-
-
Notifications
You must be signed in to change notification settings - Fork 466
Expand file tree
/
Copy pathrecognition.py
More file actions
476 lines (426 loc) · 20.3 KB
/
Copy pathrecognition.py
File metadata and controls
476 lines (426 loc) · 20.3 KB
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
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
# coding=utf-8
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import os, json, time, sys
import shutil
import time
import os.path
import requests
from uuid import uuid1
import numpy as np
from utilslib.save2gst import generate_protocol_string
import classifier_classify_new as classifier
from faces import save_embedding
all_face_index = 0 #每当识别出一个人脸就+1，当2个人同时出现在图片里面并且都不认识，需要区分开来
BASEDIR = os.getenv('RUNTIME_BASEDIR',os.path.abspath(os.path.dirname(__file__)))
TMP_DIR_PATH = os.path.join(BASEDIR, 'data', 'faces', 'tmp_pic_path')
UPLOAD_FOLDER = os.path.join(BASEDIR, 'image')
DATABASE = 'sqlite:///' + os.path.join(BASEDIR, 'data', 'data.sqlite')
face_tmp_objid = None
obje_tmp_objid = None
EN_OBJECT_DETECTION = False
FACE_DETECTION_WITH_DLIB = False # Disable DLIB at this time
EN_SOFTMAX = False
SOFTMAX_ONLY = False
isUpdatingDataSet = False
webShowFace = False
EXT_IMG='png'
DO_NOT_UPLOAD_IMAGE = False
DO_NOT_REPORT_TO_SERVER = False
FOR_ARLO = True
USE_DEFAULT_DATA=True # Enable to use "groupid_default" for SVM training
SVM_CLASSIFIER_ENABLED=True
SVM_SAVE_TEST_DATASET=True
SVM_TRAIN_WITHOUT_CATEGORY=True
SVM_HIGH_SCORE_WITH_DB_CHECK=True
svm_face_dataset=None
svm_face_embedding=None
svm_tmp_dir=None
svm_face_testdataset=None
svm_stranger_testdataset=None
data_collection=None
def init_fs():
global svm_face_dataset
global svm_face_embedding
global svm_tmp_dir
global svm_face_testdataset
global svm_stranger_testdataset
try:
if not os.path.exists(UPLOAD_FOLDER):
os.makedirs(UPLOAD_FOLDER)
# if not os.path.exists(os.path.join(BASEDIR, 'data.sqlite')):
# db.create_all()
if not os.path.exists(os.path.join(BASEDIR, 'data', 'data.sqlite')):
if os.path.exists(os.path.join(BASEDIR, 'data_init')):
shutil.copyfile(os.path.join(BASEDIR, 'data_init'), os.path.join(BASEDIR, 'data', 'data.sqlite'))
if not os.path.exists(TMP_DIR_PATH):
os.makedirs(TMP_DIR_PATH)
if SVM_CLASSIFIER_ENABLED:
svm_face_dataset = os.path.join(BASEDIR, 'data', 'face_dataset')
svm_face_embedding = os.path.join(BASEDIR, 'data', 'face_embedding')
svm_tmp_dir = os.path.join(BASEDIR, 'data', 'faces', 'noname', 'person')
svm_face_testdataset = os.path.join(BASEDIR, 'data', 'face_testdataset')
svm_stranger_testdataset = os.path.join(BASEDIR, 'data', 'stranger_testdataset')
if not os.path.exists(svm_face_dataset):
os.mkdir(svm_face_dataset)
if not os.path.exists(svm_face_embedding):
os.mkdir(svm_face_embedding)
if not os.path.exists(svm_tmp_dir):
os.makedirs(svm_tmp_dir)
if not os.path.exists(svm_face_testdataset):
os.mkdir(svm_face_testdataset)
if not os.path.exists(svm_stranger_testdataset):
os.mkdir(svm_stranger_testdataset)
except Exception as e:
print(e)
def updatePeopleImgURL(ownerid, url, embedding, uuid, objid, img_type, accuracy, fuzziness, sqlId, style, img_ts, tid,
p_ids, waiting):
print('debug updatePeopleImgURL 1')
if len(url) < 1 or len(uuid) < 1 or len(objid) < 1 or len(img_type) < 1:
return
if not DO_NOT_REPORT_TO_SERVER:
