forked from soeaver/caffe-model
-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathinception_v3.py
412 lines (355 loc) · 29.4 KB
/
inception_v3.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
import caffe
from caffe import layers as L
from caffe import params as P
def fc_relu_drop(bottom, num_output=1024, dropout_ratio=0.5):
fc = L.InnerProduct(bottom, num_output=num_output,
param=[dict(lr_mult=1, decay_mult=1), dict(lr_mult=2, decay_mult=0)],
weight_filler=dict(type='xavier', std=1),
bias_filler=dict(type='constant', value=0.2))
relu = L.ReLU(fc, in_place=True)
drop = L.Dropout(fc, in_place=True,
dropout_param=dict(dropout_ratio=dropout_ratio))
return fc, relu, drop
def factorization_conv_bn_scale_relu(bottom, num_output=64, kernel_size=3, stride=1, pad=0):
conv = L.Convolution(bottom, num_output=num_output, kernel_size=kernel_size, stride=stride, pad=pad,
param=[dict(lr_mult=1, decay_mult=1), dict(lr_mult=2, decay_mult=0)],
weight_filler=dict(type='xavier', std=0.01),
bias_filler=dict(type='constant', value=0))
conv_bn = L.BatchNorm(conv, use_global_stats=False, in_place=True)
conv_scale = L.Scale(conv, scale_param=dict(bias_term=True), in_place=True)
conv_relu = L.ReLU(conv, in_place=True)
return conv, conv_bn, conv_scale, conv_relu
def factorization_conv_mxn(bottom, num_output=64, kernel_h=1, kernel_w=7, stride=1, pad_h=3, pad_w=0):
conv_mxn = L.Convolution(bottom, num_output=num_output, kernel_h=kernel_h, kernel_w=kernel_w, stride=stride,
pad_h=pad_h, pad_w=pad_w,
param=[dict(lr_mult=1, decay_mult=1), dict(lr_mult=2, decay_mult=0)],
weight_filler=dict(type='xavier', std=0.01),
bias_filler=dict(type='constant', value=0.2))
conv_mxn_bn = L.BatchNorm(conv_mxn, use_global_stats=False, in_place=True)
conv_mxn_scale = L.Scale(conv_mxn, scale_param=dict(bias_term=True), in_place=True)
conv_mxn_relu = L.ReLU(conv_mxn, in_place=True)
return conv_mxn, conv_mxn_bn, conv_mxn_scale, conv_mxn_relu
def stem_v3_299x299(bottom):
"""
input:3x299x299
output:192x35x35
:param bottom: bottom layer
:return: layers
"""
conv1_3x3_s2, conv1_3x3_s2_bn, conv1_3x3_s2_scale, conv1_3x3_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=32, kernel_size=3, stride=2) # 32x149x149
conv2_3x3_s1, conv2_3x3_s1_bn, conv2_3x3_s1_scale, conv2_3x3_s1_relu = \
factorization_conv_bn_scale_relu(conv1_3x3_s2, num_output=32, kernel_size=3) # 32x147x147
conv3_3x3_s1, conv3_3x3_s1_bn, conv3_3x3_s1_scale, conv3_3x3_s1_relu = \
factorization_conv_bn_scale_relu(conv2_3x3_s1, num_output=64, kernel_size=3, pad=1) # 64x147x147
pool1_3x3_s2 = L.Pooling(conv3_3x3_s1, kernel_size=3, stride=2, pool=P.Pooling.MAX) # 64x73x73
conv4_3x3_reduce, conv4_3x3_reduce_bn, conv4_3x3_reduce_scale, conv4_3x3_reduce_relu = \
factorization_conv_bn_scale_relu(pool1_3x3_s2, num_output=80, kernel_size=1) # 80x73x73
conv4_3x3, conv4_3x3_bn, conv4_3x3_scale, conv4_3x3_relu = \
factorization_conv_bn_scale_relu(conv4_3x3_reduce, num_output=192, kernel_size=3) # 192x71x71
pool2_3x3_s2 = L.Pooling(conv4_3x3, kernel_size=3, stride=2, pool=P.Pooling.MAX) # 192x35x35
return conv1_3x3_s2, conv1_3x3_s2_bn, conv1_3x3_s2_scale, conv1_3x3_relu, conv2_3x3_s1, conv2_3x3_s1_bn, \
conv2_3x3_s1_scale, conv2_3x3_s1_relu, conv3_3x3_s1, conv3_3x3_s1_bn, conv3_3x3_s1_scale, conv3_3x3_s1_relu, \
pool1_3x3_s2, conv4_3x3_reduce, conv4_3x3_reduce_bn, conv4_3x3_reduce_scale, conv4_3x3_reduce_relu, \
