-
Notifications
You must be signed in to change notification settings - Fork 19
/
DSOD.py
569 lines (543 loc) · 32.6 KB
/
DSOD.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
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
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
import os
import gc
import xml.etree.ElementTree as etxml
import math
import random
import skimage.io
import skimage.transform
import numpy as np
import tensorflow as tf
import tensorlayer as tl
from tensorlayer.layers import *
from tensorflow.python.ops import variables
import time
from imutils.object_detection import non_max_suppression
import imutils
import cv2
import matplotlib.pyplot as plt
batch_size = 16
running_count = 5000
file_name_list = os.listdir('./train_datasets/voc2012/JPEGImages/')
lable_arr = ['background','aeroplane','bicycle','bird','boat','bottle','bus','car','cat','chair','cow','diningtable','dog','horse','motorbike','person','pottedplant','sheep','sofa','train','tvmonitor']
img_size = [300, 300]
# 分类总数量
classes_size = 21
# 背景分类的值
background_classes_val = 0
# 每个特征图单元的default box数量
default_box_size = [6, 6, 6, 6, 6, 6]
# default box 尺寸长宽比例
box_aspect_ratio = [
[0.5, 1.0, 2.0, 3.0,1/3.0],
[0.5, 1.0, 2.0, 3.0, 1 / 3.0],
[0.5, 1.0, 2.0, 3.0, 1 / 3.0],
[0.5, 1.0, 2.0, 3.0, 1 / 3.0],
[0.5, 1.0, 2.0, 3.0, 1 / 3.0],
[0.5, 1.0, 2.0, 3.0, 1 / 3.0]
]
# 最小default box面积比例
min_box_scale = 0.1
# 最大default box面积比例
max_box_scale = 0.9
# 每个特征层的面积比例
# numpy生成等差数组，效果等同于论文中的s_k=s_min+(s_max-s_min)*(k-1)/(m-1)
default_box_scale = np.linspace(min_box_scale, max_box_scale, num=np.amax(default_box_size))
print('## default_box_scale:' + str(default_box_scale))
# 卷积步长
conv_strides_1 = [1, 1, 1, 1]
conv_strides_2 = [1, 2, 2, 1]
conv_strides_3 = [1, 3, 3, 1]
tl_strides_1 = (1, 1)
tl_strides_2 = (2, 2)
tl_strides_3 = (3, 3)
# 池化窗口
pool_size = [1, 2, 2, 1]
tl_pool_size = (2, 2)
# 池化步长
pool_strides = [1, 2, 2, 1]
tl_pool_strides = (2, 2)
# Batch Normalization 算法的 decay 参数
conv_bn_decay = 0.9999
# Batch Normalization 算法的 variance_epsilon 参数
conv_bn_epsilon = 0.001
# Jaccard相似度判断阀值
jaccard_value = 0.55
feature_maps_shape=[]
all_default_boxs_len=0
all_default_boxs=[]
jitter = 0.2
def get_traindata_voc(batch_size):
def get_actual_data_from_xml(xml_path):
actual_item = []
try:
annotation_node = etxml.parse(xml_path).getroot()
img_width = float(annotation_node.find('size').find('width').text.strip())
img_height = float(annotation_node.find('size').find('height').text.strip())
object_node_list = annotation_node.findall('object')
for obj_node in object_node_list:
lable = lable_arr.index(obj_node.find('name').text.strip())
bndbox = obj_node.find('bndbox')
x_min = float(bndbox.find('xmin').text.strip())
y_min = float(bndbox.find('ymin').text.strip())
x_max = float(bndbox.find('xmax').text.strip())
y_max = float(bndbox.find('ymax').text.strip())
# 位置数据用比例来表示，格式[center_x,center_y,width,height,lable]
actual_item.append([((x_min + x_max) / 2 / img_width), ((y_min + y_max) / 2 / img_height),
((x_max - x_min) / img_width), ((y_max - y_min) / img_height), lable])
return actual_item
except:
return None
