# attn_layer = Conv2D(8, kernel_size = (1,1), padding = 'same', activation = 'relu')(attn_layer) # attn_layer = Conv2D(1, kernel_size = (1,1), padding = 'valid',activation = 'sigmoid')(attn_layer) # # fan it out to all of the channels # up_c2_w = np.ones((1, 1, 1, pt_depth)) # up_c2 = Conv2D(pt_depth, kernel_size = (1,1), padding = 'same',activation = 'linear', use_bias = False, weights = [up_c2_w]) # up_c2.trainable = False # attn_layer = up_c2(attn_layer) # mask_features = multiply([attn_layer, bn_features]) # gap_features = GlobalAveragePooling2D()(mask_features) # gap_mask = GlobalAveragePooling2D()(attn_layer) # to account for missing values from the attention model # gap = Lambda(lambda x: x[0]/x[1], name = 'RescaleGAP')([gap_features, gap_mask]) glb_max = GlobalMaxPooling2D()(bn_features) gap_dr = Dropout(0.25)(glb_max) dr_steps = Dropout(0.25)(Dense(512, activation=LeakyReLU(alpha=0.5))(gap_dr)) dr_steps = Dropout(0.25)(Dense(512, activation=LeakyReLU(alpha=0.5))(dr_steps)) dr_steps = Dropout(0.25)(Dense(512, activation='relu')(dr_steps)) dr_steps = Dropout(0.25)(Dense(512, activation='relu')(dr_steps)) out_layer = Dense(5, activation='softmax')(dr_steps) model = Model(inputs=[in_lay], outputs=[out_layer]) # from keras.metrics import top_k_categorical_accuracy # def top_2_accuracy(in_gt, in_pred): # return top_k_categorical_accuracy(in_gt, in_pred, k=2) # retina_model.compile(optimizer = 'adam', loss = 'categorical_crossentropy', # metrics = ['categorical_accuracy', top_2_accuracy])
def InceptionResNetV2(include_top=True,
weights='imagenet',
input_tensor=None,
input_shape=None,
pooling=None,
classes=1000,
dropout_keep_prob=0.8):
"""Instantiates the Inception-ResNet v2 architecture.
Optionally loads weights pre-trained on ImageNet.
Note that when using TensorFlow, for best performance you should
set `"image_data_format": "channels_last"` in your Keras config
at `~/.keras/keras.json`.
The model and the weights are compatible with both TensorFlow and Theano.
The data format convention used by the model is the one specified in your
Keras config file.
Note that the default input image size for this model is 299x299, instead
of 224x224 as in the VGG16 and ResNet models. Also, the input preprocessing
function is different (i.e., do not use `imagenet_utils.preprocess_input()`
with this model. Use `preprocess_input()` defined in this module instead).
# Arguments
include_top: whether to include the fully-connected
layer at the top of the network.
weights: one of `None` (random initialization)
or `'imagenet'` (pre-training on ImageNet).
input_tensor: optional Keras tensor (i.e. output of `layers.Input()`)
to use as image input for the model.
input_shape: optional shape tuple, only to be specified
if `include_top` is `False` (otherwise the input shape
has to be `(299, 299, 3)` (with `channels_last` data format)
or `(3, 299, 299)` (with `channels_first` data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 139.
E.g. `(150, 150, 3)` would be one valid value.
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.
dropout_keep_prob: dropout keep rate after pooling and before the
classification layer, only to be specified if `include_top` is `True`.
# Returns
A Keras `Model` instance.
# Raises
ValueError: in case of invalid argument for `weights`,
or invalid input shape.
"""
if weights not in {'imagenet', None}:
raise ValueError('The `weights` argument should be either '
'`None` (random initialization) or `imagenet` '
'(pre-training on ImageNet).')
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')
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=299,
min_size=139,
data_format=K.image_data_format(),
require_flatten=False,
weights=weights)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
# Stem block: 35 x 35 x 192
x = conv2d_bn(img_input,
32,
3,
strides=2,
padding='valid',
name='Conv2d_1a_3x3')
x = conv2d_bn(x, 32, 3, padding='valid', name='Conv2d_2a_3x3')
x = conv2d_bn(x, 64, 3, name='Conv2d_2b_3x3')
x = MaxPooling2D(3, strides=2, name='MaxPool_3a_3x3')(x)
x = conv2d_bn(x, 80, 1, padding='valid', name='Conv2d_3b_1x1')
x = conv2d_bn(x, 192, 3, padding='valid', name='Conv2d_4a_3x3')
x = MaxPooling2D(3, strides=2, name='MaxPool_5a_3x3')(x)
# Mixed 5b (Inception-A block): 35 x 35 x 320
channel_axis = 1 if K.image_data_format() == 'channels_first' else 3
name_fmt = partial(_generate_layer_name, prefix='Mixed_5b')
branch_0 = conv2d_bn(x, 96, 1, name=name_fmt('Conv2d_1x1', 0))
branch_1 = conv2d_bn(x, 48, 1, name=name_fmt('Conv2d_0a_1x1', 1))
branch_1 = conv2d_bn(branch_1, 64, 5, name=name_fmt('Conv2d_0b_5x5', 1))
branch_2 = conv2d_bn(x, 64, 1, name=name_fmt('Conv2d_0a_1x1', 2))
branch_2 = conv2d_bn(branch_2, 96, 3, name=name_fmt('Conv2d_0b_3x3', 2))
branch_2 = conv2d_bn(branch_2, 96, 3, name=name_fmt('Conv2d_0c_3x3', 2))
branch_pool = AveragePooling2D(3,
strides=1,
padding='same',
name=name_fmt('AvgPool_0a_3x3', 3))(x)
branch_pool = conv2d_bn(branch_pool,
64,
1,
name=name_fmt('Conv2d_0b_1x1', 3))
branches = [branch_0, branch_1, branch_2, branch_pool]
x = Concatenate(axis=channel_axis, name='Mixed_5b')(branches)
# 10x Block35 (Inception-ResNet-A block): 35 x 35 x 320
for block_idx in range(1, 11):
x = _inception_resnet_block(x,
scale=0.17,
block_type='Block35',
block_idx=block_idx)
# Mixed 6a (Reduction-A block): 17 x 17 x 1088
name_fmt = partial(_generate_layer_name, prefix='Mixed_6a')
branch_0 = conv2d_bn(x,
384,
3,
strides=2,
padding='valid',
name=name_fmt('Conv2d_1a_3x3', 0))
branch_1 = conv2d_bn(x, 256, 1, name=name_fmt('Conv2d_0a_1x1', 1))
branch_1 = conv2d_bn(branch_1, 256, 3, name=name_fmt('Conv2d_0b_3x3', 1))
branch_1 = conv2d_bn(branch_1,
384,
3,
strides=2,
padding='valid',
name=name_fmt('Conv2d_1a_3x3', 1))
branch_pool = MaxPooling2D(3,
strides=2,
padding='valid',
name=name_fmt('MaxPool_1a_3x3', 2))(x)
branches = [branch_0, branch_1, branch_pool]
x = Concatenate(axis=channel_axis, name='Mixed_6a')(branches)
# 20x Block17 (Inception-ResNet-B block): 17 x 17 x 1088
for block_idx in range(1, 21):
x = _inception_resnet_block(x,
scale=0.1,
block_type='Block17',
block_idx=block_idx)
# Mixed 7a (Reduction-B block): 8 x 8 x 2080
name_fmt = partial(_generate_layer_name, prefix='Mixed_7a')
branch_0 = conv2d_bn(x, 256, 1, name=name_fmt('Conv2d_0a_1x1', 0))
branch_0 = conv2d_bn(branch_0,
384,
3,
strides=2,
padding='valid',
name=name_fmt('Conv2d_1a_3x3', 0))
branch_1 = conv2d_bn(x, 256, 1, name=name_fmt('Conv2d_0a_1x1', 1))
branch_1 = conv2d_bn(branch_1,
288,
3,
strides=2,
padding='valid',
name=name_fmt('Conv2d_1a_3x3', 1))
branch_2 = conv2d_bn(x, 256, 1, name=name_fmt('Conv2d_0a_1x1', 2))
branch_2 = conv2d_bn(branch_2, 288, 3, name=name_fmt('Conv2d_0b_3x3', 2))
branch_2 = conv2d_bn(branch_2,
320,
3,
strides=2,
padding='valid',
name=name_fmt('Conv2d_1a_3x3', 2))
branch_pool = MaxPooling2D(3,
strides=2,
padding='valid',
name=name_fmt('MaxPool_1a_3x3', 3))(x)
branches = [branch_0, branch_1, branch_2, branch_pool]
x = Concatenate(axis=channel_axis, name='Mixed_7a')(branches)
# 10x Block8 (Inception-ResNet-C block): 8 x 8 x 2080
for block_idx in range(1, 10):
x = _inception_resnet_block(x,
scale=0.2,
block_type='Block8',
block_idx=block_idx)
x = _inception_resnet_block(x,
scale=1.,
activation=None,
block_type='Block8',
block_idx=10)
# Final convolution block
x = conv2d_bn(x, 1536, 1, name='Conv2d_7b_1x1')
if include_top:
# Classification block
x = GlobalAveragePooling2D(name='AvgPool')(x)
x = Dropout(1.0 - dropout_keep_prob, name='Dropout')(x)
x = Dense(classes, name='Logits')(x)
x = Activation('softmax', name='Predictions')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D(name='AvgPool')(x)
elif pooling == 'max':
x = GlobalMaxPooling2D(name='MaxPool')(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model
model = Model(inputs, x, name='inception_resnet_v2')
# Load weights
if weights == 'imagenet':
if K.image_data_format() == 'channels_first':
if K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
if include_top:
weights_filename = 'inception_resnet_v2_weights_tf_dim_ordering_tf_kernels.h5'
weights_path = get_file(
weights_filename,
BASE_WEIGHT_URL + weights_filename,
cache_subdir='models',
md5_hash='e693bd0210a403b3192acc6073ad2e96')
else:
weights_filename = 'inception_resnet_v2_weights_tf_dim_ordering_tf_kernels_notop.h5'
weights_path = get_file(
weights_filename,
BASE_WEIGHT_URL + weights_filename,
cache_subdir='models',
md5_hash='d19885ff4a710c122648d3b5c3b684e4')
model.load_weights(weights_path)
return model
def EfficientNet(width_coefficient,
depth_coefficient,
default_size,
dropout_rate=0.2,
drop_connect_rate=0.2,
depth_divisor=8,
activation_fn=swish,
blocks_args=DEFAULT_BLOCKS_ARGS,
model_name='efficientnet',
include_top=True,
weights='imagenet',
input_tensor=None,
input_shape=None,
pooling=None,
num_classes=1000,
**kwargs):
# Instantiates the EfficientNet architecture using given scaling coefficients.
"""
# Arguments
width_coefficient: float, scaling coefficient for network width.
depth_coefficient: float, scaling coefficient for network depth.
default_size: integer, default input image size.
dropout_rate: float, dropout rate before final classifier layer.
drop_connect_rate: float, dropout rate at skip connections.
depth_divisor: integer, a unit of network width.
activation_fn: activation function.
blocks_args: list of dicts, parameters to construct block modules.
model_name: string, model name.
include_top: whether to include the FC layer at the top of the network.
weights: `None` (random initialization), 'imagenet' or the path to any weights.
input_tensor: optional Keras tensor (output of `layers.Input()`)
input_shape: tuple, only to be specified if `include_top` is False.
pooling: Optional mode for feature extraction when `include_top` is `False`.