conv4_3x3, conv4_3x3_bn, conv4_3x3_scale, conv4_3x3_relu, pool2_3x3_s2
def inception_v3_a(bottom, pool_proj_num_output=32):
"""
input:192or256or288x35x35
output:256or288x35x35
:param pool_proj_num_output: num_output of pool_proj
:param bottom: bottom layer
:return: layers
"""
conv_1x1, conv_1x1_bn, conv_1x1_scale, conv_1x1_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=64, kernel_size=1) # 64x35x35
conv_5x5_reduce, conv_5x5_reduce_bn, conv_5x5_reduce_scale, conv_5x5_reduce_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=48, kernel_size=1) # 48x35x35
conv_5x5, conv_5x5_bn, conv_5x5_scale, conv_5x5_relu = \
factorization_conv_bn_scale_relu(conv_5x5_reduce, num_output=64, kernel_size=5, pad=2) # 64x35x35
conv_3x3_reduce, conv_3x3_reduce_bn, conv_3x3_reduce_scale, conv_3x3_reduce_relu = \
factorization_conv_bn_scale_relu(bottom, kernel_size=1, num_output=64) # 64x35x35
conv_3x3, conv_3x3_bn, conv_3x3_scale, conv_3x3_relu = \
factorization_conv_bn_scale_relu(conv_3x3_reduce, kernel_size=3, num_output=96, pad=1) # 96x35x35
conv_3x3_2, conv_3x3_2_bn, conv_3x3_2_scale, conv_3x3_2_relu = \
factorization_conv_bn_scale_relu(conv_3x3, kernel_size=3, num_output=96, pad=1) # 96x35x35
pool = L.Pooling(bottom, kernel_size=3, stride=1, pad=1, pool=P.Pooling.AVE) # 192x35x35
pool_proj, pool_proj_bn, pool_proj_scale, pool_proj_relu = \
factorization_conv_bn_scale_relu(pool, kernel_size=1, num_output=pool_proj_num_output) # 32x35x35
concat = L.Concat(conv_1x1, conv_5x5, conv_3x3_2, pool_proj) # 256or288(64+64+96+32or64)x35x35
return conv_1x1, conv_1x1_bn, conv_1x1_scale, conv_1x1_relu, conv_5x5_reduce, conv_5x5_reduce_bn, \
conv_5x5_reduce_scale, conv_5x5_reduce_relu, conv_5x5, conv_5x5_bn, conv_5x5_scale, conv_5x5_relu, \
conv_3x3_reduce, conv_3x3_reduce_bn, conv_3x3_reduce_scale, conv_3x3_reduce_relu, conv_3x3, conv_3x3_bn, \
conv_3x3_scale, conv_3x3_relu, conv_3x3_2, conv_3x3_2_bn, conv_3x3_2_scale, conv_3x3_2_relu, pool, \
pool_proj, pool_proj_bn, pool_proj_scale, pool_proj_relu, concat
def reduction_v3_a(bottom):
"""
input:288x35x35
output:768x17x17
:param bottom: bottom layer
:return: layers
"""
pool = L.Pooling(bottom, kernel_size=3, stride=2, pool=P.Pooling.MAX) # 384x17x17
conv_3x3, conv_3x3_bn, conv_3x3_scale, conv_3x3_relu = \
factorization_conv_bn_scale_relu(bottom, kernel_size=3, num_output=384, stride=2) # 384x17x17
conv_3x3_2_reduce, conv_3x3_2_reduce_bn, conv_3x3_2_reduce_scale, conv_3x3_2_reduce_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=64, kernel_size=1) # 64x35x35
conv_3x3_2, conv_3x3_2_bn, conv_3x3_2_scale, conv_3x3_2_relu = \
factorization_conv_bn_scale_relu(conv_3x3_2_reduce, num_output=96, kernel_size=3, pad=1) # 96x35x35
conv_3x3_3, conv_3x3_3_bn, conv_3x3_3_scale, conv_3x3_3_relu = \
factorization_conv_bn_scale_relu(conv_3x3_2, num_output=96, kernel_size=3, stride=2) # 96x17x17
concat = L.Concat(pool, conv_3x3, conv_3x3_3) # 768(288+384+96)x17x17
return pool, conv_3x3, conv_3x3_bn, conv_3x3_scale, conv_3x3_relu, conv_3x3_2_reduce, conv_3x3_2_reduce_bn, \
conv_3x3_2_reduce_scale, conv_3x3_2_reduce_relu, conv_3x3_2, conv_3x3_2_bn, conv_3x3_2_scale, \
conv_3x3_2_relu, conv_3x3_3, conv_3x3_3_bn, conv_3x3_3_scale, conv_3x3_3_relu, concat
def inception_v3_b(bottom, outs=128):
"""
input:768x17x17