train_data = []
actual_data = []
file_list = random.sample(file_name_list, batch_size)
for f_name in file_list:
img_path = './train_datasets/voc2012/JPEGImages/' + f_name
xml_path = './train_datasets/voc2012/Annotations/' + f_name.replace('.jpg', '.xml')
if os.path.splitext(img_path)[1].lower() == '.jpg':
actual_item = get_actual_data_from_xml(xml_path)
img = skimage.io.imread(img_path)
if actual_item != None:
countwhile=0
while True:
clas=[]
coords=[]
for x in actual_item:
clas.append(x[4])
coords.append([x[0],x[1],x[2],x[3]])
tmp0 = random.randint(-30, 50)
tmp1 = random.randint(-30, 50)
imgr=img.copy()
scale = np.max((400 / float(img.shape[1]),
400 / float(img.shape[0])))
im, coords = tl.prepro.obj_box_imresize(imgr, coords,
[int(img.shape[0] * scale) + tmp0, int(img.shape[1] * scale) + tmp1],
is_rescale=True, interp='bicubic')
# print(im.shape)
# print(coords)
for wi in range(7):
imt, clast, coordst = tl.prepro.obj_box_zoom(im, clas, coords, zoom_range=(1.0, 2.2),
fill_mode='nearest',
order=1, is_rescale=True, is_center=True,
is_random=True,
thresh_wh=0.04, thresh_wh2=8.0)
# print(im.shape)
if clast!=[]:
im=imt
clas= clast
coords =coordst
break
if wi>=6:
im, clas, coords = tl.prepro.obj_box_zoom(im, clas, coords, zoom_range=(0.7, 1.2),
fill_mode='nearest',
order=1, is_rescale=True, is_center=True,
is_random=True,
thresh_wh=0.05, thresh_wh2=8.0)
im, coords = tl.prepro.obj_box_left_right_flip(im,
coords, is_rescale=True, is_center=True, is_random=True)
# print(coords)
for wi in range(8):
imt, clast, coordst = tl.prepro.obj_box_crop(im, clas, coords,
wrg=300, hrg=300,
is_rescale=True, is_center=True, is_random=True,
thresh_wh=0.07, thresh_wh2=7.0)
if clast!=[]:
im=imt
clas= clast
coords =coordst
break
if wi==7:
im, clas, coords = tl.prepro.obj_box_crop(im, clas, coords,
wrg=300, hrg=300,
is_rescale=True, is_center=True,
is_random=True,
thresh_wh=0.07, thresh_wh2=8.0)
im = tl.prepro.illumination(im, gamma=(0.2, 1.2),
contrast=(0.2, 1.2), saturation=(0.2, 1.2), is_random=True)
im = tl.prepro.adjust_hue(im, hout=0.1, is_offset=True,
is_clip=True, is_random=True)
im = im / 127.5 - 1.
aitems = []
if clas!=[]:
for x in range(len(clas)):
aitem=[coords[x][0],coords[x][1],coords[x][2],coords[x][3],clas[x]]
aitems.append(aitem)
actual_data.append(aitems)
train_data.append(im)
break
countwhile+=1
if countwhile>=4:
clas = []
coords = []
for x in actual_item:
clas.append(x[4])
coords.append([x[0], x[1], x[2], x[3]])
tmp0 = random.randint(1, 30)
tmp1 = random.randint(1, 30)
imgr = img.copy()
im, coords = tl.prepro.obj_box_imresize(imgr, coords,
[300 + tmp0,
300 + tmp1],
is_rescale=True, interp='bicubic')
im, coords = tl.prepro.obj_box_left_right_flip(im,
coords, is_rescale=True, is_center=True,
is_random=True)
im, clas, coords = tl.prepro.obj_box_crop(im, clas, coords,
wrg=300, hrg=300,
is_rescale=True, is_center=True,
is_random=True,
thresh_wh=0.02, thresh_wh2=10.0)
im = tl.prepro.illumination(im, gamma=(0.8, 1.2),
contrast=(0.8, 1.2), saturation=(0.8, 1.2), is_random=True)
im = tl.prepro.pixel_value_scale(im, 0.1, [0, 255], is_random=True)
im = im / 127.5 - 1.