- `None`: the output of model is the 4D tensor of the last conv layer
- `avg` means global average pooling and the output as a 2D tensor.
- `max` means global max pooling will be applied.
num_classes: specified if `include_top` is True
# Returns
A Keras model instance.
# Raises
ValueError: in case of invalid argument for `weights` or invalid input shape.
"""
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 num_classes != 1000:
raise ValueError(
'If using `weights` as `"imagenet"` with `include_top`'
' as true, `classes` should be 1000')
# Determine the proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=default_size,
min_size=32,
data_format=K.image_data_format(),
require_flatten=include_top,
weights=weights)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
bn_axis = -1 if K.image_data_format() == 'channels_last' else 1
def round_filters(filters, divisor=depth_divisor):
# Round number of filters based on depth multiplier.
filters *= width_coefficient
new_filters = max(divisor,
int(filters + divisor / 2) // divisor * divisor)
# Ensure the round-down does not go down by more than 10%.
if new_filters < 0.9 * filters:
new_filters += divisor
return int(new_filters)
def round_repeats(repeats):
# Round number of repeats based on depth multiplier.
return int(math.ceil(depth_coefficient * repeats))
# Build the stem
x = img_input
x = ZeroPadding2D(padding=correct_pad(K, x, 3), name='stem_conv_pad')(x)
x = Conv2D(round_filters(32),
kernel_size=(3, 3),
strides=2,
padding='valid',
use_bias=False,
kernel_initializer=CONV_KERNEL_INITIALIZER,
name='stem_conv')(x)
x = BatchNormalization(axis=bn_axis, name='stem_bn')(x)
x = Activation(activation_fn, name='stem_activation')(x)
# Build the blocks
from copy import deepcopy
blocks_args = deepcopy(blocks_args) # See the above blocks_args
b = 0
blocks = float(sum(args['repeats'] for args in blocks_args))
for (i, args) in enumerate(blocks_args):
assert args['repeats'] > 0
# Update the block input and output filters based on depth multiplier.
args['filters_in'] = round_filters(args['filters_in'])
args['filters_out'] = round_filters(args['filters_out'])
for j in range(round_repeats(args.pop('repeats'))):
# The first block needs to take care of stride and filter size growth.
if j > 0:
args['strides'] = 1
args['filters_in'] = args['filters_out']
x = block(x,
activation_fn,
drop_connect_rate * b / blocks,
name='block{}{}_'.format(i + 1, chr(j + 97)),
**args)
b += 1
# Build the top
x = Conv2D(round_filters(1280),
kernel_size=(1, 1),
padding='same',
use_bias=False,
kernel_initializer=CONV_KERNEL_INITIALIZER,
name='top_conv')(x)
x = BatchNormalization(axis=bn_axis, name='top_bn')(x)
x = Activation(activation_fn, name='top_activation')(x)
if include_top:
x = GlobalAveragePooling2D(name='avg_pool')(x)
if dropout_rate > 0:
x = Dropout(dropout_rate, name='top_dropout')(x)
x = Dense(num_classes,
activation='softmax',
kernel_initializer=DENSE_KERNEL_INITIALIZER,
name='probs')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D(name='avg_pool')(x)
elif pooling == 'max':
x = GlobalMaxPooling2D(name='max_pool')(x)
# Ensure the model considers any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Build the model.
model = Model(inputs, x, name=model_name)
# Load weights.
if weights == 'imagenet':
if include_top:
file_suff = '_weights_tf_dim_ordering_tf_kernels_autoaugment.h5'
file_hash = WEIGHTS_HASHES[model_name[-2:]][0]
else:
file_suff = '_weights_tf_dim_ordering_tf_kernels_autoaugment_notop.h5'
file_hash = WEIGHTS_HASHES[model_name[-2:]][1]
file_name = model_name + file_suff
weights_path = get_file(file_name,
BASE_WEIGHTS_PATH + file_name,
cache_subdir='models',
file_hash=file_hash)
model.load_weights(weights_path)
elif weights is not None:
model.load_weights(weights)
return model
def ResNet(input_shape=None, classes=10, block='bottleneck', residual_unit='v2', repetitions=None,
initial_filters=64, activation='softmax', include_top=True, input_tensor=None, dropout=None,
transition_dilation_rate=(1, 1), initial_strides=(2, 2), initial_kernel_size=(7, 7),
initial_pooling='max', final_pooling=None, top='classification'):
"""Builds a custom ResNet like architecture. Defaults to ResNet50 v2.
Args:
input_shape: optional shape tuple, only to be specified
if `include_top` is False (otherwise the input shape
has to be `(224, 224, 3)` (with `channels_last` dim ordering)
or `(3, 224, 224)` (with `channels_first` dim ordering).
It should have exactly 3 dimensions,
and width and height should be no smaller than 8.
E.g. `(224, 224, 3)` would be one valid value.
classes: The number of outputs at final softmax layer
block: The block function to use. This is either `'basic'` or `'bottleneck'`.
The original paper used `basic` for layers < 50.
repetitions: Number of repetitions of various block units.
At each block unit, the number of filters are doubled and the input size is halved.
Default of None implies the ResNet50v2 values of [3, 4, 6, 3].
residual_unit: the basic residual unit, 'v1' for conv bn relu, 'v2' for bn relu conv.
See [Identity Mappings in Deep Residual Networks](https://arxiv.org/abs/1603.05027)
for details.
dropout: None for no dropout, otherwise rate of dropout from 0 to 1.
Based on [Wide Residual Networks.(https://arxiv.org/pdf/1605.07146) paper.
transition_dilation_rate: Dilation rate for transition layers. For semantic
segmentation of images use a dilation rate of (2, 2).
initial_strides: Stride of the very first residual unit and MaxPooling2D call,
with default (2, 2), set to (1, 1) for small images like cifar.
initial_kernel_size: kernel size of the very first convolution, (7, 7) for imagenet
and (3, 3) for small image datasets like tiny imagenet and cifar.
See [ResNeXt](https://arxiv.org/abs/1611.05431) paper for details.
initial_pooling: Determine if there will be an initial pooling layer,
'max' for imagenet and None for small image datasets.
See [ResNeXt](https://arxiv.org/abs/1611.05431) paper for details.
final_pooling: Optional pooling mode for feature extraction at the final model layer
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.
top: Defines final layers to evaluate based on a specific problem type. Options are
'classification' for ImageNet style problems, 'segmentation' for problems like
the Pascal VOC dataset, and None to exclude these layers entirely.
Returns:
The keras `Model`.
"""
if activation not in ['softmax', 'sigmoid', None]:
raise ValueError('activation must be one of "softmax", "sigmoid", or None')
if activation == 'sigmoid' and classes != 1:
raise ValueError('sigmoid activation can only be used when classes = 1')
if repetitions is None:
repetitions = [3, 4, 6, 3]
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=32,
min_size=8,
data_format=K.image_data_format(),
require_flatten=include_top)
_handle_dim_ordering()
if len(input_shape) != 3:
raise Exception("Input shape should be a tuple (nb_channels, nb_rows, nb_cols)")
if block == 'basic':
block_fn = basic_block
elif block == 'bottleneck':
block_fn = bottleneck
elif isinstance(block, six.string_types):
block_fn = _string_to_function(block)
else:
block_fn = block
if residual_unit == 'v2':
residual_unit = _bn_relu_conv
elif residual_unit == 'v1':
residual_unit = _conv_bn_relu
elif isinstance(residual_unit, six.string_types):
residual_unit = _string_to_function(residual_unit)
else:
residual_unit = residual_unit
# Permute dimension order if necessary
if K.image_data_format() == 'channels_first':
input_shape = (input_shape[1], input_shape[2], input_shape[0])
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=32,
min_size=8,
data_format=K.image_data_format(),
require_flatten=include_top)
img_input = Input(shape=input_shape, tensor=input_tensor)
x = _conv_bn_relu(filters=initial_filters, kernel_size=initial_kernel_size, strides=initial_strides)(img_input)
if initial_pooling == 'max':
x = MaxPooling2D(pool_size=(3, 3), strides=initial_strides, padding="same")(x)
block = x
filters = initial_filters
for i, r in enumerate(repetitions):
transition_dilation_rates = [transition_dilation_rate] * r
transition_strides = [(1, 1)] * r
if transition_dilation_rate == (1, 1):
transition_strides[0] = (2, 2)
block = _residual_block(block_fn, filters=filters,
stage=i, blocks=r,
is_first_layer=(i == 0),
dropout=dropout,
transition_dilation_rates=transition_dilation_rates,
transition_strides=transition_strides,
residual_unit=residual_unit)(block)
filters *= 2
# Last activation
x = _bn_relu(block)
# Classifier block
if include_top and top is 'classification':
x = GlobalAveragePooling2D()(x)
x = Dense(units=classes, activation=activation, kernel_initializer="he_normal")(x)
elif include_top and top is 'segmentation':
x = Conv2D(classes, (1, 1), activation='linear', padding='same')(x)
if K.image_data_format() == 'channels_first':
channel, row, col = input_shape
else:
row, col, channel = input_shape
x = Reshape((row * col, classes))(x)
x = Activation(activation)(x)
x = Reshape((row, col, classes))(x)
elif final_pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif final_pooling == 'max':
x = GlobalMaxPooling2D()(x)
model = Model(inputs=img_input, outputs=x)
return model
def VGG16(include_top=True,
weights='imagenet',
input_tensor=None,
input_shape=None,
pooling=None,
classes=1000):
if weights not in {'imagenet', None}:
raise ValueError('The `weights` argument should be either '
'`None` (random initialization) or `imagenet` '
'(pre-training on ImageNet).')