output:768x17x17
:param outs: num_outputs
:param bottom: bottom layer
:return: layers
"""
pool_ave = L.Pooling(bottom, kernel_size=3, stride=1, pad=1, pool=P.Pooling.AVE) # 768x17x17
conv_1x1, conv_1x1_bn, conv_1x1_scale, conv_1x1_relu = \
factorization_conv_bn_scale_relu(pool_ave, num_output=192, kernel_size=1) # 192x17x17
conv_1x1_2, conv_1x1_2_bn, conv_1x1_2_scale, conv_1x1_2_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=192, kernel_size=1) # 192x17x17
conv_1x7_reduce, conv_1x7_reduce_bn, conv_1x7_reduce_scale, conv_1x7_reduce_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=outs, kernel_size=1) # outsx17x17
conv_1x7, conv_1x7_bn, conv_1x7_scale, conv_1x7_relu = \
factorization_conv_mxn(conv_1x7_reduce, num_output=outs, kernel_h=1, kernel_w=7, pad_h=0, pad_w=3) # outsx17x17
conv_7x1, conv_7x1_bn, conv_7x1_scale, conv_7x1_relu = \
factorization_conv_mxn(conv_1x7, num_output=192, kernel_h=7, kernel_w=1, pad_h=3, pad_w=0) # 192x17x17
conv_7x1_reduce, conv_7x1_reduce_bn, conv_7x1_reduce_scale, conv_7x1_reduce_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=outs, kernel_size=1) # outsx17x17
conv_7x1_2, conv_7x1_2_bn, conv_7x1_2_scale, conv_7x1_2_relu = \
factorization_conv_mxn(conv_7x1_reduce, num_output=outs, kernel_h=7, kernel_w=1, pad_h=3, pad_w=0) # outsx17x17
conv_1x7_2, conv_1x7_2_bn, conv_1x7_2_scale, conv_1x7_2_relu = \
factorization_conv_mxn(conv_7x1_2, num_output=outs, kernel_h=1, kernel_w=7, pad_h=0, pad_w=3) # outsx17x17
conv_7x1_3, conv_7x1_3_bn, conv_7x1_3_scale, conv_7x1_3_relu = \
factorization_conv_mxn(conv_1x7_2, num_output=outs, kernel_h=7, kernel_w=1, pad_h=3, pad_w=0) # outsx17x17
conv_1x7_3, conv_1x7_3_bn, conv_1x7_3_scale, conv_1x7_3_relu = \
factorization_conv_mxn(conv_7x1_3, num_output=192, kernel_h=1, kernel_w=7, pad_h=0, pad_w=3) # 192x17x17
concat = L.Concat(conv_1x1_2, conv_7x1, conv_1x7_3, conv_1x1) # 768(192+192+192+192)x17x17
return pool_ave, conv_1x1, conv_1x1_bn, conv_1x1_scale, conv_1x1_relu, conv_1x1_2, conv_1x1_2_bn, conv_1x1_2_scale, \
conv_1x1_2_relu, conv_1x7_reduce, conv_1x7_reduce_bn, conv_1x7_reduce_scale, conv_1x7_reduce_relu, \
conv_1x7, conv_1x7_bn, conv_1x7_scale, conv_1x7_relu, conv_7x1, conv_7x1_bn, conv_7x1_scale, conv_7x1_relu, \
conv_7x1_reduce, conv_7x1_reduce_bn, conv_7x1_reduce_scale, conv_7x1_reduce_relu, conv_7x1_2, conv_7x1_2_bn, \
conv_7x1_2_scale, conv_7x1_2_relu, conv_1x7_2, conv_1x7_2_bn, conv_1x7_2_scale, conv_1x7_2_relu, conv_7x1_3, \
conv_7x1_3_bn, conv_7x1_3_scale, conv_7x1_3_relu, conv_1x7_3, conv_1x7_3_bn, conv_1x7_3_scale, conv_1x7_3_relu, \
concat
def reduction_v3_b(bottom):
"""
input:768x17x17
output:1280x8x8
:param bottom: bottom layer
:return: layers
"""
pool = L.Pooling(bottom, kernel_size=3, stride=2, pool=P.Pooling.MAX) # 768x8x8
conv_3x3_reduce, conv_3x3_reduce_bn, conv_3x3_reduce_scale, conv_3x3_reduce_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=192, kernel_size=1) # 192x17x17
conv_3x3, conv_3x3_bn, conv_3x3_scale, conv_3x3_relu = \
factorization_conv_bn_scale_relu(conv_3x3_reduce, num_output=320, kernel_size=3, stride=2) # 320x8x8
conv_1x7_reduce, conv_1x7_reduce_bn, conv_1x7_reduce_scale, conv_1x7_reduce_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=192, kernel_size=1) # 192x17x17
conv_1x7, conv_1x7_bn, conv_1x7_scale, conv_1x7_relu = \