print('save2gst')
save2gst(uuid, objid, url, '', 'face', accuracy, int(fuzziness), int(sqlId), style, img_ts, tid, p_ids, waiting) # 发送请求给workai
def upload_forecast_result(key, forecast_result, json_data, num_p):
uuid = forecast_result['uuid']
face_id = forecast_result['face_id']
face_accuracy = forecast_result['face_accuracy']
people_sqlId = forecast_result['people_sqlId']
align_image_path = forecast_result['align_image_path']
img_style_str = forecast_result['img_style_str']
ts = forecast_result['ts']
trackerId = forecast_result['trackerId']
face_fuzziness = forecast_result['face_fuzziness']
people = forecast_result['people']
p_ids = forecast_result['p_ids']
embedding_string = forecast_result['embedding_string']
#embedding_bytes = embedding_string.encode('utf-8')
img_type = forecast_result['img_type']
waiting = forecast_result['waiting']
do_not_report_to_server = DO_NOT_REPORT_TO_SERVER
uploadedimgurl = None
return generate_protocol_string(key, face_id, align_image_path,
embedding='', uuid=uuid,
DO_NOT_REPORT_TO_SERVER=do_not_report_to_server,block=False,
objid=face_id, img_type=img_type,
accuracy=face_accuracy, fuzziness=face_fuzziness, sqlId=people_sqlId,
style=img_style_str, ts=ts, tid=str(trackerId), p_ids=p_ids, waiting = waiting)
def face_recognition_on_embedding(align_image_path, embedding, totalPeople, blury, uuid,
current_groupid, style, trackerId,
timestamp1, ts, embedding_path):
img_objid = trackerId
print("img_objid = {}".format(img_objid))
print("number of people=%d" % (totalPeople))
if totalPeople > 1:
trackerId = str(uuid1())
number_people=totalPeople
img_style_str = style
img_style = img_style_str
json_data = {'detected':False, 'recognized': False, 'style': img_style_str}
forecast_result = {'people':None,
'img_type': 'face',
'people_sqlId': 0,
'face_id': None, # 当前这个人脸图片的objId/face_id
'trackerId': trackerId,
'uuid': uuid,
'ts': ts,
'img_style_str': img_style_str,
'align_image_path': align_image_path,
'face_accuracy': 0,
'face_fuzziness': 0,
'embedding_string': '',
'waiting': False,
'p_ids': None,
} # 保存发送给uploadImage的数据
forecast_result['face_fuzziness'] = blury
print(">>> blury=%d" %(blury))
json_data['detected'] = True
print("2 %.2f seconds" % (time.time() - timestamp1))
if forecast_result['face_fuzziness'] < int(data_collection.get("blury_threhold")):
json_data, forecast_result = get_empty_faceid(current_groupid, uuid, embedding,
img_style, number_people, img_objid,
forecast_result)
print("Too blurry image, skip it img_objid={}, trackerId={}".format(img_objid, trackerId))
elif SVM_CLASSIFIER_ENABLED is True:
#img_style = 'front'
# embedding = embedding.reshape((1, -1))
forecast_result['waiting'] = False
json_data, forecast_result = SVM_classifier(embedding,align_image_path,
uuid,current_groupid,img_style,number_people,img_objid,json_data,forecast_result, embedding_path)
print("3 %.2f seconds" % (time.time() - timestamp1))
if json_data['recognized'] is True:
if webShowFace is True:
showRecognizedImage(forecast_result['align_image_path'], 1)
forecast_result['trackerId'] = trackerId
# 人脸预测结果发送
key = str(uuid1())
_,api_url,payload = upload_forecast_result(key, forecast_result, json_data, number_people)
json_data['key'] = key
json_data['face_fuzziness'] = blury
return json_data, {'api_url':api_url,'payload':payload}