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')
#determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=224,
min_size=48,
data_format=K.image_data_format(),
include_top=include_top)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
#Block_1
x = Conv2D(64, (3, 3),
activation='relu',
padding='same',
name='block1_conv1')(img_input)
x = Conv2D(64, (3, 3),
activation='relu',
padding='same',
name='block1_conv2')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block1_pool')(x)
#Block_2
x = Conv2D(128, (3, 3),
activation='relu',
padding='same',
name='block2_conv1')(x)
x = Conv2D(128, (3, 3),
activation='relu',
padding='same',
name='block2_conv2')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block2_pool')(x)
#Block_3
x = Conv2D(256, (3, 3),
activation='relu',
padding='same',
name='block3_conv1')(x)
x = Conv2D(256, (3, 3),
activation='relu',
padding='same',
name='block3_conv2')(x)
x = Conv2D(256, (3, 3),
activation='relu',
padding='same',
name='block3_conv3')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block3_pool')(x)
#Block_4
x = Conv2D(512, (3, 3),
activation='relu',
padding='same',
name='block4_conv1')(x)
x = Conv2D(512, (3, 3),
activation='relu',
padding='same',
name='block4_conv2')(x)
x = Conv2D(512, (3, 3),
activation='relu',
padding='same',
name='block4_conv3')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block4_pool')(x)
#Block_5
x = Conv2D(512, (3, 3),
activation='relu',
padding='same',
name='block5_conv1')(x)
x = Conv2D(512, (3, 3),
activation='relu',
padding='same',
name='block5_conv2')(x)
x = Conv2D(512, (3, 3),
activation='relu',
padding='same',
name='block5_conv3')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block5_pool')(x)
if include_top:
x = Flatten(name='flatten')(x)
x = Dense(4096, activation='relu', name='fc1')(x)
x = Dense(4096, activation='relu', name='fc2')(x)
x = Dense(classes, activation='softmax', name='predictions')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
#Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
#create model
model = Model(inputs, x, name='vgg16')
#load weights
if weights == 'imagenet':
if include_top:
weights_path = get_file(
'vgg16_weights_tf_dim_ordering_tf_kernels.h5',
WEIGHTS_PATH,
cache_subdir='models')
else:
weights_path = get_file(
'vgg16_weights_tf_dim_ordering_tf_kernels_notop.h5',
WEIGHTS_PATH_NO_TOP,
cache_subdir='models')
model.load_weights(weights_path)
if K.backend() == 'theano':
layer_utils.convert_all_kernels_in_model(model)
if K.image_data_format() == 'channels_first':
if include_top:
maxpool = model.get_layer(name='block5_pool')
shape = maxpool.output_shape[1:]
dense = model.get_layer(name='fc1')
layer_utils.convert_dense_weights_data_format(
dense, shape, 'channels_first')
if K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
return model
def VGG16(include_top=True, weights='vggface',
input_tensor=None, input_shape=None,
pooling=None,
classes=2622):
input_shape = _obtain_input_shape(input_shape,
default_size=224,
min_size=48,
data_format=K.image_data_format(),
require_flatten=include_top)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
# Block 1
x = Conv2D(64, (3, 3), activation='relu', padding='same', name='conv1_1')(
img_input)
x = Conv2D(64, (3, 3), activation='relu', padding='same', name='conv1_2')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='pool1')(x)
# Block 2
x = Conv2D(128, (3, 3), activation='relu', padding='same', name='conv2_1')(
x)
x = Conv2D(128, (3, 3), activation='relu', padding='same', name='conv2_2')(
x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='pool2')(x)
# Block 3
x = Conv2D(256, (3, 3), activation='relu', padding='same', name='conv3_1')(
x)
x = Conv2D(256, (3, 3), activation='relu', padding='same', name='conv3_2')(
x)
x = Conv2D(256, (3, 3), activation='relu', padding='same', name='conv3_3')(
x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='pool3')(x)
# Block 4
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='conv4_1')(
x)
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='conv4_2')(
x)
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='conv4_3')(
x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='pool4')(x)
# Block 5
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='conv5_1')(
x)
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='conv5_2')(
x)
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='conv5_3')(
x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='pool5')(x)
if include_top:
# Classification block
x = Flatten(name='flatten')(x)
x = Dense(4096, name='fc6')(x)
x = Activation('relu', name='fc6/relu')(x)
x = Dense(4096, name='fc7')(x)
x = Activation('relu', name='fc7/relu')(x)
x = Dense(classes, name='fc8')(x)
x = Activation('softmax', name='fc8/softmax')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model.
model = Model(inputs, x, name='vggface_vgg16') # load weights
if weights == 'vggface':
if include_top:
weights_path = get_file('rcmalli_vggface_tf_vgg16.h5',
utils.
VGG16_WEIGHTS_PATH,
cache_subdir=utils.VGGFACE_DIR)
else:
weights_path = get_file('rcmalli_vggface_tf_notop_vgg16.h5',
utils.VGG16_WEIGHTS_PATH_NO_TOP,
cache_subdir=utils.VGGFACE_DIR)
model.load_weights(weights_path, by_name=True)
if K.backend() == 'theano':
layer_utils.convert_all_kernels_in_model(model)
if K.image_data_format() == 'channels_first':
if include_top:
maxpool = model.get_layer(name='pool5')
shape = maxpool.output_shape[1:]
dense = model.get_layer(name='fc6')
layer_utils.convert_dense_weights_data_format(dense, shape,
'channels_first')
if K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
return model
def InceptionV3(include_top=True,
weights='imagenet',
input_tensor=None,
input_shape=None,
pooling=None,
classes=1000):
"""Instantiates the Inception v3 architecture.
Optionally loads weights pre-trained
on ImageNet. Note that when using TensorFlow,
for best performance you should set
`image_data_format="channels_last"` in your Keras config
at ~/.keras/keras.json.
The model and the weights are compatible with both
TensorFlow and Theano. The data format
convention used by the model is the one
specified in your Keras config file.
Note that the default input image size for this model is 299x299.
Arguments:
include_top: whether to include the fully-connected
layer at the top of the network.
weights: one of `None` (random initialization)
or "imagenet" (pre-training on ImageNet).
input_tensor: optional Keras tensor (i.e. output of `layers.Input()`)
to use as image input for the model.
input_shape: optional shape tuple, only to be specified
if `include_top` is False (otherwise the input shape
has to be `(299, 299, 3)` (with `channels_last` data format)
or `(3, 299, 299)` (with `channels_first` data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 139.
E.g. `(150, 150, 3)` would be one valid value.
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.
"""
if weights not in {'imagenet', None}:
raise ValueError('The `weights` argument should be either '
'`None` (random initialization) or `imagenet` '
'(pre-training on ImageNet).')