factorization_conv_mxn(conv_1x7_reduce, num_output=192, kernel_h=1, kernel_w=7, pad_h=0, pad_w=3) # 192x17x17
conv_7x1, conv_7x1_bn, conv_7x1_scale, conv_7x1_relu = \
factorization_conv_mxn(conv_1x7, num_output=192, kernel_h=7, kernel_w=1, pad_h=3, pad_w=0) # 192x17x17
conv_3x3_2, conv_3x3_2_bn, conv_3x3_2_scale, conv_3x3_2_relu = \
factorization_conv_bn_scale_relu(conv_7x1, num_output=192, kernel_size=3, stride=2) # 192x8x8
concat = L.Concat(pool, conv_3x3, conv_3x3_2) # 1280(768+320+192)x8x8
return pool, conv_3x3_reduce, conv_3x3_reduce_bn, conv_3x3_reduce_scale, conv_3x3_reduce_relu, conv_3x3, conv_3x3_bn, \
conv_3x3_scale, conv_3x3_relu, conv_1x7_reduce, conv_1x7_reduce_bn, conv_1x7_reduce_scale, conv_1x7_reduce_relu, \
conv_1x7, conv_1x7_bn, conv_1x7_scale, conv_1x7_relu, conv_7x1, conv_7x1_bn, conv_7x1_scale, conv_7x1_relu, \
conv_3x3_2, conv_3x3_2_bn, conv_3x3_2_scale, conv_3x3_2_relu, concat
def inception_v3_c(bottom, pool=P.Pooling.AVE):
"""
input:1280or2048x8x8
output:2048x8x8
:param pool: pool_type
:param bottom: bottom layer
:return: layers
"""
pool = L.Pooling(bottom, kernel_size=3, stride=1, pad=1, pool=pool) # 1280or2048x8x8
conv_1x1, conv_1x1_bn, conv_1x1_scale, conv_1x1_relu = \
factorization_conv_bn_scale_relu(pool, num_output=192, kernel_size=1) # 192x8x8
conv_1x1_2, conv_1x1_2_bn, conv_1x1_2_scale, conv_1x1_2_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=320, kernel_size=1) # 320x8x8
conv_1x3_reduce, conv_1x3_reduce_bn, conv_1x3_reduce_scale, conv_1x3_reduce_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=384, kernel_size=1) # 384x8x8
conv_1x3, conv_1x3_bn, conv_1x3_scale, conv_1x3_relu = \
factorization_conv_mxn(conv_1x3_reduce, num_output=384, kernel_h=1, kernel_w=3, pad_h=0, pad_w=1) # 384x8x8
conv_3x1, conv_3x1_bn, conv_3x1_scale, conv_3x1_relu = \
factorization_conv_mxn(conv_1x3_reduce, num_output=384, kernel_h=3, kernel_w=1, pad_h=1, pad_w=0) # 384x8x8
conv_3x3_reduce, conv_3x3_reduce_bn, conv_3x3_reduce_scale, conv_3x3_reduce_relu = \
factorization_conv_bn_scale_relu(bottom, num_output=448, kernel_size=1) # 448x8x8
conv_3x3, conv_3x3_bn, conv_3x3_scale, conv_3x3_relu = \
factorization_conv_bn_scale_relu(conv_3x3_reduce, num_output=384, kernel_size=3, pad=1) # 384x8x8
conv_1x3_2, conv_1x3_2_bn, conv_1x3_2_scale, conv_1x3_2_relu = \
factorization_conv_mxn(conv_3x3, num_output=384, kernel_h=1, kernel_w=3, pad_h=0, pad_w=1) # 384x8x8
conv_3x1_2, conv_3x1_2_bn, conv_3x1_2_scale, conv_3x1_2_relu = \
factorization_conv_mxn(conv_3x3, num_output=384, kernel_h=3, kernel_w=1, pad_h=1, pad_w=0) # 384x8x8
concat = L.Concat(conv_1x1_2, conv_1x3, conv_3x1, conv_1x3_2, conv_3x1_2, conv_1x1) # 2048(192+320+384+384+384+384)x8x8
return pool, conv_1x1, conv_1x1_bn, conv_1x1_scale, conv_1x1_relu, conv_1x1_2, conv_1x1_2_bn, conv_1x1_2_scale, \
conv_1x1_2_relu, conv_1x3_reduce, conv_1x3_reduce_bn, conv_1x3_reduce_scale, conv_1x3_reduce_relu, conv_1x3, \
conv_1x3_bn, conv_1x3_scale, conv_1x3_relu, conv_3x1, conv_3x1_bn, conv_3x1_scale, conv_3x1_relu, \
conv_3x3_reduce, conv_3x3_reduce_bn, conv_3x3_reduce_scale, conv_3x3_reduce_relu, conv_3x3, conv_3x3_bn, \
conv_3x3_scale, conv_3x3_relu, conv_1x3_2, conv_1x3_2_bn, conv_1x3_2_scale, conv_1x3_2_relu, conv_3x1_2, \
conv_3x1_2_bn, conv_3x1_2_scale, conv_3x1_2_relu, concat
class InceptionV3(object):