aitems = []
if len(clas) != 0:
for x in range(len(clas)):
aitem = [coords[x][0], coords[x][1], coords[x][2], coords[x][3], clas[x]]
aitems.append(aitem)
actual_data.append(aitems)
train_data.append(im)
break
else:
print('Error : ' + xml_path)
continue
return train_data, actual_data, file_list
def generate_groundtruth_data(input_actual_data):
# 生成空数组，用于保存groundtruth
input_actual_data_len = len(input_actual_data)
gt_class = np.zeros((input_actual_data_len, all_default_boxs_len))
gt_location = np.zeros((input_actual_data_len, all_default_boxs_len, 4))
gt_positives_jacc = np.zeros((input_actual_data_len, all_default_boxs_len))
gt_positives = np.zeros((input_actual_data_len, all_default_boxs_len))
gt_negatives = np.zeros((input_actual_data_len, all_default_boxs_len))
background_jacc = max(0, (jaccard_value - 0.2))
# 初始化正例训练数据
for img_index in range(input_actual_data_len):
for pre_actual in input_actual_data[img_index]:
gt_class_val = pre_actual[-1:][0]
if gt_class_val>20 or gt_class_val<0:
gt_class_val=0
gt_box_val = pre_actual[:-1]
for boxe_index in range(all_default_boxs_len):
jacc,gt_box_val_loc = jaccard(gt_box_val, all_default_boxs[boxe_index])
if jacc > jaccard_value or jacc == jaccard_value:
gt_class[img_index][boxe_index] = gt_class_val
gt_location[img_index][boxe_index] = gt_box_val_loc
gt_positives_jacc[img_index][boxe_index] = jacc
gt_positives[img_index][boxe_index] = 1
gt_negatives[img_index][boxe_index] = 0
# 如果没有正例，则随机创建一个正例，预防nan
if np.sum(gt_positives[img_index]) == 0:
# print('【没有匹配jacc】:'+str(input_actual_data[img_index]))
random_pos_index = np.random.randint(low=0, high=all_default_boxs_len, size=1)[0]
gt_class[img_index][random_pos_index] = background_classes_val
gt_location[img_index][random_pos_index] = [0.00001, 0.00001, 0.00001, 0.00001]
gt_positives_jacc[img_index][random_pos_index] = jaccard_value
gt_positives[img_index][random_pos_index] = 1
gt_negatives[img_index][random_pos_index] = 0
gt_neg_end_count = int(np.sum(gt_positives[img_index]) * 3)
if (gt_neg_end_count + np.sum(gt_positives[img_index])) > all_default_boxs_len:
gt_neg_end_count = all_default_boxs_len - np.sum(gt_positives[img_index])
gt_neg_index = np.random.randint(low=0, high=all_default_boxs_len, size=gt_neg_end_count)
for r_index in gt_neg_index:
if gt_positives_jacc[img_index][r_index] < background_jacc and gt_positives[img_index][r_index] != 1:
gt_class[img_index][r_index] = background_classes_val
gt_positives[img_index][r_index] = 0
gt_negatives[img_index][r_index] = 1
gt_class = check_numerics(gt_class, 'gt_class')
gt_location = check_numerics(gt_location, 'gt_class')
gt_positives = check_numerics(gt_positives, 'gt_positives')
gt_negatives = check_numerics(gt_negatives, 'gt_negatives')
return gt_class, gt_location, gt_positives, gt_negatives
def jaccard(rect1, rect2):
x_overlap = max(0, (min(rect1[0] + (rect1[2] / 2), rect2[0] + (rect2[2] / 2)) - max(rect1[0] - (rect1[2] / 2),
rect2[0] - (rect2[2] / 2))))
y_overlap = max(0, (min(rect1[1] + (rect1[3] / 2), rect2[1] + (rect2[3] / 2)) - max(rect1[1] - (rect1[3] / 2),
rect2[1] - (rect2[3] / 2))))
intersection = x_overlap * y_overlap
# 删除超出图像大小的部分
rect1_width_sub = 0
rect1_height_sub = 0
rect2_width_sub = 0
rect2_height_sub = 0
if (rect1[0] - rect1[2] / 2) < 0: rect1_width_sub += 0 - (rect1[0] - rect1[2] / 2)
if (rect1[0] + rect1[2] / 2) > 1: rect1_width_sub += (rect1[0] + rect1[2] / 2) - 1
if (rect1[1] - rect1[3] / 2) < 0: rect1_height_sub += 0 - (rect1[1] - rect1[3] / 2)
if (rect1[1] + rect1[3] / 2) > 1: rect1_height_sub += (rect1[1] + rect1[3] / 2) - 1
if (rect2[0] - rect2[2] / 2) < 0: rect2_width_sub += 0 - (rect2[0] - rect2[2] / 2)
if (rect2[0] + rect2[2] / 2) > 1: rect2_width_sub += (rect2[0] + rect2[2] / 2) - 1