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')
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=299,
min_size=139,
data_format=K.image_data_format(),
include_top=include_top)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
img_input = Input(tensor=input_tensor, shape=input_shape)
if K.image_data_format() == 'channels_first':
channel_axis = 1
else:
channel_axis = 3
x = conv2d_bn(img_input, 32, 3, 3, strides=(2, 2), padding='valid')
x = conv2d_bn(x, 32, 3, 3, padding='valid')
x = conv2d_bn(x, 64, 3, 3)
x = MaxPooling2D((3, 3), strides=(2, 2))(x)
x = conv2d_bn(x, 80, 1, 1, padding='valid')
x = conv2d_bn(x, 192, 3, 3, padding='valid')
x = MaxPooling2D((3, 3), strides=(2, 2))(x)
# mixed 0, 1, 2: 35 x 35 x 256
branch1x1 = conv2d_bn(x, 64, 1, 1)
branch5x5 = conv2d_bn(x, 48, 1, 1)
branch5x5 = conv2d_bn(branch5x5, 64, 5, 5)
branch3x3dbl = conv2d_bn(x, 64, 1, 1)
branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1), padding='same')(x)
branch_pool = conv2d_bn(branch_pool, 32, 1, 1)
x = layers.concatenate([branch1x1, branch5x5, branch3x3dbl, branch_pool],
axis=channel_axis,
name='mixed0')
# mixed 1: 35 x 35 x 256
branch1x1 = conv2d_bn(x, 64, 1, 1)
branch5x5 = conv2d_bn(x, 48, 1, 1)
branch5x5 = conv2d_bn(branch5x5, 64, 5, 5)
branch3x3dbl = conv2d_bn(x, 64, 1, 1)
branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1), padding='same')(x)
branch_pool = conv2d_bn(branch_pool, 64, 1, 1)
x = layers.concatenate([branch1x1, branch5x5, branch3x3dbl, branch_pool],
axis=channel_axis,
name='mixed1')
# mixed 2: 35 x 35 x 256
branch1x1 = conv2d_bn(x, 64, 1, 1)
branch5x5 = conv2d_bn(x, 48, 1, 1)
branch5x5 = conv2d_bn(branch5x5, 64, 5, 5)
branch3x3dbl = conv2d_bn(x, 64, 1, 1)
branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1), padding='same')(x)
branch_pool = conv2d_bn(branch_pool, 64, 1, 1)
x = layers.concatenate([branch1x1, branch5x5, branch3x3dbl, branch_pool],
axis=channel_axis,
name='mixed2')
# mixed 3: 17 x 17 x 768
branch3x3 = conv2d_bn(x, 384, 3, 3, strides=(2, 2), padding='valid')
branch3x3dbl = conv2d_bn(x, 64, 1, 1)
branch3x3dbl = conv2d_bn(branch3x3dbl, 96, 3, 3)
branch3x3dbl = conv2d_bn(branch3x3dbl,
96,
3,
3,
strides=(2, 2),
padding='valid')
branch_pool = MaxPooling2D((3, 3), strides=(2, 2))(x)
x = layers.concatenate([branch3x3, branch3x3dbl, branch_pool],
axis=channel_axis,
name='mixed3')
# mixed 4: 17 x 17 x 768
branch1x1 = conv2d_bn(x, 192, 1, 1)
branch7x7 = conv2d_bn(x, 128, 1, 1)
branch7x7 = conv2d_bn(branch7x7, 128, 1, 7)
branch7x7 = conv2d_bn(branch7x7, 192, 7, 1)
branch7x7dbl = conv2d_bn(x, 128, 1, 1)
branch7x7dbl = conv2d_bn(branch7x7dbl, 128, 7, 1)
branch7x7dbl = conv2d_bn(branch7x7dbl, 128, 1, 7)
branch7x7dbl = conv2d_bn(branch7x7dbl, 128, 7, 1)
branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 1, 7)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1), padding='same')(x)
branch_pool = conv2d_bn(branch_pool, 192, 1, 1)
x = layers.concatenate([branch1x1, branch7x7, branch7x7dbl, branch_pool],
axis=channel_axis,
name='mixed4')
# mixed 5, 6: 17 x 17 x 768
for i in range(2):
branch1x1 = conv2d_bn(x, 192, 1, 1)
branch7x7 = conv2d_bn(x, 160, 1, 1)
branch7x7 = conv2d_bn(branch7x7, 160, 1, 7)
branch7x7 = conv2d_bn(branch7x7, 192, 7, 1)
branch7x7dbl = conv2d_bn(x, 160, 1, 1)
branch7x7dbl = conv2d_bn(branch7x7dbl, 160, 7, 1)
branch7x7dbl = conv2d_bn(branch7x7dbl, 160, 1, 7)
branch7x7dbl = conv2d_bn(branch7x7dbl, 160, 7, 1)
branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 1, 7)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1),
padding='same')(x)
branch_pool = conv2d_bn(branch_pool, 192, 1, 1)
x = layers.concatenate(
[branch1x1, branch7x7, branch7x7dbl, branch_pool],
axis=channel_axis,
name='mixed' + str(5 + i))
# mixed 7: 17 x 17 x 768
branch1x1 = conv2d_bn(x, 192, 1, 1)
branch7x7 = conv2d_bn(x, 192, 1, 1)
branch7x7 = conv2d_bn(branch7x7, 192, 1, 7)
branch7x7 = conv2d_bn(branch7x7, 192, 7, 1)
branch7x7dbl = conv2d_bn(x, 192, 1, 1)
branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 7, 1)
branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 1, 7)
branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 7, 1)
branch7x7dbl = conv2d_bn(branch7x7dbl, 192, 1, 7)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1), padding='same')(x)
branch_pool = conv2d_bn(branch_pool, 192, 1, 1)
x = layers.concatenate([branch1x1, branch7x7, branch7x7dbl, branch_pool],
axis=channel_axis,
name='mixed7')
# mixed 8: 8 x 8 x 1280
branch3x3 = conv2d_bn(x, 192, 1, 1)
branch3x3 = conv2d_bn(branch3x3,
320,
3,
3,
strides=(2, 2),
padding='valid')
branch7x7x3 = conv2d_bn(x, 192, 1, 1)
branch7x7x3 = conv2d_bn(branch7x7x3, 192, 1, 7)
branch7x7x3 = conv2d_bn(branch7x7x3, 192, 7, 1)
branch7x7x3 = conv2d_bn(branch7x7x3,
192,
3,
3,
strides=(2, 2),
padding='valid')
branch_pool = MaxPooling2D((3, 3), strides=(2, 2))(x)
x = layers.concatenate([branch3x3, branch7x7x3, branch_pool],
axis=channel_axis,
name='mixed8')
# mixed 9: 8 x 8 x 2048
for i in range(2):
branch1x1 = conv2d_bn(x, 320, 1, 1)
branch3x3 = conv2d_bn(x, 384, 1, 1)
branch3x3_1 = conv2d_bn(branch3x3, 384, 1, 3)
branch3x3_2 = conv2d_bn(branch3x3, 384, 3, 1)
branch3x3 = layers.concatenate([branch3x3_1, branch3x3_2],
axis=channel_axis,
name='mixed9_' + str(i))
branch3x3dbl = conv2d_bn(x, 448, 1, 1)
branch3x3dbl = conv2d_bn(branch3x3dbl, 384, 3, 3)
branch3x3dbl_1 = conv2d_bn(branch3x3dbl, 384, 1, 3)
branch3x3dbl_2 = conv2d_bn(branch3x3dbl, 384, 3, 1)
branch3x3dbl = layers.concatenate([branch3x3dbl_1, branch3x3dbl_2],
axis=channel_axis)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1),
padding='same')(x)
branch_pool = conv2d_bn(branch_pool, 192, 1, 1)
x = layers.concatenate(
[branch1x1, branch3x3, branch3x3dbl, branch_pool],
axis=channel_axis,
name='mixed' + str(9 + i))
if include_top:
# Classification block
x = GlobalAveragePooling2D(name='avg_pool')(x)
x = Dense(classes, activation='softmax', name='predictions')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model.
model = Model(inputs, x, name='inception_v3')
# load weights
if weights == 'imagenet':
if K.image_data_format() == 'channels_first':
if K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
if include_top:
weights_path = get_file(
'inception_v3_weights_tf_dim_ordering_tf_kernels.h5',
WEIGHTS_PATH,
cache_subdir='models',
md5_hash='9a0d58056eeedaa3f26cb7ebd46da564')
else:
weights_path = get_file(
'inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5',
WEIGHTS_PATH_NO_TOP,
cache_subdir='models',
md5_hash='bcbd6486424b2319ff4ef7d526e38f63')
model.load_weights(weights_path)
if K.backend() == 'theano':
convert_all_kernels_in_model(model)
return model
def test_resnet():
# for reproducibility
rand_seed = 1337
np.random.seed(rand_seed)
out_size = (49, 2)
roi_size = 128
input_shape = (roi_size, roi_size, 1)
img_input = Input(shape=input_shape)
x = Conv2D(32, (3, 3), padding='same', use_bias=False, name='conv0')(img_input)
x = BatchNormalization(name='conv0a_bn')(x)
x = Activation('relu', name='conv0_act')(x)
x = MaxPooling2D((2, 2), name='conv0_pool')(x)
x = Conv2D(64, (3, 3), padding='same', use_bias=False, name='conv1a')(x)
x = BatchNormalization(name='conv1a_bn')(x)
x = Activation('relu', name='conv1a_act')(x)
residual = Conv2D(128, (1, 1), strides=(2, 2),
padding='same', use_bias=False)(x)
residual = BatchNormalization()(residual)
x = Conv2D(128, (3, 3), padding='same', use_bias=False, name='block2_sepconv1')(x)
x = BatchNormalization(name='block2_sepconv1_bn')(x)
x = Activation('relu', name='block2_sepconv2_act')(x)
x = Conv2D(128, (3, 3), padding='same', use_bias=False, name='block2_sepconv2')(x)
x = BatchNormalization(name='block2_sepconv2_bn')(x)
x = MaxPooling2D((2, 2), padding='same', name='block2_pool')(x)
x = layers.add([x, residual])
residual = Conv2D(512, (1, 1), strides=(2, 2),
padding='same', use_bias=False)(x)
residual = BatchNormalization()(residual)
x = Activation('relu', name='block3_sepconv1_act')(x)
x = Conv2D(256, (3, 3), padding='same', use_bias=False, name='block3_sepconv1')(x)
x = BatchNormalization(name='block3_sepconv1_bn')(x)
x = Activation('relu', name='block3_sepconv2_act')(x)
x = Conv2D(512, (3, 3), padding='same', use_bias=False, name='block3_sepconv2')(x)
x = BatchNormalization(name='block3_sepconv2_bn')(x)
x = MaxPooling2D((2, 2), padding='same', name='block3_pool')(x)
x = layers.add([x, residual])
x = Conv2D(1024, (3, 3), padding='same', use_bias=False, name='block14_sepconv2')(x)
x = BatchNormalization(name='block14_sepconv2_bn')(x)
x = Activation('relu', name='block14_sepconv2_act')(x)
x = GlobalMaxPooling2D(name='avg_pool')(x)
#layers seems to be key to capture the worm shape
#i need to use ELU instead of RELU otherwise the skeletons converges to 0
x = Dense(1024, name='dense0', activation='elu')(x)
x = Dropout(0.4)(x)
x = Dense(512, name='dense1', activation='elu')(x)
x = Dropout(0.4)(x)
x = Dense(256, name='dense2', activation='elu')(x)
x = Dropout(0.2)(x)
x = Dense(128, name='dense3', activation='elu')(x)
x = Dropout(0.1)(x)
x = Dense(out_size[0]*out_size[1], activation='elu', name='skeleton')(x)
x = Reshape((out_size))(x)
model = Model(img_input, x)
optimizer = Adam(lr=1e-3)#, decay=0.05)
model.compile(loss='mean_absolute_error',
optimizer=optimizer,
metrics=['mean_absolute_error', 'mean_squared_error', 'mean_absolute_percentage_error'])
return model
def test_simple_v2(out_size = (49, 2), roi_size = 128):
# for reproducibility
rand_seed = 1337
np.random.seed(rand_seed)
input_shape = (roi_size, roi_size, 1)
img_input = Input(shape=input_shape)
x = Conv2D(32, (3, 3), padding='same', name='conv0')(img_input)
x = Activation('relu', name='conv0_act')(x)
x = MaxPooling2D((2, 2), name='conv0_pool')(x)
x = Conv2D(64, (3, 3), padding='same', name='conv1a')(x)
x = BatchNormalization(name='conv1a_bn')(x)
x = Activation('relu', name='conv1a_act')(x)
x = Conv2D(64, (3, 3), padding='same', name='conv1b')(x)
x = BatchNormalization(name='conv1b_bn')(x)
x = Activation('relu', name='conv1b_act')(x)
x = MaxPooling2D((2, 2), name='conv1_pool')(x)
x = Conv2D(128, (3, 3), padding='same', name='conv2a')(x)
x = BatchNormalization(name='conv2a_bn')(x)
x = Activation('relu', name='conv2a_act')(x)
