forked from dmonterom/face_recognition_TF2
-
Notifications
You must be signed in to change notification settings - Fork 0
/
resnet_groupNorm.py
360 lines (318 loc) · 15.3 KB
/
resnet_groupNorm.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
import tensorflow as tf
import os
from arcface import Arcfacelayer
bn_axis = -1
initializer = 'glorot_normal'
class GroupNormalization(tf.keras.layers.Layer):
"""Group normalization layer
Group Normalization divides the channels into groups and computes within each group
the mean and variance for normalization. GN's computation is independent of batch sizes,
and its accuracy is stable in a wide range of batch sizes
# Arguments
groups: Integer, the number of groups for Group Normalization.
axis: Integer, the axis that should be normalized
(typically the features axis).
For instance, after a `Conv2D` layer with
`data_format="channels_first"`,
set `axis=1` in `BatchNormalization`.
epsilon: Small float added to variance to avoid dividing by zero.
center: If True, add offset of `beta` to normalized tensor.
If False, `beta` is ignored.
scale: If True, multiply by `gamma`.
If False, `gamma` is not used.
When the next layer is linear (also e.g. `nn.relu`),
this can be disabled since the scaling
will be done by the next layer.
beta_initializer: Initializer for the beta weight.
gamma_initializer: Initializer for the gamma weight.
beta_regularizer: Optional regularizer for the beta weight.
gamma_regularizer: Optional regularizer for the gamma weight.
beta_constraint: Optional constraint for the beta weight.
gamma_constraint: Optional constraint for the gamma weight.
# Input shape
Arbitrary. Use the keyword argument `input_shape`
(tuple of integers, does not include the samples axis)
when using this layer as the first layer in a model.
# Output shape
Same shape as input.
# References
- [Group Normalization](https://arxiv.org/abs/1803.08494)
"""
def __init__(self,
groups=32,
axis=-1,
epsilon=1e-5,
center=True,
scale=True,
beta_initializer='zeros',
gamma_initializer='ones',
beta_regularizer=None,
gamma_regularizer=None,
beta_constraint=None,
gamma_constraint=None,
**kwargs):
super(GroupNormalization, self).__init__(**kwargs)
self.supports_masking = True
self.groups = groups
self.axis = axis
self.epsilon = epsilon
self.center = center
self.scale = scale
self.beta_initializer = tf.keras.initializers.get(beta_initializer)
self.gamma_initializer = tf.keras.initializers.get(gamma_initializer)
self.beta_regularizer = tf.keras.regularizers.get(beta_regularizer)
self.gamma_regularizer = tf.keras.regularizers.get(gamma_regularizer)
self.beta_constraint = tf.keras.constraints.get(beta_constraint)
self.gamma_constraint = tf.keras.constraints.get(gamma_constraint)
def build(self, input_shape):
dim = input_shape[self.axis]
if dim is None:
raise ValueError('Axis ' + str(self.axis) + ' of '
'input tensor should have a defined dimension '
'but the layer received an input with shape ' +
str(input_shape) + '.')
if dim < self.groups:
raise ValueError('Number of groups (' + str(self.groups) + ') cannot be '
'more than the number of channels (' +
str(dim) + ').')
if dim % self.groups != 0:
raise ValueError('Number of groups (' + str(self.groups) + ') must be a '
'multiple of the number of channels (' +
str(dim) + ').')
self.input_spec = tf.keras.layers.InputSpec(ndim=len(input_shape),
axes={self.axis: dim})
shape = (dim,)
if self.scale:
self.gamma = self.add_weight(shape=shape,
name='gamma',
initializer=self.gamma_initializer,
regularizer=self.gamma_regularizer,
constraint=self.gamma_constraint)
else:
self.gamma = None
if self.center:
self.beta = self.add_weight(shape=shape,
name='beta',
initializer=self.beta_initializer,
regularizer=self.beta_regularizer,
constraint=self.beta_constraint)
else:
self.beta = None
self.built = True
def call(self, inputs, **kwargs):
input_shape = tf.keras.backend.int_shape(inputs)
tensor_input_shape = tf.keras.backend.shape(inputs)