if (rect2[1] - rect2[3] / 2) < 0: rect2_height_sub += 0 - (rect2[1] - rect2[3] / 2)
if (rect2[1] + rect2[3] / 2) > 1: rect2_height_sub += (rect2[1] + rect2[3] / 2) - 1
area_box_a = (rect1[2] - rect1_width_sub) * (rect1[3] - rect1_height_sub)
area_box_b = (rect2[2] - rect2_width_sub) * (rect2[3] - rect2_height_sub)
union = area_box_a + area_box_b - intersection
if intersection > 0 and union > 0:
return intersection / union,[(rect1[0]-(rect2[0]))/rect2[2],(rect1[1]-(rect2[1]))/rect2[3],math.log(rect1[2]/rect2[2]),math.log(rect1[3]/rect2[3])]
else:
return 0,[0.00001,0.00001,0.00001,0.00001]
def denseblock(input,blocknum=1,step=48,firstchannel=192,is_train=True,name='denseblock',reuse=None):
with tf.variable_scope(name, reuse=reuse):
tl.layers.set_name_reuse(reuse)
nettemp=LambdaLayer(input, lambda x: tf.identity(x), name="INPUTS")
for x in range(blocknum):
netbn = BatchNormLayer(nettemp, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='bn/' + str(x))
net=Conv2d(netbn, firstchannel, (1, 1), (1, 1), padding='SAME',name='neta/'+str(x))
netbn = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name=name + 'bn2/' + str(x))
net=Conv2d(netbn, step, (3, 3), (1, 1), padding='SAME',name='netb/'+str(x))
nettemp= ConcatLayer([nettemp,net], -1,name='concattemp/'+str(x))
net = nettemp
return net
def denseblockpl(input,step=256,firstchannel=256,is_train=True,name='densepl',reuse=None):
with tf.variable_scope(name, reuse=reuse):
tl.layers.set_name_reuse(reuse)
input = LambdaLayer(input, lambda x: tf.identity(x), name="INPUTS")
netbn2=MaxPool2d(input,(2,2),(2,2),padding='SAME', name='bnpool2')
netbn2 = BatchNormLayer(netbn2, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name=name + 'bn2pl' )
netbn2 = Conv2d(netbn2, firstchannel, (1, 1), (1, 1), padding='SAME', name='bnconv2' )
netbn = BatchNormLayer(input, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name= 'bn' )
net=Conv2d(netbn, firstchannel, (1, 1), (1, 1), padding='SAME',name='neta')
netbn = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='bn2')
net=Conv2d(netbn, step, (3, 3), (2, 2), padding='SAME',name='netb')
nettemp = ConcatLayer([net,netbn2], -1,name='concat')
return nettemp
def denseblockfin(input,step=256,firstchannel=256,is_train=True,name='densepl',reuse=None):
with tf.variable_scope(name, reuse=reuse):
tl.layers.set_name_reuse(reuse)
input = LambdaLayer(input, lambda x: tf.identity(x), name="INPUTS")
netbn2=MaxPool2d(input,(3,3),(1,1),padding='VALID', name='bnpool2')
netbn2 = BatchNormLayer(netbn2, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name=name + 'bn2pl' )
netbn2 = Conv2d(netbn2, firstchannel, (1, 1), (1, 1), padding='SAME', name='bnconv2' )
netbn = BatchNormLayer(input, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name= 'bn' )
net=Conv2d(netbn, firstchannel, (1, 1), (1, 1), padding='SAME',name='neta')
netbn = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='bn2')
net=Conv2d(netbn, step, (3, 3), (1, 1), padding='VALID',name='netb')
nettemp = ConcatLayer([net,netbn2], -1,name='concat')
return nettemp
def inference(inputs, is_train, reuse):
W_init = tf.contrib.layers.xavier_initializer()
with tf.variable_scope("model", reuse=reuse):
tl.layers.set_name_reuse(reuse)
net = InputLayer(inputs, name='input')
net = Conv2d(net, 64, (3, 3), (2, 2), padding='SAME',
W_init=W_init, name='stem1')
net = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='stem1_bn')
net = Conv2d(net, 64, (3, 3), (1, 1), padding='SAME',
W_init=W_init, name='stem2')
net = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='stem2_bn')
net = Conv2d(net, 128, (3, 3), (1, 1), padding='SAME',
W_init=W_init, name='stem3')
net = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='stem3_bn')
net = MaxPool2d(net, filter_size=(2, 2), strides=(2, 2), name='stem3_pool')