x = Conv2D(128, (3, 3), padding='same', name='conv2b')(x)
x = BatchNormalization(name='conv2b_bn')(x)
x = Activation('relu', name='conv2b_act')(x)
x = MaxPooling2D((2, 2), name='conv2_pool')(x)
x = Conv2D(256, (3, 3), padding='same', name='conv3a')(x)
x = BatchNormalization(name='conv3a_bn')(x)
x = Activation('relu', name='conv3a_act')(x)
x = Conv2D(256, (3, 3), padding='same', name='conv3b')(x)
x = BatchNormalization(name='conv3b_bn')(x)
x = Activation('relu', name='conv3b_act')(x)
x = MaxPooling2D((2, 2), name='conv3_pool')(x)
x = Conv2D(512, (3, 3), padding='same', name='conv4a')(x)
x = BatchNormalization(name='conv4a_bn')(x)
x = Activation('relu', name='conv4a_act')(x)
x = Conv2D(512, (3, 3), padding='same', name='conv4b')(x)
x = BatchNormalization(name='conv4b_bn')(x)
x = Activation('relu', name='conv4b_act')(x)
x = GlobalMaxPooling2D(name='avg_pool')(x)
x = Dense(1024, name='dense0', activation='elu')(x)
x = Dropout(0.4)(x)
x = Dense(1024, name='dense1', activation='elu')(x)
x = Dropout(0.4)(x)
x = Dense(np.prod(out_size), activation='elu', name='skeleton')(x)
x = Reshape((out_size))(x)
model = Model(img_input, x)
optimizer = Adam(lr=1e-3, decay=0.05)
model.compile(loss='mean_absolute_error',
optimizer=optimizer,
metrics=['mean_absolute_error', 'mean_squared_error', 'mean_absolute_percentage_error'])
model_name = 'simple_v2'
return model, model_name
def extract_folds_features(argv):
# Parse command line to read with which network should the script extract image information
parser = argparse.ArgumentParser(
description='Receives the desired network model for extraction')
parser.add_argument(
'--network',
'-n',
default="vgg16",
help=
'The desired network. We only support vgg16, inceptionV3 and resnet50')
parser.add_argument(
'--pooling',
'-p',
default="GAP",
help=
'Used pooling layer. We support: Global Max Poolling [GAP], Global Average Pooling[GMP], None [None]'
)
parser.add_argument('--fold',
'-f',
default="all",
help='Select which fold to extract features')
args = parser.parse_args()
# Check if network is suported
# We now only have vgg16
if not ((args.network == "vgg16") or (args.network == "inceptionV3") or
(args.network == "resnet50")):
print_error(
"We currently only suport the followuing models: vgg16, inceptionv3 and resnet50"
)
exit(1)
if not ((args.pooling == "GAP") or (args.pooling == "GMP") or
(args.pooling == "None")):
print_error("We currently only suport GAP, GMP and None layers")
exit(1)
# ------------------ Create extracting model
inputs = Input(CONST_VEC_DATASET_OUTPUT_IMAGE_SHAPE)
# ------------------ Convolution ------------------
if args.network == "vgg16":
convolutional_layer = VGG16(
weights='imagenet',
include_top=False,
input_shape=CONST_VEC_DATASET_OUTPUT_IMAGE_SHAPE)
for layer in convolutional_layer.layers[:]:
layer.trainable = False #Freezes all layers in the vgg16
output_shape = CONST_VEC_NETWORK_VGG16_OUTPUTSHAPE
if args.network == "inceptionV3":
convolutional_layer = InceptionV3(
weights='imagenet',
include_top=False,
input_shape=CONST_VEC_DATASET_OUTPUT_IMAGE_SHAPE)
for layer in convolutional_layer.layers[:]:
layer.trainable = False #Freezes all layers in the vgg16
output_shape = CONST_VEC_NETWORK_INCEPTIONV3_OUTPUTSHAPE
if args.network == "resnet50":
convolutional_layer = ResNet50(
weights='imagenet',
include_top=False,
input_shape=CONST_VEC_DATASET_OUTPUT_IMAGE_SHAPE)
for layer in convolutional_layer.layers[:]:
layer.trainable = False #Freezes all layers in the vgg16
output_shape = CONST_VEC_NETWORK_RESNET50_OUTPUTSHAPE
convolution_output = convolutional_layer(inputs)
# ------------------ Pooling ------------------
if args.pooling == "None":
outputs = convolution_output
elif args.pooling == "GAP":
outputs = GlobalAveragePooling2D(data_format=None)(convolution_output)
elif args.pooling == "GMP":
outputs = GlobalMaxPooling2D(data_format=None)(convolution_output)
model = Model(inputs=inputs, outputs=outputs)
print_info("Extracting network summary:")
model.summary()
# ------------------ Extracting process
telegramSendMessage("Starting extracting process with network " +
args.network + " and pooling layer " + args.pooling)
print_info("Starting extracting process with network " + args.network +
" and pooling layer " + args.pooling)
# Create folder to host all extracted models
if (args.network == "vgg16"):
extraction_datapath = os.path.join(CONST_STR_DATASET_BASE_PATH,
CONST_STR_DATASET_DATAPATH,
CONST_STR_DATASET_FOLDS_DATAPATH,
CONST_STR_DATASET_FEATURES_VGG16)
if (args.network == "inceptionV3"):
extraction_datapath = os.path.join(
CONST_STR_DATASET_BASE_PATH, CONST_STR_DATASET_DATAPATH,
CONST_STR_DATASET_FOLDS_DATAPATH,
CONST_STR_DATASET_FEATURES_INCEPTIONV3)
if (args.network == "resnet50"):
extraction_datapath = os.path.join(
CONST_STR_DATASET_BASE_PATH, CONST_STR_DATASET_DATAPATH,
CONST_STR_DATASET_FOLDS_DATAPATH,
CONST_STR_DATASET_FEATURES_RESNET50)
try:
if not os.path.exists(extraction_datapath):
os.makedirs(extraction_datapath)
except OSError:
print_error("Could not make directory for features extraction")
telegramSendMessage("Error: Creating directory")
exit(1)
if args.fold == "all":
for fold in folds:
# Load fold dataset
print_info("Loading dataset from " + fold["name"])
telegramSendMessage("Loading dataset from " + fold["name"])
try:
input_train_data = np.load(
os.path.join(
CONST_STR_DATASET_BASE_PATH,
CONST_STR_DATASET_DATAPATH,
CONST_STR_DATASET_FOLDS_DATAPATH,
"input_training_data_" + fold['name'] + ".npy"))
input_test_data = np.load(
os.path.join(CONST_STR_DATASET_BASE_PATH,
CONST_STR_DATASET_DATAPATH,
CONST_STR_DATASET_FOLDS_DATAPATH,
"input_testing_data_" + fold['name'] +
".npy"))
except:
print_error("Could not find dataset. Did you build it?")
telegramSendMessage(
"Could not find dataset. Did you build it?")
exit(1)
# Extracting features of fold
print_info("Extracting training features")
telegramSendMessage("Extracting training features")
train_features = []
number_of_images = input_train_data.shape[0]
for index in range(number_of_images):
# Expand dimention of image from (224,224,3) to (1,224,224,3)
image = np.expand_dims(input_train_data[index], 0)
# Pass image throught the model
image_feature = model.predict(image)
# Append to the train_features array
train_features.append(image_feature)
#Transform array into ndarray
train_features = np.array(train_features).astype("float32")
train_features = np.reshape(train_features,
(number_of_images, output_shape[2]))
#Save the extracted features
print_info("Saving training features")
telegramSendMessage("Saving training features")
np.save(
os.path.join(
extraction_datapath, "input_training_data_" +
args.pooling + "_" + fold["name"]), train_features)
### Repeat to test dataset
print_info("Extracting testing features")
telegramSendMessage("Extracting testing features")
test_features = []
number_of_images = input_test_data.shape[0]
for index in range(number_of_images):
image = np.expand_dims(input_test_data[index], 0)
image_feature = model.predict(image)
test_features.append(image_feature)
test_features = np.array(test_features).astype("float32")
test_features = np.reshape(test_features,
(number_of_images, output_shape[2]))
#Save the extracted features
print_info("Saving testing features")
telegramSendMessage("Saving testing features")
np.save(
os.path.join(
extraction_datapath,
"input_testing_data_" + args.pooling + "_" + fold["name"]),
test_features)
# Forcefully delete input datas from memory
del input_train_data
del input_test_data
del train_features
del test_features
gc.collect()
else:
fold_name = args.fold
# Load fold dataset
print_info("Loading dataset from " + fold_name)
telegramSendMessage("Loading dataset from " + fold_name)
try:
input_train_data = np.load(
os.path.join(CONST_STR_DATASET_BASE_PATH,
CONST_STR_DATASET_DATAPATH,
CONST_STR_DATASET_FOLDS_DATAPATH,
"input_training_data_" + fold_name + ".npy"))
input_test_data = np.load(
os.path.join(CONST_STR_DATASET_BASE_PATH,
CONST_STR_DATASET_DATAPATH,
CONST_STR_DATASET_FOLDS_DATAPATH,
"input_testing_data_" + fold_name + ".npy"))
except:
print_error("Could not find dataset. Did you build it?")
telegramSendMessage("Could not find dataset. Did you build it?")
exit(1)
# Extracting features of fold
print_info("Extracting training features")
telegramSendMessage("Extracting training features")
train_features = []
number_of_images = input_train_data.shape[0]
for index in range(number_of_images):
# Expand dimention of image from (224,224,3) to (1,224,224,3)
image = np.expand_dims(input_train_data[index], 0)
# Pass image throught the model
image_feature = model.predict(image)
# Append to the train_features array
train_features.append(image_feature)
#Transform array into ndarray
train_features = np.array(train_features).astype("float32")
train_features = np.reshape(train_features,
(number_of_images, output_shape[2]))
#Save the extracted features
print_info("Saving training features")
telegramSendMessage("Saving training features")
np.save(
os.path.join(
extraction_datapath,
"input_training_data_" + args.pooling + "_" + fold_name),
train_features)
### Repeat to test dataset
print_info("Extracting testing features")
telegramSendMessage("Extracting testing features")
test_features = []
number_of_images = input_test_data.shape[0]
for index in range(number_of_images):
image = np.expand_dims(input_test_data[index], 0)
image_feature = model.predict(image)
test_features.append(image_feature)
test_features = np.array(test_features).astype("float32")
test_features = np.reshape(test_features,
(number_of_images, output_shape[2]))
#Save the extracted features
print_info("Saving testing features")
telegramSendMessage("Saving testing features")
np.save(
os.path.join(
extraction_datapath,
"input_testing_data_" + args.pooling + "_" + fold_name),
test_features)
# Forcefully delete input datas from memory
del input_train_data
del input_test_data
del train_features
del test_features
gc.collect()
print_info("Extraction script end")
telegramSendMessage("Extraction script end")
def get_test_model_full():
"""Returns a maximally complex test model,
using all supported layer types with different parameter combination.
