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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754
|
"""NASNet-A models for Keras.
NASNet refers to Neural Architecture Search Network, a family of models
that were designed automatically by learning the model architectures
directly on the dataset of interest.
Here we consider NASNet-A, the highest performance model that was found
for the CIFAR-10 dataset, and then extended to ImageNet 2012 dataset,
obtaining state of the art performance on CIFAR-10 and ImageNet 2012.
Only the NASNet-A models, and their respective weights, which are suited
for ImageNet 2012 are provided.
The below table describes the performance on ImageNet 2012:
--------------------------------------------------------------------------------
Architecture | Top-1 Acc | Top-5 Acc | Multiply-Adds | Params (M)
--------------------------------------------------------------------------------
| NASNet-A (4 @ 1056) | 74.0 % | 91.6 % | 564 M | 5.3 |
| NASNet-A (6 @ 4032) | 82.7 % | 96.2 % | 23.8 B | 88.9 |
--------------------------------------------------------------------------------
Weights obtained from the official TensorFlow repository found at
https://github.com/tensorflow/models/tree/master/research/slim/nets/nasnet
# References
- [Learning Transferable Architectures for Scalable Image Recognition]
(https://arxiv.org/abs/1707.07012) (CVPR 2018)
This model is based on the following implementations:
- [TF Slim Implementation]
(https://github.com/tensorflow/models/blob/master/research/slim/nets/nasnet/nasnet.py)
- [TensorNets implementation]
(https://github.com/taehoonlee/tensornets/blob/master/tensornets/nasnets.py)
"""
from __future__ import print_function
from __future__ import absolute_import
from __future__ import division
import os
import warnings
from . import correct_pad
from . import get_submodules_from_kwargs
from . import imagenet_utils
from .imagenet_utils import decode_predictions
from .imagenet_utils import _obtain_input_shape
BASE_WEIGHTS_PATH = ('https://github.com/titu1994/Keras-NASNet/'
'releases/download/v1.2/')
NASNET_MOBILE_WEIGHT_PATH = BASE_WEIGHTS_PATH + 'NASNet-mobile.h5'
NASNET_MOBILE_WEIGHT_PATH_NO_TOP = BASE_WEIGHTS_PATH + 'NASNet-mobile-no-top.h5'
NASNET_LARGE_WEIGHT_PATH = BASE_WEIGHTS_PATH + 'NASNet-large.h5'
NASNET_LARGE_WEIGHT_PATH_NO_TOP = BASE_WEIGHTS_PATH + 'NASNet-large-no-top.h5'
backend = None
layers = None
models = None
keras_utils = None
def NASNet(input_shape=None,
penultimate_filters=4032,
num_blocks=6,
stem_block_filters=96,
skip_reduction=True,
filter_multiplier=2,
include_top=True,
weights=None,
input_tensor=None,
pooling=None,
classes=1000,
default_size=None,
**kwargs):
'''Instantiates a NASNet model.
Optionally loads weights pre-trained on ImageNet.
Note that the data format convention used by the model is
the one specified in your Keras config at `~/.keras/keras.json`.
# Arguments
input_shape: Optional shape tuple, the input shape
is by default `(331, 331, 3)` for NASNetLarge and
`(224, 224, 3)` for NASNetMobile.
It should have exactly 3 input channels,
and width and height should be no smaller than 32.
E.g. `(224, 224, 3)` would be one valid value.
penultimate_filters: Number of filters in the penultimate layer.
NASNet models use the notation `NASNet (N @ P)`, where:
- N is the number of blocks
- P is the number of penultimate filters
num_blocks: Number of repeated blocks of the NASNet model.
NASNet models use the notation `NASNet (N @ P)`, where:
- N is the number of blocks
- P is the number of penultimate filters
stem_block_filters: Number of filters in the initial stem block
skip_reduction: Whether to skip the reduction step at the tail
end of the network.
filter_multiplier: Controls the width of the network.
- If `filter_multiplier` < 1.0, proportionally decreases the number
of filters in each layer.
- If `filter_multiplier` > 1.0, proportionally increases the number
of filters in each layer.
