]['shape_attributes']['all_points_y']
corners = getPolygonCorners(x,y)
area = CaculateArea(Polygon2Mask([width,height],corners,category_id+1))
xy,tx,ty = corners2xy(corners)
bbox = getBoundingBox(x,y)
if bbox[2]==0 or bbox[3]==0:
bad_region=1
if bad_region == 0:
for index_xy in range(len(xy)):
if index_xy%2==0:
xy[index_xy]-=s_x
else:
xy[index_xy]-=s_y
if xy[index_xy]<0:
xy[index_xy] = 0
bbox[0]=bbox[0]-s_x
bbox[1]=bbox[1]-s_y
if bbox[0]<0:
bbox[0]=0
if bbox[1]<0:
bbox[1]=0
if flip:
flip_xy = xy
flip_bbox = bbox[:]
for index_xy in range(len(flip_xy)):
if index_xy%2==0:
flip_xy[index_xy] = width - flip_xy[index_xy]-1
flip_bbox[0] = width - flip_bbox[0]-flip_bbox[2] -1
if flag_v+random_v == image_id_c:
category_val_st[category_id_todo.index(category_id)]+=1
coco_data_val['annotations'].append({'id':region_id,
'image_id':image_id_c,
'category_id':category_id,
'iscrowd':0,
'area':area,
'bbox':bbox,
'segmentation':[xy]})
create_classification(img,bbox,val_file_list,'val',region_id,category_id-1)
if flip:
coco_data_val['annotations'].append({'id':region_id+flip_region_base_number,
'image_id':image_id_c+flip_img_base_number,
'category_id':category_id,
'iscrowd':0,
'area':area,
'bbox':flip_bbox,
'segmentation':[flip_xy]})
create_classification(flip_img,flip_bbox,val_file_list,'val',region_id+flip_region_base_number,category_id-1)
else:
category_train_st[category_id_todo.index(category_id)]+=1
coco_data['annotations'].append({'id':region_id,
'image_id':image_id_c,
'category_id':category_id,
'iscrowd':0,
'area':area,
'bbox':bbox,
'segmentation':[xy]})
create_classification(img,bbox,train_file_list,'train',region_id,category_id-1)
if flip:
coco_data['annotations'].append({'id':region_id+flip_region_base_number,
'image_id':image_id_c+flip_img_base_number,
'category_id':category_id,
'iscrowd':0,
'area':area,
'bbox':flip_bbox,
'segmentation':[flip_xy]})
create_classification(flip_img,flip_bbox,train_file_list,'train',region_id+flip_region_base_number,category_id-1)
labels_mask = segToMask(xy,height,width,labels_mask,category_id)
instances_mask = segToMask(xy,height,width,instances_mask,category_id*1000+instances_count[category_id])
instances_count[category_id] += 1
region_id+=1
save_folder = ''
if flag_v+random_v == image_id_c:
save_folder = './val2014/'
lstline = str(image_id_c) + '\t' +filelst+'\t'+save_folder+str(image_id_c)+'_labelIds.png\n'
vallst.write(lstline)
else:
save_folder = './train2014/'
lstline = str(image_id_c) + '\t' +filelst+'\t'+save_folder+str(image_id_c)+'_labelIds.png\n'
trainlst.write(lstline)
labels_mask_im = Image.fromarray(labels_mask)
labels_mask_im.save(save_folder+str(image_id_c)+'_labelIds.png')
instances_mask_im = Image.fromarray(instances_mask)
instances_mask_im.save(save_folder+str(image_id_c)+'_instanceIds.png')
image_id_c+=1
dir_count+=1
train_file_list.close()
val_file_list.close()
log_file.close()
vallst.close()
trainlst.close()
print('save file...')
with open('./annotations/instances_train2014.json', 'w') as outfile:
json.dump(coco_data, outfile)
outfile.close()
with open('./annotations/instances_val2014.json', 'w') as outfile:
json.dump(coco_data_val, outfile)
outfile.close()
print('total_images:'+str(image_id_c))
print('total_regions:'+str(region_id))
print('train:'+str(category_train_st))
print('val:'+str(category_val_st))