def __init__(self, lmdb_train, lmdb_test, num_output):
self.train_data = lmdb_train
self.test_data = lmdb_test
self.classifier_num = num_output
def inception_v3_proto(self, batch_size, phase='TRAIN'):
n = caffe.NetSpec()
if phase == 'TRAIN':
source_data = self.train_data
mirror = True
else:
source_data = self.test_data
mirror = False
n.data, n.label = L.Data(source=source_data, backend=P.Data.LMDB, batch_size=batch_size, ntop=2,
transform_param=dict(crop_size=299, mean_value=[104, 117, 123], mirror=mirror))
# stem
n.conv1_3x3_s2, n.conv1_3x3_s2_bn, n.conv1_3x3_s2_scale, n.conv1_3x3_relu, n.conv2_3x3_s1, n.conv2_3x3_s1_bn, \
n.conv2_3x3_s1_scale, n.conv2_3x3_relu, n.conv3_3x3_s1, n.conv3_3x3_s1_bn, n.conv3_3x3_s1_scale, n.conv3_3x3_relu, \
n.pool1_3x3_s2, n.conv4_3x3_reduce, n.conv4_3x3_reduce_bn, n.conv4_relu_3x3_reduce, n.conv4_3x3_reduce_scale, \
n.conv4_3x3, n.conv4_3x3_bn, n.conv4_3x3_scale, n.conv4_relu_3x3, n.pool2_3x3_s2 = \
stem_v3_299x299(n.data) # 192x35x35
# 3 x inception_v3_a
n.inception_a1_1x1, n.inception_a1_1x1_bn, n.inception_a1_1x1_scale, n.inception_a1_1x1_relu, n.inception_a1_5x5_reduce, \
n.inception_a1_5x5_reduce_bn, n.inception_a1_5x5_reduce_scale, n.inception_a1_5x5_reduce_relu, n.inception_a1_5x5, \
n.inception_a1_5x5_bn, n.inception_a1_5x5_scale, n.inception_a1_5x5_relu, n.inception_a1_3x3_reduce, \
n.inception_a1_3x3_reduce_bn, n.inception_a1_3x3_reduce_scale, n.inception_a1_3x3_reduce_relu, n.inception_a1_3x3_1, \
n.inception_a1_3x3_bn, n.inception_a1_3x3_scale, n.inception_a1_3x3_relu, n.inception_a1_3x3_2, \
n.inception_a1_3x3_2_bn, n.inception_a1_3x3_2_scale, n.inception_a1_3x3_2_relu, n.inception_a1_pool, \
n.inception_a1_pool_proj, n.inception_a1_pool_proj_bn, n.inception_a1_pool_proj_scale, n.inception_a1_pool_proj_relu, \
n.inception_a1_output = \
inception_v3_a(n.pool2_3x3_s2) # 256x35x35
n.inception_a2_1x1, n.inception_a2_1x1_bn, n.inception_a2_1x1_scale, n.inception_a2_1x1_relu, n.inception_a2_5x5_reduce, \
n.inception_a2_5x5_reduce_bn, n.inception_a2_5x5_reduce_scale, n.inception_a2_5x5_reduce_relu, n.inception_a2_5x5, \
n.inception_a2_5x5_bn, n.inception_a2_5x5_scale, n.inception_a2_5x5_relu, n.inception_a2_3x3_reduce, \
n.inception_a2_3x3_reduce_bn, n.inception_a2_3x3_reduce_scale, n.inception_a2_3x3_reduce_relu, n.inception_a2_3x3_1, \
n.inception_a2_3x3_bn, n.inception_a2_3x3_scale, n.inception_a2_3x3_relu, n.inception_a2_3x3_2, \
n.inception_a2_3x3_2_bn, n.inception_a2_3x3_2_scale, n.inception_a2_3x3_2_relu, n.inception_a2_pool, \
n.inception_a2_pool_proj, n.inception_a2_pool_proj_bn, n.inception_a2_pool_proj_scale, n.inception_a2_pool_proj_relu, \
n.inception_a2_output = \
inception_v3_a(n.inception_a1_output, pool_proj_num_output=64) # 288x35x35
n.inception_a3_1x1, n.inception_a3_1x1_bn, n.inception_a3_1x1_scale, n.inception_a3_1x1_relu, n.inception_a3_5x5_reduce, \
n.inception_a3_5x5_reduce_bn, n.inception_a3_5x5_reduce_scale, n.inception_a3_5x5_reduce_relu, n.inception_a3_5x5, \
n.inception_a3_5x5_bn, n.inception_a3_5x5_scale, n.inception_a3_5x5_relu, n.inception_a3_3x3_reduce, \
n.inception_a3_3x3_reduce_bn, n.inception_a3_3x3_reduce_scale, n.inception_a3_3x3_reduce_relu, n.inception_a3_3x3_1, \
n.inception_a3_3x3_bn, n.inception_a3_3x3_scale, n.inception_a3_3x3_relu, n.inception_a3_3x3_2, \
n.inception_a3_3x3_2_bn, n.inception_a3_3x3_2_scale, n.inception_a3_3x3_2_relu, n.inception_a3_pool, \