- If `filter_multiplier` = 1, default number of filters from the
paper are used at each layer.
include_top: Whether to include the fully-connected
layer at the top of the network.
weights: `None` (random initialization) or
`imagenet` (ImageNet weights)
input_tensor: Optional Keras tensor (i.e. output of
`layers.Input()`)
to use as image input for the model.
pooling: Optional pooling mode for feature extraction
when `include_top` is `False`.
- `None` means that the output of the model
will be the 4D tensor output of the
last convolutional block.
- `avg` means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a
2D tensor.
- `max` means that global max pooling will
be applied.
classes: Optional number of classes to classify images
into, only to be specified if `include_top` is True, and
if no `weights` argument is specified.
default_size: Specifies the default image size of the model
# Returns
A Keras model instance.
# Raises
ValueError: In case of invalid argument for `weights`,
invalid input shape or invalid `penultimate_filters` value.
'''
global backend, layers, models, keras_utils
backend, layers, models, keras_utils = get_submodules_from_kwargs(kwargs)
if not (weights in {'imagenet', None} or os.path.exists(weights)):
raise ValueError('The `weights` argument should be either '
'`None` (random initialization), `imagenet` '
'(pre-training on ImageNet), '
'or the path to the weights file to be loaded.')
if weights == 'imagenet' and include_top and classes != 1000:
raise ValueError('If using `weights` as `"imagenet"` with `include_top` '
'as true, `classes` should be 1000')
if (isinstance(input_shape, tuple) and
None in input_shape and
weights == 'imagenet'):
raise ValueError('When specifying the input shape of a NASNet'
' and loading `ImageNet` weights, '
'the input_shape argument must be static '
'(no None entries). Got: `input_shape=' +
str(input_shape) + '`.')
if default_size is None:
default_size = 331
# Determine proper input shape and default size.
input_shape = _obtain_input_shape(input_shape,
default_size=default_size,
min_size=32,
data_format=backend.image_data_format(),
require_flatten=True,
weights=weights)
if backend.image_data_format() != 'channels_last':
warnings.warn('The NASNet family of models is only available '
'for the input data format "channels_last" '
'(width, height, channels). '
'However your settings specify the default '
'data format "channels_first" (channels, width, height).'
' You should set `image_data_format="channels_last"` '
'in your Keras config located at ~/.keras/keras.json. '
'The model being returned right now will expect inputs '
'to follow the "channels_last" data format.')
backend.set_image_data_format('channels_last')
old_data_format = 'channels_first'
else:
old_data_format = None
if input_tensor is None:
img_input = layers.Input(shape=input_shape)
else:
if not backend.is_keras_tensor(input_tensor):
img_input = layers.Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
if penultimate_filters % (24 * (filter_multiplier ** 2)) != 0:
raise ValueError(
'For NASNet-A models, the `penultimate_filters` must be a multiple '
'of 24 * (`filter_multiplier` ** 2). Current value: %d' %
penultimate_filters)
channel_dim = 1 if backend.image_data_format() == 'channels_first' else -1
filters = penultimate_filters // 24
x = layers.Conv2D(stem_block_filters, (3, 3),
strides=(2, 2),
padding='valid',
use_bias=False,
name='stem_conv1',
kernel_initializer='he_normal')(img_input)
x = layers.BatchNormalization(
axis=channel_dim, momentum=0.9997, epsilon=1e-3, name='stem_bn1')(x)
p = None
x, p = _reduction_a_cell(x, p, filters // (filter_multiplier ** 2),
block_id='stem_1')
x, p = _reduction_a_cell(x, p, filters // filter_multiplier,
block_id='stem_2')
for i in range(num_blocks):
x, p = _normal_a_cell(x, p, filters, block_id='%d' % (i))
x, p0 = _reduction_a_cell(x, p, filters * filter_multiplier,
block_id='reduce_%d' % (num_blocks))
p = p0 if not skip_reduction else p
for i in range(num_blocks):
x, p = _normal_a_cell(x, p, filters * filter_multiplier,
block_id='%d' % (num_blocks + i + 1))
x, p0 = _reduction_a_cell(x, p, filters * filter_multiplier ** 2,
block_id='reduce_%d' % (2 * num_blocks))
p = p0 if not skip_reduction else p
for i in range(num_blocks):
x, p = _normal_a_cell(x, p, filters * filter_multiplier ** 2,
block_id='%d' % (2 * num_blocks + i + 1))
x = layers.Activation('relu')(x)
if include_top:
x = layers.GlobalAveragePooling2D()(x)
x = layers.Dense(classes, activation='softmax', name='predictions')(x)
else:
if pooling == 'avg':
x = layers.GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = layers.GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = keras_utils.get_source_inputs(input_tensor)
else:
inputs = img_input
model = models.Model(inputs, x, name='NASNet')