net = denseblock(net, blocknum=6, step=48, firstchannel=192, is_train=is_train, name='denseblock0', reuse=reuse)
net = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='denseblock0_bn')
net = Conv2d(net, 416, (1, 1), (1, 1), padding='SAME',
W_init=W_init, name='denseblock0_cnn')
net = MaxPool2d(net, filter_size=(2, 2), strides=(2, 2), name='denseblock0_pool')
net = denseblock(net, blocknum=8, step=48, firstchannel=192, is_train=is_train, name='denseblock1', reuse=reuse)
net = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='denseblock1_bn')
net = Conv2d(net, 800, (1, 1), (1, 1), padding='SAME',
W_init=W_init, name='denseblock1_cnn')
netfirst=BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='feature_first_bn')
net = MaxPool2d(net, filter_size=(2, 2), strides=(2, 2), name='denseblock2_pool1')
net = denseblock(net, blocknum=8, step=48, firstchannel=192, is_train=is_train, name='denseblock2', reuse=reuse)
net = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='denseblock2_bn')
net = Conv2d(net, 1184, (1, 1), (1, 1), padding='SAME',
W_init=W_init, name='denseblock2_cnn')
net = denseblock(net, blocknum=8, step=48, firstchannel=192, is_train=is_train, name='denseblock3', reuse=reuse)
net = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='denseblock3_bn')
net = Conv2d(net, 256, (1, 1), (1, 1), padding='SAME',
W_init=W_init, name='denseblock2_cnna')
netpl=MaxPool2d(netfirst, filter_size=(2, 2), strides=(2, 2), name='First_pool')
netpl=BatchNormLayer(netpl, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='First_bn')
netpl = Conv2d(netpl, 256, (1, 1), (1, 1), padding='SAME',
W_init=W_init, name='denseblock2_cnnb')
net=ConcatLayer([net,netpl],-1,"Second_Cat")
netsecond = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu, name='feature_second_bn')
net = denseblockpl(net, step=256, firstchannel=256, is_train=is_train, name='denseplz1', reuse=reuse)
netthird = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu,
name='feature_third_bn')
net = denseblockpl(net, step=128, firstchannel=128, is_train=is_train, name='denseplz2', reuse=reuse)
netfourth = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu,
name='feature_fourth_bn')
net = denseblockpl(net, step=128, firstchannel=128, is_train=is_train, name='denseplz3', reuse=reuse)
netfifth = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu,
name='feature_fifth_bn')
net = denseblockfin(net, step=128, firstchannel=128, is_train=is_train, name='denseplz4', reuse=reuse)
netsixth = BatchNormLayer(net, is_train=is_train, decay=conv_bn_decay, act=tf.nn.relu,
name='feature_sixth_bn')
outfirst=Conv2d(netfirst, default_box_size[0] * (classes_size + 4), (3, 3), (1, 1), padding='SAME',
W_init=W_init, name='firstout')
outsecond=Conv2d(netsecond, default_box_size[1] * (classes_size + 4), (3, 3), (1, 1), padding='SAME',
W_init=W_init, name='secondout')
outthird=Conv2d(netthird, default_box_size[2] * (classes_size + 4), (3, 3), (1, 1), padding='SAME',
W_init=W_init, name='thirdout')
outfourth=Conv2d(netfourth, default_box_size[3] * (classes_size + 4), (3, 3), (1, 1), padding='SAME',
W_init=W_init, name='fourthout')
outfifth=Conv2d(netfifth, default_box_size[4] * (classes_size + 4), (3, 3), (1, 1), padding='SAME',
W_init=W_init, name='fifthout')
outsixth=Conv2d(netsixth, default_box_size[5] * (classes_size + 4), (3, 3), (1, 1), padding='SAME',
W_init=W_init, name='sixthout')
features1=outfirst.outputs
features2=outsecond.outputs
features3=outthird.outputs
features4=outfourth.outputs
features5=outfifth.outputs
features6=outsixth.outputs
feature_maps = [features1, features2, features3, features4, features5,features6]
global feature_maps_shape
feature_maps_shape = [m.get_shape().as_list() for m in feature_maps]