n.inception_a3_pool_proj, n.inception_a3_pool_proj_bn, n.inception_a3_pool_proj_scale, n.inception_a3_pool_proj_relu, \
n.inception_a3_output = \
inception_v3_a(n.inception_a2_output, pool_proj_num_output=64) # 288x35x35
# reduction_v3_a
n.reduction_a_pool, n.reduction_a_3x3, n.reduction_a_3x3_bn, n.reduction_a_3x3_scale, n.reduction_a_3x3_relu, \
n.reduction_a_3x3_2_reduce, n.reduction_a_3x3_2_reduce_bn, n.reduction_a_3x3_2_reduce_scale, n.reduction_a_3x3_2_reduce_relu, \
n.reduction_a_3x3_2, n.reduction_a_3x3_2_bn, n.reduction_a_3x3_2_scale, n.reduction_a_3x3_2_relu, n.reduction_a_3x3_3, \
n.reduction_a_3x3_3_bn, n.reduction_a_3x3_3_scale, n.reduction_a_3x3_3_relu, n.reduction_a_concat = \
reduction_v3_a(n.inception_a3_output) # 768x17x17
# 4 x inception_v3_b
n.inception_b1_pool_ave, n.inception_b1_1x1, n.inception_b1_1x1_bn, n.inception_b1_1x1_scale, n.inception_b1_1x1_relu, \
n.inception_b1_1x1_2, n.inception_b1_1x1_2_bn, n.inception_b1_1x1_2_scale, n.inception_b1_1x1_2_relu, \
n.inception_b1_1x7_reduce, n.inception_b1_1x7_reduce_bn, n.inception_b1_1x7_reduce_scale, n.inception_b1_1x7_reduce_relu, \
n.inception_b1_1x7, n.inception_b1_1x7_bn, n.inception_b1_1x7_scale, n.inception_b1_1x7_relu, n.inception_b1_7x1, \
n.inception_b1_7x1_bn, n.inception_b1_7x1_scale, n.inception_b1_7x1_relu, n.inception_b1_7x1_reduce, n.inception_b1_7x1_reduce_bn, \
n.inception_b1_7x1_reduce_scale, n.inception_b1_7x1_reduce_relu, n.inception_b1_7x1_2, n.inception_b1_7x1_2_bn, \
n.inception_b1_7x1_2_scale, n.inception_b1_7x1_2_relu, n.inception_b1_1x7_2, n.inception_b1_1x7_2_bn, n.inception_b1_1x7_2_scale, \
n.inception_b1_1x7_2_relu, n.inception_b1_7x1_3, n.inception_b1_7x1_3_bn, n.inception_b1_7x1_3_scale, n.inception_b1_7x1_3_relu, \
n.inception_b1_1x7_3, n.inception_b1_1x7_3_bn, n.inception_b1_1x7_3_scale, n.inception_b1_1x7_3_relu, n.inception_b1_concat = \
inception_v3_b(n.reduction_a_concat, outs=128) # 768x17x17
n.inception_b2_pool_ave, n.inception_b2_1x1, n.inception_b2_1x1_bn, n.inception_b2_1x1_scale, n.inception_b2_1x1_relu, \
n.inception_b2_1x1_2, n.inception_b2_1x1_2_bn, n.inception_b2_1x1_2_scale, n.inception_b2_1x1_2_relu, \
n.inception_b2_1x7_reduce, n.inception_b2_1x7_reduce_bn, n.inception_b2_1x7_reduce_scale, n.inception_b2_1x7_reduce_relu, \
n.inception_b2_1x7, n.inception_b2_1x7_bn, n.inception_b2_1x7_scale, n.inception_b2_1x7_relu, n.inception_b2_7x1, \
n.inception_b2_7x1_bn, n.inception_b2_7x1_scale, n.inception_b2_7x1_relu, n.inception_b2_7x1_reduce, n.inception_b2_7x1_reduce_bn, \
n.inception_b2_7x1_reduce_scale, n.inception_b2_7x1_reduce_relu, n.inception_b2_7x1_2, n.inception_b2_7x1_2_bn, \
n.inception_b2_7x1_2_scale, n.inception_b2_7x1_2_relu, n.inception_b2_1x7_2, n.inception_b2_1x7_2_bn, n.inception_b2_1x7_2_scale, \
n.inception_b2_1x7_2_relu, n.inception_b2_7x1_3, n.inception_b2_7x1_3_bn, n.inception_b2_7x1_3_scale, n.inception_b2_7x1_3_relu, \
n.inception_b2_1x7_3, n.inception_b2_1x7_3_bn, n.inception_b2_1x7_3_scale, n.inception_b2_1x7_3_relu, n.inception_b2_concat = \
inception_v3_b(n.inception_b1_concat, outs=160) # 768x17x17
n.inception_b3_pool_ave, n.inception_b3_1x1, n.inception_b3_1x1_bn, n.inception_b3_1x1_scale, n.inception_b3_1x1_relu, \
n.inception_b3_1x1_2, n.inception_b3_1x1_2_bn, n.inception_b3_1x1_2_scale, n.inception_b3_1x1_2_relu, \