# Load weights.
if weights == 'imagenet':
if default_size == 224: # mobile version
if include_top:
weights_path = keras_utils.get_file(
'nasnet_mobile.h5',
NASNET_MOBILE_WEIGHT_PATH,
cache_subdir='models',
file_hash='020fb642bf7360b370c678b08e0adf61')
else:
weights_path = keras_utils.get_file(
'nasnet_mobile_no_top.h5',
NASNET_MOBILE_WEIGHT_PATH_NO_TOP,
cache_subdir='models',
file_hash='1ed92395b5b598bdda52abe5c0dbfd63')
model.load_weights(weights_path)
elif default_size == 331: # large version
if include_top:
weights_path = keras_utils.get_file(
'nasnet_large.h5',
NASNET_LARGE_WEIGHT_PATH,
cache_subdir='models',
file_hash='11577c9a518f0070763c2b964a382f17')
else:
weights_path = keras_utils.get_file(
'nasnet_large_no_top.h5',
NASNET_LARGE_WEIGHT_PATH_NO_TOP,
cache_subdir='models',
file_hash='d81d89dc07e6e56530c4e77faddd61b5')
model.load_weights(weights_path)
else:
raise ValueError(
'ImageNet weights can only be loaded with NASNetLarge'
' or NASNetMobile')
elif weights is not None:
model.load_weights(weights)
if old_data_format:
backend.set_image_data_format(old_data_format)
return model
def NASNetLarge(input_shape=None,
include_top=True,
weights='imagenet',
input_tensor=None,
pooling=None,
classes=1000,
**kwargs):
'''Instantiates a NASNet model in ImageNet mode.
Optionally loads weights pre-trained on ImageNet.
Note that the data format convention used by the model is
the one specified in your Keras config at `~/.keras/keras.json`.
# Arguments
input_shape: Optional shape tuple, only to be specified
if `include_top` is False (otherwise the input shape
has to be `(331, 331, 3)` for NASNetLarge.
It should have exactly 3 inputs channels,
and width and height should be no smaller than 32.
E.g. `(224, 224, 3)` would be one valid value.
include_top: Whether to include the fully-connected
layer at the top of the network.
weights: `None` (random initialization) or
`imagenet` (ImageNet weights)
input_tensor: Optional Keras tensor (i.e. output of
`layers.Input()`)
to use as image input for the model.
pooling: Optional pooling mode for feature extraction
when `include_top` is `False`.
- `None` means that the output of the model
will be the 4D tensor output of the
last convolutional layer.
- `avg` means that global average pooling
will be applied to the output of the
last convolutional layer, and thus
the output of the model will be a
2D tensor.
- `max` means that global max pooling will
be applied.
classes: Optional number of classes to classify images
into, only to be specified if `include_top` is True, and
if no `weights` argument is specified.
# Returns
A Keras model instance.
# Raises
ValueError: in case of invalid argument for `weights`,
or invalid input shape.
RuntimeError: If attempting to run this model with a
backend that does not support separable convolutions.
'''
return NASNet(input_shape,
penultimate_filters=4032,
num_blocks=6,
stem_block_filters=96,
skip_reduction=True,
filter_multiplier=2,
include_top=include_top,
weights=weights,
input_tensor=input_tensor,
pooling=pooling,
classes=classes,
default_size=331,
**kwargs)
def NASNetMobile(input_shape=None,
include_top=True,
weights='imagenet',
input_tensor=None,
pooling=None,
classes=1000,
**kwargs):
'''Instantiates a Mobile NASNet model in ImageNet mode.