tmp_all_feature = []
for i, fmap in zip(range(len(feature_maps)), feature_maps):
width = feature_maps_shape[i][1]
height = feature_maps_shape[i][2]
tmp_all_feature.append(
tf.reshape(fmap, [-1, (width * height * default_box_size[i]), (classes_size + 4)]))
tmp_all_feature = tf.concat(tmp_all_feature, axis=1)
feature_class = tmp_all_feature[:, :, :classes_size]
feature_location = tmp_all_feature[:, :, classes_size:]
print('## feature_class shape : ' + str(feature_class.get_shape().as_list()))
print('## feature_location shape : ' + str(feature_location.get_shape().as_list()))
# 生成所有default boxs
global all_default_boxs
all_default_boxs = generate_all_default_boxs()
# print(all_default_boxs)
global all_default_boxs_len
all_default_boxs_len = len(all_default_boxs)
print('## all default boxs : ' + str(all_default_boxs_len))
return feature_class,feature_location,all_default_boxs,all_default_boxs_len
def smooth_L1(x):
return tf.where(tf.less_equal(tf.abs(x), 1.0), tf.multiply(0.5, tf.pow(x, 2.0)), tf.subtract(tf.abs(x), 0.5))
def elloss(feature_class,feature_location,groundtruth_class,groundtruth_location,groundtruth_positives,groundtruth_count):
softmax_cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(logits=feature_class,
labels=groundtruth_class)
loss_location = tf.div(tf.reduce_sum(tf.multiply(
tf.reduce_sum(smooth_L1(tf.subtract(groundtruth_location, feature_location)),
reduction_indices=2), groundtruth_positives), reduction_indices=1),
tf.reduce_sum(groundtruth_positives, reduction_indices=1))
loss_class = tf.div(
tf.reduce_sum(tf.multiply(softmax_cross_entropy, groundtruth_count), reduction_indices=1),
tf.reduce_sum(groundtruth_count, reduction_indices=1))
loss_all = tf.reduce_sum(tf.add(loss_class, loss_location*5))
return loss_all,loss_class,loss_location
def generate_all_default_boxs():
all_default_boxes = []
for index, map_shape in zip(range(len(feature_maps_shape)), feature_maps_shape):
width = int(map_shape[1])
height = int(map_shape[2])
cell_scale = default_box_scale[index]
for x in range(width):
for y in range(height):
for ratio in box_aspect_ratio[index]:
center_x = (x / float(width)) + (0.5 / float(width))
center_y = (y / float(height)) + (0.5 / float(height))
box_width = cell_scale*np.sqrt(ratio)/1.2
box_height = cell_scale/np.sqrt(ratio)/1.2
all_default_boxes.append([center_x, center_y, box_width, box_height])
all_default_boxes.append([(x / float(width)) + (0.5 / float(width)), (y / float(height)) + (0.5 / float(height)), cell_scale*1.5,cell_scale*1.4])
all_default_boxes = np.array(all_default_boxes)
all_default_boxes = check_numerics(all_default_boxes, 'all_default_boxes')
return all_default_boxes
def check_numerics(input_dataset, message):
if str(input_dataset).find('Tensor') == 0:
input_dataset = tf.check_numerics(input_dataset, message)
else:
dataset = np.array(input_dataset)
nan_count = np.count_nonzero(dataset != dataset)
inf_count = len(dataset[dataset == float("inf")])
n_inf_count = len(dataset[dataset == float("-inf")])
if nan_count > 0 or inf_count > 0 or n_inf_count > 0:
data_error = '【' + message + '】出现数据错误！【nan：' + str(nan_count) + '|inf：' + str(
inf_count) + '|-inf：' + str(n_inf_count) + '】'
raise Exception(data_error)
return input_dataset
if __name__ == '__main__':
imageinput=tf.placeholder(tf.float32,[None,300,300,3],"inputsimage")
imageinputtest = tf.placeholder(tf.float32, [None, 300, 300, 3], "inputsimage")
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.9)
fc, fl, _, _ = inference(imageinput, True, None)
fc2, fl2, _, _ = inference(imageinputtest, False, True)
groundtruth_class = tf.placeholder(shape=[None, all_default_boxs_len], dtype=tf.int32,
name='groundtruth_class')
groundtruth_location = tf.placeholder(shape=[None, all_default_boxs_len, 4], dtype=tf.float32,