n.inception_b3_1x7_reduce, n.inception_b3_1x7_reduce_bn, n.inception_b3_1x7_reduce_scale, n.inception_b3_1x7_reduce_relu, \
n.inception_b3_1x7, n.inception_b3_1x7_bn, n.inception_b3_1x7_scale, n.inception_b3_1x7_relu, n.inception_b3_7x1, \
n.inception_b3_7x1_bn, n.inception_b3_7x1_scale, n.inception_b3_7x1_relu, n.inception_b3_7x1_reduce, n.inception_b3_7x1_reduce_bn, \
n.inception_b3_7x1_reduce_scale, n.inception_b3_7x1_reduce_relu, n.inception_b3_7x1_2, n.inception_b3_7x1_2_bn, \
n.inception_b3_7x1_2_scale, n.inception_b3_7x1_2_relu, n.inception_b3_1x7_2, n.inception_b3_1x7_2_bn, n.inception_b3_1x7_2_scale, \
n.inception_b3_1x7_2_relu, n.inception_b3_7x1_3, n.inception_b3_7x1_3_bn, n.inception_b3_7x1_3_scale, n.inception_b3_7x1_3_relu, \
n.inception_b3_1x7_3, n.inception_b3_1x7_3_bn, n.inception_b3_1x7_3_scale, n.inception_b3_1x7_3_relu, n.inception_b3_concat = \
inception_v3_b(n.inception_b2_concat, outs=160) # 768x17x17
n.inception_b4_pool_ave, n.inception_b4_1x1, n.inception_b4_1x1_bn, n.inception_b4_1x1_scale, n.inception_b4_1x1_relu, \
n.inception_b4_1x1_2, n.inception_b4_1x1_2_bn, n.inception_b4_1x1_2_scale, n.inception_b4_1x1_2_relu, \
n.inception_b4_1x7_reduce, n.inception_b4_1x7_reduce_bn, n.inception_b4_1x7_reduce_scale, n.inception_b4_1x7_reduce_relu, \
n.inception_b4_1x7, n.inception_b4_1x7_bn, n.inception_b4_1x7_scale, n.inception_b4_1x7_relu, n.inception_b4_7x1, \
n.inception_b4_7x1_bn, n.inception_b4_7x1_scale, n.inception_b4_7x1_relu, n.inception_b4_7x1_reduce, n.inception_b4_7x1_reduce_bn, \
n.inception_b4_7x1_reduce_scale, n.inception_b4_7x1_reduce_relu, n.inception_b4_7x1_2, n.inception_b4_7x1_2_bn, \
n.inception_b4_7x1_2_scale, n.inception_b4_7x1_2_relu, n.inception_b4_1x7_2, n.inception_b4_1x7_2_bn, n.inception_b4_1x7_2_scale, \
n.inception_b4_1x7_2_relu, n.inception_b4_7x1_3, n.inception_b4_7x1_3_bn, n.inception_b4_7x1_3_scale, n.inception_b4_7x1_3_relu, \
n.inception_b4_1x7_3, n.inception_b4_1x7_3_bn, n.inception_b4_1x7_3_scale, n.inception_b4_1x7_3_relu, n.inception_b4_concat = \
inception_v3_b(n.inception_b3_concat, outs=192) # 768x17x17
# loss 1
n.auxiliary_loss_ave_pool = L.Pooling(n.inception_b4_concat, kernel_size=5, stride=3,
pool=P.Pooling.AVE) # 768x5x5
n.auxiliary_loss_conv, n.auxiliary_loss_conv_bn, n.auxiliary_loss_conv_scale, n.auxiliary_loss_relu_conv = \
factorization_conv_bn_scale_relu(n.auxiliary_loss_ave_pool, num_output=128, kernel_size=1) # 128x1x1
n.auxiliary_loss_fc = L.InnerProduct(n.auxiliary_loss_conv, num_output=768,
param=[dict(lr_mult=1, decay_mult=1), dict(lr_mult=2, decay_mult=0)],
weight_filler=dict(type='xavier', std=0.01),
bias_filler=dict(type='constant', value=0))
n.auxiliary_loss_fc_relu = L.ReLU(n.auxiliary_loss_fc, in_place=True)
n.auxiliary_loss_classifier = L.InnerProduct(n.auxiliary_loss_fc, num_output=self.classifier_num,
param=[dict(lr_mult=1, decay_mult=1),
dict(lr_mult=2, decay_mult=0)],
weight_filler=dict(type='xavier'),
bias_filler=dict(type='constant', value=0))
n.auxiliary_loss = L.SoftmaxWithLoss(n.auxiliary_loss_classifier, n.label, loss_weight=0.4)
# reduction_v3_b
n.reduction_b_pool, n.reduction_b_3x3_reduce, n.reduction_b_3x3_reduce_bn, n.reduction_b_3x3_reduce_scale, \
n.reduction_b_3x3_reduce_relu, n.reduction_b_3x3, n.reduction_b_3x3_bn, n.reduction_b_3x3_scale, n.reduction_b_3x3_relu, \