Optionally loads weights pre-trained on ImageNet.
Note that the data format convention used by the model is
the one specified in your Keras config at `~/.keras/keras.json`.
# Arguments
input_shape: Optional shape tuple, only to be specified
if `include_top` is False (otherwise the input shape
has to be `(224, 224, 3)` for NASNetMobile
It should have exactly 3 inputs channels,
and width and height should be no smaller than 32.
E.g. `(224, 224, 3)` would be one valid value.
include_top: Whether to include the fully-connected
layer at the top of the network.
weights: `None` (random initialization) or
`imagenet` (ImageNet weights)
input_tensor: Optional Keras tensor (i.e. output of
`layers.Input()`)
to use as image input for the model.
pooling: Optional pooling mode for feature extraction
when `include_top` is `False`.
- `None` means that the output of the model
will be the 4D tensor output of the
last convolutional layer.
- `avg` means that global average pooling
will be applied to the output of the
last convolutional layer, and thus
the output of the model will be a
2D tensor.
- `max` means that global max pooling will
be applied.
classes: Optional number of classes to classify images
into, only to be specified if `include_top` is True, and
if no `weights` argument is specified.
# Returns
A Keras model instance.
# Raises
ValueError: In case of invalid argument for `weights`,
or invalid input shape.
RuntimeError: If attempting to run this model with a
backend that does not support separable convolutions.
'''
return NASNet(input_shape,
penultimate_filters=1056,
num_blocks=4,
stem_block_filters=32,
skip_reduction=False,
filter_multiplier=2,
include_top=include_top,
weights=weights,
input_tensor=input_tensor,
pooling=pooling,
classes=classes,
default_size=224,
**kwargs)
def _separable_conv_block(ip, filters,
kernel_size=(3, 3),
strides=(1, 1),
block_id=None):
'''Adds 2 blocks of [relu-separable conv-batchnorm].
# Arguments
ip: Input tensor
filters: Number of output filters per layer
kernel_size: Kernel size of separable convolutions
strides: Strided convolution for downsampling
block_id: String block_id
# Returns
A Keras tensor
'''
channel_dim = 1 if backend.image_data_format() == 'channels_first' else -1
with backend.name_scope('separable_conv_block_%s' % block_id):
x = layers.Activation('relu')(ip)
if strides == (2, 2):
x = layers.ZeroPadding2D(
padding=correct_pad(backend, x, kernel_size),
name='separable_conv_1_pad_%s' % block_id)(x)
conv_pad = 'valid'
else:
conv_pad = 'same'
x = layers.SeparableConv2D(filters, kernel_size,
strides=strides,
name='separable_conv_1_%s' % block_id,
padding=conv_pad, use_bias=False,
kernel_initializer='he_normal')(x)
x = layers.BatchNormalization(
axis=channel_dim,
momentum=0.9997,
epsilon=1e-3,
name='separable_conv_1_bn_%s' % (block_id))(x)
x = layers.Activation('relu')(x)
x = layers.SeparableConv2D(filters, kernel_size,
name='separable_conv_2_%s' % block_id,
padding='same',
use_bias=False,
kernel_initializer='he_normal')(x)
x = layers.BatchNormalization(
axis=channel_dim,
momentum=0.9997,
epsilon=1e-3,
name='separable_conv_2_bn_%s' % (block_id))(x)
return x
def _adjust_block(p, ip, filters, block_id=None):
'''Adjusts the input `previous path` to match the shape of the `input`.
Used in situations where the output number of filters needs to be changed.