name='groundtruth_location')
groundtruth_positives = tf.placeholder(shape=[None, all_default_boxs_len], dtype=tf.float32,
name='groundtruth_positives')
groundtruth_negatives = tf.placeholder(shape=[None, all_default_boxs_len], dtype=tf.float32,
name='groundtruth_negatives')
groundtruth_count = tf.add(groundtruth_positives, groundtruth_negatives)
learning_rt=0.000001
learning_rate = tf.placeholder(tf.float32, None, 'learning_rate')
loss_allt, loss_classt, loss_locationt = elloss(fc, fl, groundtruth_class, groundtruth_location, groundtruth_positives, groundtruth_count)
train = tf.train.MomentumOptimizer(learning_rate,momentum=0.9).minimize(loss_allt)
tf.summary.scalar('loss_all_train', loss_allt)
tf.summary.scalar('loss_class_train', tf.reduce_sum(loss_classt) )
tf.summary.scalar('loss_location_train', tf.reduce_sum(loss_locationt))
merged = tf.summary.merge_all()
with tf.Session(config=tf.ConfigProto(gpu_options=gpu_options)) as sess:
trainwrite = tf.summary.FileWriter("logs/", sess.graph)
sess.run(tf.global_variables_initializer())
saver2 = tf.train.Saver(var_list=tf.trainable_variables())
zzz = variables._all_saveable_objects().copy()
print(zzz)
saver = tf.train.Saver()
if os.path.exists('./session_paramsdddaleasy/session2.ckpt.index') :
print('\nStart Restore')
saver2.restore(sess, './session_paramsdddaleasy/session2.ckpt')
print('\nEnd Restore')
print('\nStart Training')
min_loss_location = 100000.
min_loss_class = 100000.
avg_loss=0
avg_lossloc=0
avg_losclass=0
ptlos=0
ptlosc=0
ptlosl=0
while((min_loss_location + min_loss_class) > 0.001 and running_count < 100000):
running_count += 1
train_data, actual_data, _ = get_traindata_voc(batch_size)
starttime = time.time()
gt_class, gt_location, gt_positives, gt_negatives=generate_groundtruth_data(actual_data)
if len(train_data) > 0:
loss_all,loss_class,loss_location,_,pred_class,pred_location = sess.run([loss_allt, loss_classt, loss_locationt,train,fc, fl],feed_dict={imageinput:train_data,groundtruth_class:gt_class,groundtruth_location:gt_location,groundtruth_positives:gt_positives,groundtruth_negatives:gt_negatives,learning_rate:learning_rt})
l = np.sum(loss_location)
c = np.sum(loss_class)
avg_loss +=loss_all
avg_lossloc += loss_class
avg_losclass += loss_location
if min_loss_location > l:
min_loss_location = l
if min_loss_class > c:
min_loss_class = c
print('Running:【' + str(running_count) + '】|Loss All:【' + str(
min_loss_location + min_loss_class) + '|' + str(loss_all) + '】|Location:【' + str(
np.sum(loss_location)) + '】|Class:【' + str(np.sum(loss_class)) + '】|pred_class:【' + str(
np.sum(pred_class)) + '|' + str(np.amax(pred_class)) + '|' + str(
np.min(pred_class)) + '】|pred_location:【' + str(np.sum(pred_location)) + '|' + str(
np.amax(pred_location)) + '|' + str(np.min(pred_location)) + '】TIME:'+str(time.time()-starttime))
if running_count % 100 == 0:
print('---------')
print('avgloss')
print(avg_loss/100.)
print(np.sum(avg_lossloc/100.) )
print(np.sum(avg_losclass/100.) )
print(ptlos-avg_loss/100.)
print(ptlosc-np.sum(avg_lossloc/100.) )
print(ptlosl-np.sum(avg_losclass/100.) )
ptlos = avg_loss/100.
ptlosc = np.sum(avg_lossloc/100. )
ptlosl = np.sum(avg_losclass/100. )
print('---------')
avg_loss=0
avg_lossloc = 0
avg_losclass = 0
if running_count % 100 == 0:
results = sess.run(merged,feed_dict={imageinput:train_data,groundtruth_class:gt_class,groundtruth_location:gt_location,groundtruth_positives:gt_positives,groundtruth_negatives:gt_negatives,learning_rate:learning_rt})
trainwrite.add_summary(results, running_count)
if running_count % 500 == 0:
saver.save(sess, './session_paramsdddaleasy/session.ckpt')
print('session.ckpt has been saved.')
gc.collect()
else:
print('No Data Exists!')
break