"""
input_shapes = [
(26, 28, 3),
(4, 4, 3),
(4, 4, 3),
(4,),
(2, 3),
(27, 29, 1),
(17, 1),
(17, 4),
(2, 3),
]
inputs = [Input(shape=s) for s in input_shapes]
outputs = []
for inp in inputs[6:8]:
for padding in ['valid', 'same']:
for s in range(1, 6):
for out_channels in [1, 2]:
for d in range(1, 4):
outputs.append(
Conv1D(out_channels, s, padding=padding,
dilation_rate=d)(inp))
for padding_size in range(0, 5):
outputs.append(ZeroPadding1D(padding_size)(inp))
for crop_left in range(0, 2):
for crop_right in range(0, 2):
outputs.append(Cropping1D((crop_left, crop_right))(inp))
for upsampling_factor in range(1, 5):
outputs.append(UpSampling1D(upsampling_factor)(inp))
for padding in ['valid', 'same']:
for pool_factor in range(1, 6):
for s in range(1, 4):
outputs.append(
MaxPooling1D(pool_factor, strides=s,
padding=padding)(inp))
outputs.append(
AveragePooling1D(pool_factor, strides=s,
padding=padding)(inp))
outputs.append(GlobalMaxPooling1D()(inp))
outputs.append(GlobalAveragePooling1D()(inp))
for inp in [inputs[0], inputs[5]]:
for padding in ['valid', 'same']:
for h in range(1, 6):
for out_channels in [1, 2]:
for d in range(1, 4):
outputs.append(
Conv2D(out_channels, (h, 1), padding=padding,
dilation_rate=(d, 1))(inp))
outputs.append(
SeparableConv2D(out_channels, (h, 1), padding=padding,
dilation_rate=(d, 1))(inp))
for sy in range(1, 4):
outputs.append(
Conv2D(out_channels, (h, 1), strides=(1, sy),
padding=padding)(inp))
outputs.append(
SeparableConv2D(out_channels, (h, 1),
strides=(sy, sy),
padding=padding)(inp))
for sy in range(1, 4):
outputs.append(
DepthwiseConv2D((h, 1),
strides=(sy, sy),
padding=padding)(inp))
outputs.append(
MaxPooling2D((h, 1), strides=(1, sy),
padding=padding)(inp))
for w in range(1, 6):
for out_channels in [1, 2]:
for d in range(1, 4) if sy == 1 else [1]:
outputs.append(
Conv2D(out_channels, (1, w), padding=padding,
dilation_rate=(1, d))(inp))
outputs.append(
SeparableConv2D(out_channels, (1, w), padding=padding,
dilation_rate=(1, d))(inp))
for sx in range(1, 4):
outputs.append(
Conv2D(out_channels, (1, w), strides=(sx, 1),
padding=padding)(inp))
outputs.append(
SeparableConv2D(out_channels, (1, w),
strides=(sx, sx),
padding=padding)(inp))
for sx in range(1, 4):
outputs.append(
DepthwiseConv2D((1, w),
strides=(sy, sy),
padding=padding)(inp))
outputs.append(
MaxPooling2D((1, w), strides=(1, sx),
padding=padding)(inp))
outputs.append(ZeroPadding2D(2)(inputs[0]))
outputs.append(ZeroPadding2D((2, 3))(inputs[0]))
outputs.append(ZeroPadding2D(((1, 2), (3, 4)))(inputs[0]))
outputs.append(Cropping2D(2)(inputs[0]))
outputs.append(Cropping2D((2, 3))(inputs[0]))
outputs.append(Cropping2D(((1, 2), (3, 4)))(inputs[0]))
for y in range(1, 3):
for x in range(1, 3):
outputs.append(UpSampling2D(size=(y, x))(inputs[0]))
outputs.append(GlobalAveragePooling2D()(inputs[0]))
outputs.append(GlobalMaxPooling2D()(inputs[0]))
outputs.append(AveragePooling2D((2, 2))(inputs[0]))
outputs.append(MaxPooling2D((2, 2))(inputs[0]))
outputs.append(UpSampling2D((2, 2))(inputs[0]))
outputs.append(keras.layers.concatenate([inputs[0], inputs[0]]))
outputs.append(Dropout(0.5)(inputs[0]))
outputs.append(BatchNormalization()(inputs[0]))
outputs.append(BatchNormalization(center=False)(inputs[0]))
outputs.append(BatchNormalization(scale=False)(inputs[0]))
outputs.append(Conv2D(2, (3, 3), use_bias=True)(inputs[0]))
outputs.append(Conv2D(2, (3, 3), use_bias=False)(inputs[0]))
outputs.append(SeparableConv2D(2, (3, 3), use_bias=True)(inputs[0]))
outputs.append(SeparableConv2D(2, (3, 3), use_bias=False)(inputs[0]))
outputs.append(DepthwiseConv2D(2, (3, 3), use_bias=True)(inputs[0]))
outputs.append(DepthwiseConv2D(2, (3, 3), use_bias=False)(inputs[0]))
outputs.append(Dense(2, use_bias=True)(inputs[3]))
outputs.append(Dense(2, use_bias=False)(inputs[3]))
shared_conv = Conv2D(1, (1, 1),
padding='valid', name='shared_conv', activation='relu')
up_scale_2 = UpSampling2D((2, 2))
x1 = shared_conv(up_scale_2(inputs[1])) # (1, 8, 8)
x2 = shared_conv(up_scale_2(inputs[2])) # (1, 8, 8)
x3 = Conv2D(1, (1, 1), padding='valid')(up_scale_2(inputs[2])) # (1, 8, 8)
x = keras.layers.concatenate([x1, x2, x3]) # (3, 8, 8)
outputs.append(x)
x = Conv2D(3, (1, 1), padding='same', use_bias=False)(x) # (3, 8, 8)
outputs.append(x)
x = Dropout(0.5)(x)
outputs.append(x)
x = keras.layers.concatenate([
MaxPooling2D((2, 2))(x),
AveragePooling2D((2, 2))(x)]) # (6, 4, 4)
outputs.append(x)
x = Flatten()(x) # (1, 1, 96)
x = Dense(4, use_bias=False)(x)
outputs.append(x)
x = Dense(3)(x) # (1, 1, 3)
outputs.append(x)
outputs.append(Activation(relu6)(inputs[7]))
outputs.append(keras.layers.Add()([inputs[4], inputs[8], inputs[8]]))
outputs.append(keras.layers.Subtract()([inputs[4], inputs[8]]))
outputs.append(keras.layers.Multiply()([inputs[4], inputs[8], inputs[8]]))
outputs.append(keras.layers.Average()([inputs[4], inputs[8], inputs[8]]))
outputs.append(keras.layers.Maximum()([inputs[4], inputs[8], inputs[8]]))
outputs.append(keras.layers.Concatenate()([inputs[4], inputs[8], inputs[8]]))
intermediate_input_shape = (3,)
intermediate_in = Input(intermediate_input_shape)
intermediate_x = intermediate_in
intermediate_x = Dense(8)(intermediate_x)
intermediate_x = Dense(5)(intermediate_x)
intermediate_model = Model(
inputs=[intermediate_in], outputs=[intermediate_x],
name='intermediate_model')
intermediate_model.compile(loss='mse', optimizer='nadam')
x = intermediate_model(x) # (1, 1, 5)
intermediate_model_2 = Sequential()
intermediate_model_2.add(Dense(7, input_shape=(5,)))
intermediate_model_2.add(Dense(5))
intermediate_model_2.compile(optimizer='rmsprop',
loss='categorical_crossentropy')
x = intermediate_model_2(x) # (1, 1, 5)
x = Dense(3)(x) # (1, 1, 3)
shared_activation = Activation('tanh')
outputs = outputs + [
Activation('tanh')(inputs[3]),
Activation('hard_sigmoid')(inputs[3]),
Activation('selu')(inputs[3]),
Activation('sigmoid')(inputs[3]),
Activation('softplus')(inputs[3]),
Activation('softmax')(inputs[3]),
Activation('relu')(inputs[3]),
LeakyReLU()(inputs[3]),
ELU()(inputs[3]),
PReLU()(inputs[2]),
PReLU()(inputs[3]),
PReLU()(inputs[4]),
shared_activation(inputs[3]),
inputs[4],
inputs[1],
x,
shared_activation(x),
]
print('Model has {} outputs.'.format(len(outputs)))
model = Model(inputs=inputs, outputs=outputs, name='test_model_full')
model.compile(loss='mse', optimizer='nadam')
# fit to dummy data
training_data_size = 1
batch_size = 1
epochs = 10
data_in = generate_input_data(training_data_size, input_shapes)
data_out = generate_output_data(training_data_size, outputs)
model.fit(data_in, data_out, epochs=epochs, batch_size=batch_size)
return model
def ResNet152(include_top=True,
weights=None,
input_tensor=None,
input_shape=None,
large_input=False,
pooling=None,
classes=1000):
"""Instantiate the ResNet152 architecture.