# Prepare broadcasting shape.
reduction_axes = list(range(len(input_shape)))
del reduction_axes[self.axis]
broadcast_shape = [1] * len(input_shape)
broadcast_shape[self.axis] = input_shape[self.axis] // self.groups
broadcast_shape.insert(1, self.groups)
reshape_group_shape = tf.keras.backend.shape(inputs)
group_axes = [reshape_group_shape[i] for i in range(len(input_shape))]
group_axes[self.axis] = input_shape[self.axis] // self.groups
group_axes.insert(1, self.groups)
# reshape inputs to new group shape
group_shape = [group_axes[0], self.groups] + group_axes[2:]
group_shape = tf.keras.backend.stack(group_shape)
inputs = tf.keras.backend.reshape(inputs, group_shape)
group_reduction_axes = list(range(len(group_axes)))
group_reduction_axes = group_reduction_axes[2:]
mean = tf.keras.backend.mean(
inputs, axis=group_reduction_axes, keepdims=True)
variance = tf.keras.backend.var(
inputs, axis=group_reduction_axes, keepdims=True)
inputs = (inputs - mean) / \
(tf.keras.backend.sqrt(variance + self.epsilon))
# prepare broadcast shape
inputs = tf.keras.backend.reshape(inputs, group_shape)
outputs = inputs
# In this case we must explicitly broadcast all parameters.
if self.scale:
broadcast_gamma = tf.keras.backend.reshape(
self.gamma, broadcast_shape)
outputs = outputs * broadcast_gamma
if self.center:
broadcast_beta = tf.keras.backend.reshape(
self.beta, broadcast_shape)
outputs = outputs + broadcast_beta
outputs = tf.keras.backend.reshape(outputs, tensor_input_shape)
return outputs
def get_config(self):
config = {
'groups': self.groups,
'axis': self.axis,
'epsilon': self.epsilon,
'center': self.center,
'scale': self.scale,
'beta_initializer': tf.keras.initializers.serialize(self.beta_initializer),
'gamma_initializer': tf.keras.initializers.serialize(self.gamma_initializer),
'beta_regularizer': tf.keras.regularizers.serialize(self.beta_regularizer),
'gamma_regularizer': tf.keras.regularizers.serialize(self.gamma_regularizer),
'beta_constraint': tf.keras.constraints.serialize(self.beta_constraint),
'gamma_constraint': tf.keras.constraints.serialize(self.gamma_constraint)
}
base_config = super(GroupNormalization, self).get_config()
return dict(list(base_config.items()) + list(config.items()))
def compute_output_shape(self, input_shape):
return input_shape
tf.keras.utils.get_custom_objects().update(
{'GroupNormalization': GroupNormalization})
def residual_unit_v3(input, num_filter, stride, dim_match, name):
x = GroupNormalization(axis=bn_axis,
scale=True,
epsilon=2e-5,
# beta_regularizer=tf.keras.regularizers.l2(
# l=5e-4),
gamma_regularizer=tf.keras.regularizers.l2(
l=5e-4),
name=name + '_bn1')(input)
x = tf.keras.layers.ZeroPadding2D(
padding=(1, 1), name=name + '_conv1_pad')(x)
x = tf.keras.layers.Conv2D(num_filter, (3, 3),
strides=(1, 1),
padding='valid',
kernel_initializer=initializer,
use_bias=False,
kernel_regularizer=tf.keras.regularizers.l2(
l=5e-4),
name=name + '_conv1')(x)
x = GroupNormalization(axis=bn_axis,
scale=True,
epsilon=2e-5,
# beta_regularizer=tf.keras.regularizers.l2(
# l=5e-4),
gamma_regularizer=tf.keras.regularizers.l2(
l=5e-4),
name=name + '_bn2')(x)
x = tf.keras.layers.PReLU(name=name + '_relu1',
alpha_regularizer=tf.keras.regularizers.l2(
l=5e-4))(x)
x = tf.keras.layers.ZeroPadding2D(
padding=(1, 1), name=name + '_conv2_pad')(x)
x = tf.keras.layers.Conv2D(num_filter, (3, 3),
strides=stride,
padding='valid',
kernel_initializer=initializer,
use_bias=False,