def get_empty_faceid(current_groupid, uuid, embedding,
img_style, number_people, img_objid, forecast_result):
"""
当softmax无结果时（无模型/预测置信度低）调用遍历数据库识别
:param current_groupid:
:param uuid:
:param embedding:
:param img_style:
:param number_people:
:param img_objid:
:return:
"""
json_data = {'detected': True, 'recognized': False}
face_id = img_objid + str(all_face_index).zfill(4)
json_data['recognized'] = False
json_data['face_id'] = face_id
json_data['accuracy'] = 0
json_data['style'] = img_style
forecast_result['face_id'] = face_id
forecast_result['face_accuracy'] = 0
embedding_string = ','.join(str(x) for x in embedding)
forecast_result['embedding_string'] = embedding_string
return json_data, forecast_result
def SVM_classifier(embedding,align_image_path,uuid,current_groupid,img_style,number_people, img_objid,json_data, forecast_result, embedding_path):
#Save image to src/face_dataset_classify/group/person/
if SVM_SAVE_TEST_DATASET is True:
group_path = os.path.join(BASEDIR,svm_face_testdataset, current_groupid)
if not os.path.exists(group_path):
os.mkdir(group_path)
print('test dataset group_path=%s' % group_path)
pkl_path = ""
if SVM_TRAIN_WITHOUT_CATEGORY is True:
pkl_path = '{}/data/faces/{}/{}/classifier_182.pkl'.format(BASEDIR,current_groupid, 'front')
face_dataset_path = '{}/data/faces/{}/{}/face_dataset'.format(BASEDIR,current_groupid, 'front')
else:
pkl_path = '{}/data/faces/{}/{}/classifier_182.pkl'.format(BASEDIR,current_groupid, img_style)
face_dataset_path = '{}/data/faces/{}/{}/face_dataset'.format(BASEDIR,current_groupid, img_style)
svm_detected = False
if os.path.exists(pkl_path):
nrof_classes = 0
if os.path.exists(face_dataset_path):
classes = [path for path in os.listdir(face_dataset_path) \
if os.path.isdir(os.path.join(face_dataset_path, path))]
nrof_classes = len(classes)
print("SVM_classifier: nrof_classes={}".format(nrof_classes))
#tmp_image_path = BASEDIR + '/data/faces/noname/person/face_tmp.'+EXT_IMG
#shutil.copyfile(align_image_path, tmp_image_path)
# 输入embedding的预测方法, 速度很快
svm_stime = time.time()
_, human_string, score, top_three_name, judge_result = classifier.classify([embedding], pkl_path, embedding_path)
if top_three_name:
top_three_faceid = [name.split(' ')[1] for name in top_three_name]
else:
top_three_faceid = None
print('-> svm classify cost {}s'.format(time.time()-svm_stime))
print("current value of score_1 ", float(data_collection.get("score_1")))
print("current value of score_2 ", float(data_collection.get("score_2")))
print("current value of fuzziness_1 ", float(data_collection.get("fuzziness_1")))
print("current value of fuzziness_2 ", float(data_collection.get("fuzziness_2")))
print("current value of update interval ", float(data_collection.get("_interval")))
if human_string is not None:
message = ""
message2 = ""
face_id = human_string.split(' ')[1]
score = round(score, 2)
fuzziness = forecast_result['face_fuzziness']
if USE_DEFAULT_DATA is True:
if face_id == 'defaultfaceid':
score = 0.0
if not FOR_ARLO:
if score >= float(data_collection.get("score_1")) or (score >= float(data_collection.get("score_2")) and fuzziness >= float(data_collection.get("fuzziness_1")) and fuzziness < float(data_collection.get("fuzziness_2"))):
found, total = check_embedding_on_detected_person_forSVM(current_groupid=current_groupid,