Keyword arguments:
include_top -- whether to include the fully-connected layer at the
top of the network. (default True)
weights -- one of `None` (random initialization) or "imagenet"
(pre-training on ImageNet). (default None)
input_tensor -- optional Keras tensor (i.e. output of `layers.Input()`)
to use as image input for the model.(default None)
input_shape -- optional shape tuple, only to be specified if
`include_top` is False (otherwise the input shape has to be
`(224, 224, 3)` (with `channels_last` data format) or
`(3, 224, 224)` (with `channels_first` data format). It should
have exactly 3 inputs channels, and width and height should be
no smaller than 197. E.g. `(200, 200, 3)` would be one valid value.
(default None)
large_input -- if True, then the input shape expected will be
`(448, 448, 3)` (with `channels_last` data format) or
`(3, 448, 448)` (with `channels_first` data format). (default False)
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.
(default None)
classes -- optional number of classes to classify image into, only
to be specified if `include_top` is True, and if no `weights`
argument is specified. (default 1000)
Returns:
A Keras model instance.
Raises:
ValueError: in case of invalid argument for `weights`,
or invalid input shape.
"""
if weights not in {'imagenet', None}:
raise ValueError('The `weights` argument should be either '
'`None` (random initialization) or `imagenet` '
'(pre-training on ImageNet).')
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')
eps = 1.1e-5
if large_input:
img_size = 448
else:
img_size = 224
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=img_size,
min_size=197,
data_format=K.image_data_format(),
require_flatten=include_top)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
# handle dimension ordering for different backends
if K.image_dim_ordering() == 'tf':
bn_axis = 3
else:
bn_axis = 1
x = ZeroPadding2D((3, 3), name='conv1_zeropadding')(img_input)
x = Conv2D(64, (7, 7), strides=(2, 2), name='conv1', use_bias=False)(x)
x = BatchNormalization(epsilon=eps, axis=bn_axis, name='bn_conv1')(x)
x = Scale(axis=bn_axis, name='scale_conv1')(x)
x = Activation('relu', name='conv1_relu')(x)
x = MaxPooling2D((3, 3), strides=(2, 2), name='pool1')(x)
x = conv_block(x, 3, [64, 64, 256], stage=2, block='a', strides=(1, 1))
x = identity_block(x, 3, [64, 64, 256], stage=2, block='b')
x = identity_block(x, 3, [64, 64, 256], stage=2, block='c')
x = conv_block(x, 3, [128, 128, 512], stage=3, block='a')
for i in range(1, 8):
x = identity_block(x, 3, [128, 128, 512], stage=3, block='b' + str(i))
x = conv_block(x, 3, [256, 256, 1024], stage=4, block='a')
for i in range(1, 36):
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='b' + str(i))
x = conv_block(x, 3, [512, 512, 2048], stage=5, block='a')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='b')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='c')
if large_input:
x = AveragePooling2D((14, 14), name='avg_pool')(x)
else:
x = AveragePooling2D((7, 7), name='avg_pool')(x)
# include classification layer by default, not included for feature extraction
if include_top:
x = Flatten()(x)
x = Dense(classes, activation='softmax', name='fc1000')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model.
model = Model(inputs, x, name='resnet152')
return model
def share_stream(x_shape, outputNum=256):
input = Input(x_shape)
x_a = input
conv1 = Conv2D(filters=32,
kernel_size=(5, 5),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(x_a)
conv1 = Activation('relu')(conv1)
conv1 = Conv2D(filters=32,
kernel_size=(1, 1),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(conv1)
x_a = Conv2D(filters=32,
kernel_size=(1, 1),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(x_a)
x_a = add([x_a, conv1])
attention = Conv2D(filters=32,
kernel_size=(5, 5),
activation='sigmoid',
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(x_a)
x_a = multiply([x_a, attention])
x_a = MaxPooling2D(pool_size=(2, 2), strides=(2, 2), padding='valid')(x_a)
conv1 = Conv2D(filters=64,
kernel_size=(3, 3),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(x_a)
conv1 = Activation('relu')(conv1)
conv1 = Conv2D(filters=64,
kernel_size=(1, 1),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(conv1)
x_a = Conv2D(filters=64,
kernel_size=(1, 1),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(x_a)
x_a = add([x_a, conv1])
x_a = MaxPooling2D(pool_size=(2, 2), strides=(2, 2), padding='valid')(x_a)
x = Conv2D(filters=128,
kernel_size=(3, 3),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(x_a)
temp = Multiply()([
Conv2D(filters=128,
kernel_size=(3, 3),
padding='same',
trainable=Trainable)(Lambda(lambda x: x * 10000000.)(
Activation("tanh")(Lambda(lambda x: x / 10000000.)(x)))), x
])
for jj in range(2, taylor):
temp = Add()([
Multiply()([
Conv2D(filters=128,
kernel_size=(3, 3),
padding='same',
trainable=Trainable)(Lambda(lambda x: x * 10000000.)(
Activation("tanh")(
Lambda(lambda x: x / 10000000.)(x)))),
Lambda(lambda x: (x**jj) / math.factorial(jj))(x)
]), temp
])
temp = Add()([
Conv2D(filters=128,
kernel_size=(3, 3),
padding='same',
trainable=Trainable)(Lambda(lambda x: x * 10000000.)(
Activation("tanh")(Lambda(lambda x: x / 10000000.)(x)))),
temp
])
conv2 = Lambda(lambda x: x * 10000000.)(Activation("tanh")(
Lambda(lambda x: x / 10000000.)(temp)))
conv2 = Conv2D(filters=128,
kernel_size=(1, 1),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(conv2)
x_a = Conv2D(filters=128,
kernel_size=(1, 1),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(x_a)
x_a = add([x_a, conv2])
x_a = MaxPooling2D(pool_size=(2, 2), strides=(2, 2), padding='valid')(x_a)
conv3 = Conv2D(filters=outputNum,
kernel_size=(3, 3),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(x_a)
conv3 = Activation('relu')(conv3)
conv3 = Conv2D(filters=outputNum,
kernel_size=(1, 1),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(conv3)
x_a = Conv2D(filters=outputNum,
kernel_size=(1, 1),
strides=(1, 1),
padding='same',
use_bias=use_bias,
trainable=Trainable)(x_a)
x_a = add([x_a, conv3])
x_a = GlobalMaxPooling2D()(x_a)
shared_layer = Model(input, x_a)
return shared_layer
def SEInceptionV3(include_top=True,
weights=None,
input_tensor=None,
input_shape=None,
pooling=None,
classes=1000):
"""Instantiates the Squeeze and Excite Inception v3 architecture.
# Arguments
include_top: whether to include the fully-connected
layer at the top of the network.
weights: one of `None` (random initialization)
or "imagenet" (pre-training on ImageNet).
input_tensor: optional Keras tensor (i.e. output of `layers.Input()`)
to use as image input for the model.
input_shape: optional shape tuple, only to be specified
if `include_top` is False (otherwise the input shape
has to be `(299, 299, 3)` (with `channels_last` data format)
or `(3, 299, 299)` (with `channels_first` data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 139.
E.g. `(150, 150, 3)` would be one valid value.
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.
"""
if weights not in {'imagenet', None}:
raise ValueError('The `weights` argument should be either '
'`None` (random initialization) or `imagenet` '
'(pre-training on ImageNet).')