kernel_regularizer=tf.keras.regularizers.l2(
l=5e-4),
name=name + '_conv2')(x)
x = GroupNormalization(axis=bn_axis,
scale=True,
epsilon=2e-5,
# beta_regularizer=tf.keras.regularizers.l2(
# l=5e-4),
gamma_regularizer=tf.keras.regularizers.l2(
l=5e-4),
name=name + '_bn3')(x)
if (dim_match):
shortcut = input
else:
shortcut = tf.keras.layers.Conv2D(num_filter, (1, 1),
strides=stride,
padding='valid',
kernel_initializer=initializer,
use_bias=False,
kernel_regularizer=tf.keras.regularizers.l2(
l=5e-4),
name=name + '_conv1sc')(input)
shortcut = GroupNormalization(axis=bn_axis,
scale=True,
epsilon=2e-5,
# beta_regularizer=tf.keras.regularizers.l2(
# l=5e-4),
gamma_regularizer=tf.keras.regularizers.l2(
l=5e-4),
name=name + '_sc')(shortcut)
return x + shortcut
def get_fc1(input):
x = GroupNormalization(axis=bn_axis,
scale=True,
epsilon=2e-5,
# beta_regularizer=tf.keras.regularizers.l2(
# l=5e-4),
gamma_regularizer=tf.keras.regularizers.l2(
l=5e-4),
name='bn1')(input)
x = tf.keras.layers.Dropout(0.4)(x)
resnet_shape = input.shape
x = tf.keras.layers.Reshape(
[resnet_shape[1] * resnet_shape[2] * resnet_shape[3]], name='reshapelayer')(x)
x = tf.keras.layers.Dense(512,
name='E_DenseLayer', kernel_initializer=initializer,
kernel_regularizer=tf.keras.regularizers.l2(
l=5e-4),
bias_regularizer=tf.keras.regularizers.l2(
l=5e-4))(x)
x = GroupNormalization(axis=-1,
scale=False,
epsilon=2e-5,
# beta_regularizer=tf.keras.regularizers.l2(
# l=5e-4),
name='fc1')(x)
return x
def ResNet50():
input_shape = [112, 112, 3]
filter_list = [64, 64, 128, 256, 512]
units = [3, 4, 14, 3]
num_stages = 4
img_input = tf.keras.layers.Input(shape=input_shape)
x = tf.keras.layers.ZeroPadding2D(
padding=(1, 1), name='conv0_pad')(img_input)
x = tf.keras.layers.Conv2D(64, (3, 3),
strides=(1, 1),
padding='valid',
kernel_initializer=initializer,
use_bias=False,
kernel_regularizer=tf.keras.regularizers.l2(
l=5e-4),
name='conv0')(x)
x = GroupNormalization(axis=bn_axis,
scale=True,
epsilon=2e-5,
# beta_regularizer=tf.keras.regularizers.l2(
# l=5e-4),
gamma_regularizer=tf.keras.regularizers.l2(
l=5e-4),
name='bn0')(x)
# x = tf.keras.layers.Activation('prelu')(x)
x = tf.keras.layers.PReLU(
name='prelu0',
alpha_regularizer=tf.keras.regularizers.l2(
l=5e-4))(x)
for i in range(num_stages):
x = residual_unit_v3(x, filter_list[i + 1], (2, 2), False,
name='stage%d_unit%d' % (i + 1, 1))
for j in range(units[i] - 1):
x = residual_unit_v3(x, filter_list[i + 1], (1, 1),
True, name='stage%d_unit%d' % (i + 1, j + 2))
x = get_fc1(x)
# Create model.
model = tf.keras.models.Model(img_input, x, name='resnet50')
model.trainable = True
for i in range(len(model.layers)):
model.layers[i].trainable = True
# if ('conv0' in model.layers[i].name):
# model.layers[i].trainable = False
# if ('bn0' in model.layers[i].name):
# model.layers[i].trainable = False
# if ('prelu0' in model.layers[i].name):
# model.layers[i].trainable = False
# if ('stage1' in model.layers[i].name):
# model.layers[i].trainable = False
# if ('stage2' in model.layers[i].name):
# model.layers[i].trainable = False
# if ('stage3' in model.layers[i].name):
# model.layers[i].trainable = False
# if ('stage4' in model.layers[i].name):
# model.layers[i].trainable = False
return model
class train_model(tf.keras.Model):
def __init__(self):
super(train_model, self).__init__()
self.resnet = ResNet50()
self.arcface = Arcfacelayer()
def call(self, x, y):
x = self.resnet(x)
return self.arcface(x, y)