embedding=embedding,style=img_style,classid=face_id, nrof_classes=nrof_classes)
if found > 0:
svm_detected = True
message = "<DB Recognized> Face ID: %s %s/%s, 2nd %s/%s" % (face_id, score, img_style, found, total)
else:
message = "<DB 2nd Score Low> Face ID: %s %s/%s, 2nd %s/%s" % (face_id, score, img_style, found, total)
elif 0.35<score<0.8:
message = "Send this face to Zone Waiting: %s,%s"%(score, fuzziness)
forecast_result['waiting'] = True
else:
message = "<1st Score Low> Face ID: %s %s/%s" % (face_id, score, img_style)
else:
if score > float(data_collection.get("score_2")):#0.40
found, total = check_embedding_on_detected_person_forSVM_ByDir(current_groupid=current_groupid,
embedding=embedding,style=img_style,classid=human_string.replace(' ', '_'),nrof_classes=nrof_classes)
if found > 0:
if score > float(data_collection.get("score_1")) or judge_result != 0:#0.9
svm_detected = True
message = "<1, SVM Recognized> Face ID: %s %s/%s, 2nd %s/%s" % (face_id, score, img_style, found, total)
else:
message = "<2, SVM Recognized> Face ID: %s %s/%s, 2nd %s/%s" % (face_id, score, img_style, found, total)
else:
message = "<3, SVM Recognized Not, found=0> Face ID: %s %s/%s, 2nd %s/%s" % (face_id, score, img_style, found, total)
else:
message = "<4, SVM Recognized Not, Low score> Face ID: %s %s/%s" % (face_id, score, img_style)
print(message)
if (message2 != ""):
print(message2)
if svm_detected is True:
json_data['recognized'] = True
json_data['face_id'] = face_id
json_data['accuracy'] = int(score*100)
json_data['style'] = img_style
forecast_result['face_id'] = face_id
forecast_result['face_accuracy'] = score
embedding_string = ','.join(str(x) for x in embedding)
forecast_result['embedding_string'] = embedding_string
else:
forecast_result['p_ids'] = top_three_faceid
print("Not Recognized %s" % face_id)
#sendDebugLogToGroup(uuid, current_groupid, message)
#Save image to src/face_dataset_classify/group/person/
if SVM_SAVE_TEST_DATASET is True:
if svm_detected is True:
svm_face_testdataset_person_path = os.path.join(group_path, human_string)
else:
svm_face_testdataset_person_path = os.path.join(group_path, 'noname')
if not os.path.exists(svm_face_testdataset_person_path):
os.mkdir(svm_face_testdataset_person_path)
print('test dataset person path=%s' % svm_face_testdataset_person_path)
dir = os.path.basename(align_image_path)
name = os.path.splitext(dir)[0]
save_testdataset_filepath = os.path.join(svm_face_testdataset_person_path, name+'_'+str(int(time.time()))+'.png')
print('save classified image to path: %s' % save_testdataset_filepath)
try:
shutil.copyfile(align_image_path, save_testdataset_filepath)
except IOError:
print('cant copy file from {} to {},need check later'.format(align_image_path,save_testdataset_filepath ))
pass
if svm_detected is False:
json_data, forecast_result = get_empty_faceid(current_groupid, uuid, embedding,
img_style, number_people, img_objid,
forecast_result)
print('not in train classification or need to more train_dataset')
return json_data, forecast_result
def check_embedding_on_detected_person_forSVM(current_groupid, embedding, style, classid, nrof_classes):
total = 0
found = 0
people = None
#遍历整个数据库, 检查这个人是谁
if SVM_TRAIN_WITHOUT_CATEGORY is True:
people = People.query.filter_by(group_id=current_groupid, classId=classid).all()
else:
people = People.query.filter_by(group_id=current_groupid, style=style, classId=classid).all()
if people:
for person in people:
val = compare2(embedding, person.embed)
total = total+1
face_accuracy = val
print('face_accuracy={}'.format(face_accuracy))
threshold = 0.70
if nrof_classes <= 5 and nrof_classes > 0:
threshold = 0.90
elif nrof_classes <= 10 and nrof_classes > 0:
threshold = 0.82
if face_accuracy >= threshold:
found = found+1
if total >= 500:
break
return found, total
def check_embedding_on_detected_person_forSVM_ByDir(current_groupid, embedding, style, classid, nrof_classes):
total = 0
found = 0
people = None
if SVM_TRAIN_WITHOUT_CATEGORY is True:
facedir = '{}/data/faces/{}/{}/face_dataset/{}'.format(BASEDIR,current_groupid, 'front', classid)
else:
facedir = '{}/data/faces/{}/{}/face_dataset/{}'.format(BASEDIR,current_groupid, style, classid)
print("check embedding: facedir = {}".format(facedir))
embedding_array = []
if os.path.isdir(facedir):
image_paths = []
images = os.listdir(facedir)
if len(images) < 1:
print("Check embedding: Empty directory: facedir={}".format(facedir))
return 0, 0
for img in images:
img_path = os.path.join(facedir, img)
emb_path = save_embedding.get_embedding_path(img_path)
emb = save_embedding.read_embedding_string(emb_path)
emb = np.asarray(emb)
embedding_array.append(emb)
if len(embedding_array) > 0:
for emb in embedding_array:
val = compare2(embedding, emb)
total = total+1
face_accuracy = val
print('face_accuracy={}'.format(face_accuracy))
'''
threshold = 1.0
if nrof_classes <= 5 and nrof_classes > 0:
threshold = 1.0
elif nrof_classes <= 10 and nrof_classes > 0:
threshold = 1.0
'''
threshold = 0.42
if face_accuracy > threshold:
found = found+1
if total >= 500:
break
return found, total
def compare(emb1, emb2):
dist = np.sqrt(np.sum(np.square(np.subtract(emb1, emb2))))
# d = emb1 - emb2
# sqL2 = np.dot(d, d)
# print("+ Squared l2 distance between representations: {:0.3f}, dist is {:0.3f}".format(sqL2,dist))
# print("+ distance between representations: {:0.3f}".format(dist))
# return sqL2
return dist
def compare2(emb1, emb2):
dist = np.sum([emb2]*emb1, axis=1)
return dist
def compare3(emb1, emb2):
return np.sum(np.square(emb1-emb2))
class DataCollection(object):
def __init__(self, update_freq=10):
self.url = "http://localhost:5000/api/parameters"
self.pre_time = time.time()
self.update_freq = update_freq
self.items = self.fetch()
def fetch(self):
try:
resp = requests.get(self.url)
assert resp.status_code == 200
r = resp.json()
except Exception as e:
print(e)
#print("status_code: %s" %resp.status_code)
# if web server not work return default values.
r = None
if not FOR_ARLO:
r = {
"blury_threhold": "10",
"fuzziness_1": "40",
"fuzziness_2": "200",
"score_1": "0.75",
"score_2": "0.60",
}
else:
r = {
"blury_threhold": "60",
"fuzziness_1": "40",
"fuzziness_2": "200",
"score_1": "0.90",
"score_2": "0.40",
}
if "_interval" in r:
_freq = int(r["_interval"])
if self.update_freq != _freq:
self.update_freq = _freq
r.update({"_interval": self.update_freq})
return r
def reload(self):
cur_time = time.time()
if cur_time - self.pre_time > self.update_freq:
self.pre_time = cur_time
self.items = self.fetch()
def get(self, key):
key = key.lower()
self.reload()
return self.items.get(key, None)
data_collection = DataCollection()
init_fs()