# Arguments
p: Input tensor which needs to be modified
ip: Input tensor whose shape needs to be matched
filters: Number of output filters to be matched
block_id: String block_id
# Returns
Adjusted Keras tensor
'''
channel_dim = 1 if backend.image_data_format() == 'channels_first' else -1
img_dim = 2 if backend.image_data_format() == 'channels_first' else -2
ip_shape = backend.int_shape(ip)
if p is not None:
p_shape = backend.int_shape(p)
with backend.name_scope('adjust_block'):
if p is None:
p = ip
elif p_shape[img_dim] != ip_shape[img_dim]:
with backend.name_scope('adjust_reduction_block_%s' % block_id):
p = layers.Activation('relu',
name='adjust_relu_1_%s' % block_id)(p)
p1 = layers.AveragePooling2D(
(1, 1),
strides=(2, 2),
padding='valid',
name='adjust_avg_pool_1_%s' % block_id)(p)
p1 = layers.Conv2D(
filters // 2, (1, 1),
padding='same',
use_bias=False, name='adjust_conv_1_%s' % block_id,
kernel_initializer='he_normal')(p1)
p2 = layers.ZeroPadding2D(padding=((0, 1), (0, 1)))(p)
p2 = layers.Cropping2D(cropping=((1, 0), (1, 0)))(p2)
p2 = layers.AveragePooling2D(
(1, 1),
strides=(2, 2),
padding='valid',
name='adjust_avg_pool_2_%s' % block_id)(p2)
p2 = layers.Conv2D(
filters // 2, (1, 1),
padding='same',
use_bias=False,
name='adjust_conv_2_%s' % block_id,
kernel_initializer='he_normal')(p2)
p = layers.concatenate([p1, p2], axis=channel_dim)
p = layers.BatchNormalization(
axis=channel_dim,
momentum=0.9997,
epsilon=1e-3,
name='adjust_bn_%s' % block_id)(p)
elif p_shape[channel_dim] != filters:
with backend.name_scope('adjust_projection_block_%s' % block_id):
p = layers.Activation('relu')(p)
p = layers.Conv2D(
filters,
(1, 1),
strides=(1, 1),
padding='same',
name='adjust_conv_projection_%s' % block_id,
use_bias=False,
kernel_initializer='he_normal')(p)
p = layers.BatchNormalization(
axis=channel_dim,
momentum=0.9997,
epsilon=1e-3,
name='adjust_bn_%s' % block_id)(p)
return p
def _normal_a_cell(ip, p, filters, block_id=None):
'''Adds a Normal cell for NASNet-A (Fig. 4 in the paper).
# Arguments
ip: Input tensor `x`
p: Input tensor `p`
filters: Number of output filters
block_id: String block_id
# Returns
A Keras tensor
'''
channel_dim = 1 if backend.image_data_format() == 'channels_first' else -1
with backend.name_scope('normal_A_block_%s' % block_id):
p = _adjust_block(p, ip, filters, block_id)
h = layers.Activation('relu')(ip)
h = layers.Conv2D(
filters, (1, 1),
strides=(1, 1),
padding='same',
name='normal_conv_1_%s' % block_id,
use_bias=False,
kernel_initializer='he_normal')(h)
h = layers.BatchNormalization(
axis=channel_dim,
momentum=0.9997,
epsilon=1e-3,
name='normal_bn_1_%s' % block_id)(h)
with backend.name_scope('block_1'):
x1_1 = _separable_conv_block(
h, filters,
kernel_size=(5, 5),
block_id='normal_left1_%s' % block_id)
x1_2 = _separable_conv_block(
p, filters,
block_id='normal_right1_%s' % block_id)
x1 = layers.add([x1_1, x1_2], name='normal_add_1_%s' % block_id)
with backend.name_scope('block_2'):
x2_1 = _separable_conv_block(
p, filters, (5, 5),
block_id='normal_left2_%s' % block_id)
x2_2 = _separable_conv_block(
p, filters, (3, 3),
block_id='normal_right2_%s' % block_id)
x2 = layers.add([x2_1, x2_2], name='normal_add_2_%s' % block_id)
with backend.name_scope('block_3'):
x3 = layers.AveragePooling2D(
(3, 3),
strides=(1, 1),
padding='same',
name='normal_left3_%s' % (block_id))(h)
x3 = layers.add([x3, p], name='normal_add_3_%s' % block_id)
with backend.name_scope('block_4'):
x4_1 = layers.AveragePooling2D(
(3, 3),
strides=(1, 1),
padding='same',
name='normal_left4_%s' % (block_id))(p)
x4_2 = layers.AveragePooling2D(
(3, 3),
strides=(1, 1),
padding='same',
name='normal_right4_%s' % (block_id))(p)
x4 = layers.add([x4_1, x4_2], name='normal_add_4_%s' % block_id)
with backend.name_scope('block_5'):
x5 = _separable_conv_block(h, filters,
block_id='normal_left5_%s' % block_id)
x5 = layers.add([x5, h], name='normal_add_5_%s' % block_id)
x = layers.concatenate([p, x1, x2, x3, x4, x5],
axis=channel_dim,
name='normal_concat_%s' % block_id)
return x, ip
def _reduction_a_cell(ip, p, filters, block_id=None):
'''Adds a Reduction cell for NASNet-A (Fig. 4 in the paper).