n.reduction_b_1x7_reduce, n.reduction_b_1x7_reduce_bn, n.reduction_b_1x7_reduce_scale, n.reduction_b_1x7_reduce_relu, \
n.reduction_b_1x7, n.reduction_b_1x7_bn, n.reduction_b_1x7_scale, n.reduction_b_1x7_relu, n.reduction_b_7x1, \
n.reduction_b_7x1_bn, n.reduction_b_7x1_scale, n.reduction_b_7x1_relu, n.reduction_b_3x3_2, n.reduction_b_3x3_2_bn, \
n.reduction_b_3x3_2_scale, n.reduction_b_3x3_2_relu, n.reduction_b_concat = \
reduction_v3_b(n.inception_b4_concat) # 1280x8x8
# 2 x inception_v3_c
n.inception_c1_pool, n.inception_c1_1x1, n.inception_c1_1x1_bn, n.inception_c1_1x1_scale, n.inception_c1_1x1_relu, \
n.inception_c1_1x1_2, n.inception_c1_1x1_2_bn, n.inception_c1_1x1_2_scale, n.inception_c1_1x1_2_relu, \
n.inception_c1_1x3_reduce, n.inception_c1_1x3_reduce_bn, n.inception_c1_1x3_reduce_scale, n.inception_c1_1x3_reduce_relu, \
n.inception_c1_1x3, n.inception_c1_1x3_bn, n.inception_c1_1x3_scale, n.inception_c1_1x3_relu, n.inception_c1_3x1, \
n.inception_c1_3x1_bn, n.inception_c1_3x1_scale, n.inception_c1_3x1_relu, n.inception_c1_3x3_reduce, \
n.inception_c1_3x3_reduce_bn, n.inception_c1_3x3_reduce_scale, n.inception_c1_3x3_reduce_relu, n.inception_c1_3x3, \
n.inception_c1_3x3_bn, n.inception_c1_3x3_scale, n.inception_c1_3x3_relu, n.inception_c1_1x3_2, n.inception_c1_1x3_2_bn, \
n.inception_c1_1x3_2_scale, n.inception_c1_1x3_2_relu, n.inception_c1_3x1_2, n.inception_c1_3x1_2_bn, n.inception_c1_3x1_2_scale, \
n.inception_c1_3x1_2_relu, n.inception_c1_concat = \
inception_v3_c(n.reduction_b_concat) # 2048x8x8
n.inception_c2_pool, n.inception_c2_1x1, n.inception_c2_1x1_bn, n.inception_c2_1x1_scale, n.inception_c2_1x1_relu, \
n.inception_c2_1x1_2, n.inception_c2_1x1_2_bn, n.inception_c2_1x1_2_scale, n.inception_c2_1x1_2_relu, \
n.inception_c2_1x3_reduce, n.inception_c2_1x3_reduce_bn, n.inception_c2_1x3_reduce_scale, n.inception_c2_1x3_reduce_relu, \
n.inception_c2_1x3, n.inception_c2_1x3_bn, n.inception_c2_1x3_scale, n.inception_c2_1x3_relu, n.inception_c2_3x1, \
n.inception_c2_3x1_bn, n.inception_c2_3x1_scale, n.inception_c2_3x1_relu, n.inception_c2_3x3_reduce, \
n.inception_c2_3x3_reduce_bn, n.inception_c2_3x3_reduce_scale, n.inception_c2_3x3_reduce_relu, n.inception_c2_3x3, \
n.inception_c2_3x3_bn, n.inception_c2_3x3_scale, n.inception_c2_3x3_relu, n.inception_c2_1x3_2, n.inception_c2_1x3_2_bn, \
n.inception_c2_1x3_2_scale, n.inception_c2_1x3_2_relu, n.inception_c2_3x1_2, n.inception_c2_3x1_2_bn, n.inception_c2_3x1_2_scale, \
n.inception_c2_3x1_2_relu, n.inception_c2_concat = \
inception_v3_c(n.inception_c1_concat, pool=P.Pooling.MAX) # 2048x8x8
# loss 2
n.pool_8x8_s1 = L.Pooling(n.inception_c2_concat, kernel_size=8, pool=P.Pooling.AVE)
n.pool_8x8_s1_drop = L.Dropout(n.pool_8x8_s1, dropout_param=dict(dropout_ratio=0.2))
n.classifier = L.InnerProduct(n.pool_8x8_s1_drop, num_output=self.classifier_num,
param=[dict(lr_mult=1, decay_mult=1), dict(lr_mult=2, decay_mult=0)],
weight_filler=dict(type='xavier'),
bias_filler=dict(type='constant', value=0))
n.loss = L.SoftmaxWithLoss(n.classifier, n.label)
if phase == 'TRAIN':
pass
else:
n.accuracy_top1 = L.Accuracy(n.classifier, n.label, include=dict(phase=1))
n.accuracy_top5 = L.Accuracy(n.classifier, n.label, include=dict(phase=1),
accuracy_param=dict(top_k=5))
return n.to_proto()