# Arguments
ip: Input tensor `x`
p: Input tensor `p`
filters: Number of output filters
block_id: String block_id
# Returns
A Keras tensor
'''
channel_dim = 1 if backend.image_data_format() == 'channels_first' else -1
with backend.name_scope('reduction_A_block_%s' % block_id):
p = _adjust_block(p, ip, filters, block_id)
h = layers.Activation('relu')(ip)
h = layers.Conv2D(
filters, (1, 1),
strides=(1, 1),
padding='same',
name='reduction_conv_1_%s' % block_id,
use_bias=False,
kernel_initializer='he_normal')(h)
h = layers.BatchNormalization(
axis=channel_dim,
momentum=0.9997,
epsilon=1e-3,
name='reduction_bn_1_%s' % block_id)(h)
h3 = layers.ZeroPadding2D(
padding=correct_pad(backend, h, 3),
name='reduction_pad_1_%s' % block_id)(h)
with backend.name_scope('block_1'):
x1_1 = _separable_conv_block(
h, filters, (5, 5),
strides=(2, 2),
block_id='reduction_left1_%s' % block_id)
x1_2 = _separable_conv_block(
p, filters, (7, 7),
strides=(2, 2),
block_id='reduction_right1_%s' % block_id)
x1 = layers.add([x1_1, x1_2], name='reduction_add_1_%s' % block_id)
with backend.name_scope('block_2'):
x2_1 = layers.MaxPooling2D(
(3, 3),
strides=(2, 2),
padding='valid',
name='reduction_left2_%s' % block_id)(h3)
x2_2 = _separable_conv_block(
p, filters, (7, 7),
strides=(2, 2),
block_id='reduction_right2_%s' % block_id)
x2 = layers.add([x2_1, x2_2], name='reduction_add_2_%s' % block_id)
with backend.name_scope('block_3'):
x3_1 = layers.AveragePooling2D(
(3, 3),
strides=(2, 2),
padding='valid',
name='reduction_left3_%s' % block_id)(h3)
x3_2 = _separable_conv_block(
p, filters, (5, 5),
strides=(2, 2),
block_id='reduction_right3_%s' % block_id)
x3 = layers.add([x3_1, x3_2], name='reduction_add3_%s' % block_id)
with backend.name_scope('block_4'):
x4 = layers.AveragePooling2D(
(3, 3),
strides=(1, 1),
padding='same',
name='reduction_left4_%s' % block_id)(x1)
x4 = layers.add([x2, x4])
with backend.name_scope('block_5'):
x5_1 = _separable_conv_block(
x1, filters, (3, 3),
block_id='reduction_left4_%s' % block_id)
x5_2 = layers.MaxPooling2D(
(3, 3),
strides=(2, 2),
padding='valid',
name='reduction_right5_%s' % block_id)(h3)
x5 = layers.add([x5_1, x5_2], name='reduction_add4_%s' % block_id)
x = layers.concatenate(
[x2, x3, x4, x5],
axis=channel_dim,
name='reduction_concat_%s' % block_id)
return x, ip
def preprocess_input(x, **kwargs):
"""Preprocesses a numpy array encoding a batch of images.
# Arguments
x: a 4D numpy array consists of RGB values within [0, 255].
# Returns
Preprocessed array.
"""
return imagenet_utils.preprocess_input(x, mode='tf', **kwargs)
|