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')
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=299,
min_size=139,
data_format=K.image_data_format(),
require_flatten=include_top)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
if K.image_data_format() == 'channels_first':
channel_axis = 1
else:
channel_axis = 3
x = _conv2d_bn(img_input, 32, 3, 3, strides=(2, 2), padding='valid')
x = _conv2d_bn(x, 32, 3, 3, padding='valid')
x = _conv2d_bn(x, 64, 3, 3)
x = MaxPooling2D((3, 3), strides=(2, 2))(x)
x = _conv2d_bn(x, 80, 1, 1, padding='valid')
x = _conv2d_bn(x, 192, 3, 3, padding='valid')
x = MaxPooling2D((3, 3), strides=(2, 2))(x)
# mixed 0, 1, 2: 35 x 35 x 256
branch1x1 = _conv2d_bn(x, 64, 1, 1)
branch5x5 = _conv2d_bn(x, 48, 1, 1)
branch5x5 = _conv2d_bn(branch5x5, 64, 5, 5)
branch3x3dbl = _conv2d_bn(x, 64, 1, 1)
branch3x3dbl = _conv2d_bn(branch3x3dbl, 96, 3, 3)
branch3x3dbl = _conv2d_bn(branch3x3dbl, 96, 3, 3)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1), padding='same')(x)
branch_pool = _conv2d_bn(branch_pool, 32, 1, 1)
x = layers.concatenate([branch1x1, branch5x5, branch3x3dbl, branch_pool],
axis=channel_axis,
name='mixed0')
# squeeze and excite block
x = squeeze_excite_block(x)
# mixed 1: 35 x 35 x 256
branch1x1 = _conv2d_bn(x, 64, 1, 1)
branch5x5 = _conv2d_bn(x, 48, 1, 1)
branch5x5 = _conv2d_bn(branch5x5, 64, 5, 5)
branch3x3dbl = _conv2d_bn(x, 64, 1, 1)
branch3x3dbl = _conv2d_bn(branch3x3dbl, 96, 3, 3)
branch3x3dbl = _conv2d_bn(branch3x3dbl, 96, 3, 3)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1), padding='same')(x)
branch_pool = _conv2d_bn(branch_pool, 64, 1, 1)
x = layers.concatenate([branch1x1, branch5x5, branch3x3dbl, branch_pool],
axis=channel_axis,
name='mixed1')
# squeeze and excite block
x = squeeze_excite_block(x)
# mixed 2: 35 x 35 x 256
branch1x1 = _conv2d_bn(x, 64, 1, 1)
branch5x5 = _conv2d_bn(x, 48, 1, 1)
branch5x5 = _conv2d_bn(branch5x5, 64, 5, 5)
branch3x3dbl = _conv2d_bn(x, 64, 1, 1)
branch3x3dbl = _conv2d_bn(branch3x3dbl, 96, 3, 3)
branch3x3dbl = _conv2d_bn(branch3x3dbl, 96, 3, 3)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1), padding='same')(x)
branch_pool = _conv2d_bn(branch_pool, 64, 1, 1)
x = layers.concatenate([branch1x1, branch5x5, branch3x3dbl, branch_pool],
axis=channel_axis,
name='mixed2')
# squeeze and excite block
x = squeeze_excite_block(x)
# mixed 3: 17 x 17 x 768
branch3x3 = _conv2d_bn(x, 384, 3, 3, strides=(2, 2), padding='valid')
branch3x3dbl = _conv2d_bn(x, 64, 1, 1)
branch3x3dbl = _conv2d_bn(branch3x3dbl, 96, 3, 3)
branch3x3dbl = _conv2d_bn(branch3x3dbl,
96,
3,
3,
strides=(2, 2),
padding='valid')
branch_pool = MaxPooling2D((3, 3), strides=(2, 2))(x)
x = layers.concatenate([branch3x3, branch3x3dbl, branch_pool],
axis=channel_axis,
name='mixed3')
# squeeze and excite block
x = squeeze_excite_block(x)
# mixed 4: 17 x 17 x 768
branch1x1 = _conv2d_bn(x, 192, 1, 1)
branch7x7 = _conv2d_bn(x, 128, 1, 1)
branch7x7 = _conv2d_bn(branch7x7, 128, 1, 7)
branch7x7 = _conv2d_bn(branch7x7, 192, 7, 1)
branch7x7dbl = _conv2d_bn(x, 128, 1, 1)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 128, 7, 1)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 128, 1, 7)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 128, 7, 1)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 192, 1, 7)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1), padding='same')(x)
branch_pool = _conv2d_bn(branch_pool, 192, 1, 1)
x = layers.concatenate([branch1x1, branch7x7, branch7x7dbl, branch_pool],
axis=channel_axis,
name='mixed4')
# squeeze and excite block
x = squeeze_excite_block(x)
# mixed 5, 6: 17 x 17 x 768
for i in range(2):
branch1x1 = _conv2d_bn(x, 192, 1, 1)
branch7x7 = _conv2d_bn(x, 160, 1, 1)
branch7x7 = _conv2d_bn(branch7x7, 160, 1, 7)
branch7x7 = _conv2d_bn(branch7x7, 192, 7, 1)
branch7x7dbl = _conv2d_bn(x, 160, 1, 1)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 160, 7, 1)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 160, 1, 7)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 160, 7, 1)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 192, 1, 7)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1),
padding='same')(x)
branch_pool = _conv2d_bn(branch_pool, 192, 1, 1)
x = layers.concatenate(
[branch1x1, branch7x7, branch7x7dbl, branch_pool],
axis=channel_axis,
name='mixed' + str(5 + i))
# squeeze and excite block
x = squeeze_excite_block(x)
# mixed 7: 17 x 17 x 768
branch1x1 = _conv2d_bn(x, 192, 1, 1)
branch7x7 = _conv2d_bn(x, 192, 1, 1)
branch7x7 = _conv2d_bn(branch7x7, 192, 1, 7)
branch7x7 = _conv2d_bn(branch7x7, 192, 7, 1)
branch7x7dbl = _conv2d_bn(x, 192, 1, 1)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 192, 7, 1)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 192, 1, 7)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 192, 7, 1)
branch7x7dbl = _conv2d_bn(branch7x7dbl, 192, 1, 7)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1), padding='same')(x)
branch_pool = _conv2d_bn(branch_pool, 192, 1, 1)
x = layers.concatenate([branch1x1, branch7x7, branch7x7dbl, branch_pool],
axis=channel_axis,
name='mixed7')
# squeeze and excite block
x = squeeze_excite_block(x)
# mixed 8: 8 x 8 x 1280
branch3x3 = _conv2d_bn(x, 192, 1, 1)
branch3x3 = _conv2d_bn(branch3x3,
320,
3,
3,
strides=(2, 2),
padding='valid')
branch7x7x3 = _conv2d_bn(x, 192, 1, 1)
branch7x7x3 = _conv2d_bn(branch7x7x3, 192, 1, 7)
branch7x7x3 = _conv2d_bn(branch7x7x3, 192, 7, 1)
branch7x7x3 = _conv2d_bn(branch7x7x3,
192,
3,
3,
strides=(2, 2),
padding='valid')
branch_pool = MaxPooling2D((3, 3), strides=(2, 2))(x)
x = layers.concatenate([branch3x3, branch7x7x3, branch_pool],
axis=channel_axis,
name='mixed8')
# squeeze and excite block
x = squeeze_excite_block(x)
# mixed 9: 8 x 8 x 2048
for i in range(2):
branch1x1 = _conv2d_bn(x, 320, 1, 1)
branch3x3 = _conv2d_bn(x, 384, 1, 1)
branch3x3_1 = _conv2d_bn(branch3x3, 384, 1, 3)
branch3x3_2 = _conv2d_bn(branch3x3, 384, 3, 1)
branch3x3 = layers.concatenate([branch3x3_1, branch3x3_2],
axis=channel_axis,
name='mixed9_' + str(i))
branch3x3dbl = _conv2d_bn(x, 448, 1, 1)
branch3x3dbl = _conv2d_bn(branch3x3dbl, 384, 3, 3)
branch3x3dbl_1 = _conv2d_bn(branch3x3dbl, 384, 1, 3)
branch3x3dbl_2 = _conv2d_bn(branch3x3dbl, 384, 3, 1)
branch3x3dbl = layers.concatenate([branch3x3dbl_1, branch3x3dbl_2],
axis=channel_axis)
branch_pool = AveragePooling2D((3, 3), strides=(1, 1),
padding='same')(x)
branch_pool = _conv2d_bn(branch_pool, 192, 1, 1)
x = layers.concatenate(
[branch1x1, branch3x3, branch3x3dbl, branch_pool],
axis=channel_axis,
name='mixed' + str(9 + i))
# squeeze and excite block
x = squeeze_excite_block(x)
if include_top:
# Classification block
x = GlobalAveragePooling2D(name='avg_pool')(x)
x = Dense(classes, activation='softmax', name='predictions')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model.
model = Model(inputs, x, name='inception_v3')
return model
def _create_se_resnet(classes, img_input, include_top, initial_conv_filters,
filters, depth, width, bottleneck, weight_decay,
pooling):
'''Creates a SE ResNet model with specified parameters
Args:
initial_conv_filters: number of features for the initial convolution
include_top: Flag to include the last dense layer
filters: number of filters per block, defined as a list.
filters = [64, 128, 256, 512
depth: number or layers in the each block, defined as a list.
ResNet-50 = [3, 4, 6, 3]
ResNet-101 = [3, 6, 23, 3]
ResNet-152 = [3, 8, 36, 3]
width: width multiplier for network (for Wide ResNet)
bottleneck: adds a bottleneck conv to reduce computation
weight_decay: weight_decay (l2 norm)
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.
Returns: a Keras Model
'''
channel_axis = 1 if K.image_data_format() == 'channels_first' else -1
N = list(depth)
# branches
branches = []
for i in range(len(img_input)):
# block 1 (initial conv block)
branch = Conv2D(initial_conv_filters, (7, 7),
padding='same',
use_bias=False,
strides=(2, 2),
kernel_initializer='he_normal',
kernel_regularizer=l2(weight_decay))(img_input[i])
branch = MaxPooling2D((3, 3), strides=(2, 2), padding='same')(branch)
# block 2 (projection block)
for i in range(N[0]):
if bottleneck:
branch = _resnet_bottleneck_block(branch, filters[0], width)
else:
branch = _resnet_block(branch, filters[0], width)
branches.append(branch)
x = concatenate(branches)
'''
# block 1 (initial conv block)
x = Conv2D(initial_conv_filters, (7, 7), padding='same', use_bias=False, strides=(2, 2),
kernel_initializer='he_normal', kernel_regularizer=l2(weight_decay))(img_input)
x = MaxPooling2D((3, 3), strides=(2, 2), padding='same')(x)
# block 2 (projection block)
for i in range(N[0]):
if bottleneck:
x = _resnet_bottleneck_block(x, filters[0], width)
else:
x = _resnet_block(x, filters[0], width)
'''
# block 3 - N
for k in range(1, len(N)):
if bottleneck:
x = _resnet_bottleneck_block(x, filters[k], width, strides=(2, 2))
else:
x = _resnet_block(x, filters[k], width, strides=(2, 2))
for i in range(N[k] - 1):
if bottleneck:
x = _resnet_bottleneck_block(x, filters[k], width)
else:
x = _resnet_block(x, filters[k], width)
x = BatchNormalization(axis=channel_axis)(x)
x = Activation('relu')(x)
if include_top:
x = GlobalAveragePooling2D()(x)
x = Dense(classes,
use_bias=False,
kernel_regularizer=l2(weight_decay),
activation='sigmoid',
name='output')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
return x
def pyramid_feat2_large():
#DIDN'T CONVERNGED
out_size = (49, 2)
roi_size = 128
rand_seed = 1337
np.random.seed(rand_seed)
input_shape = (roi_size, roi_size, 1)
img_input = Input(shape=input_shape)
block1 = Conv2D(32, (3, 3), padding='same', name='conv0a')(img_input)
block1 = Activation('relu', name='conv0a_act')(block1)
block1 = Conv2D(32, (3, 3), padding='same', name='conv0b')(block1)
block1 = Activation('relu', name='conv0b_act')(block1)
block2 = MaxPooling2D((2, 2), name='conv0_pool')(block1)
block2 = Conv2D(64, (3, 3), padding='same', name='conv1a')(block2)
block2 = BatchNormalization(name='conv1a_bn')(block2)
block2 = Activation('relu', name='conv1a_act')(block2)
block2 = Conv2D(64, (3, 3), padding='same', name='conv1b')(block2)
block2 = BatchNormalization(name='conv1b_bn')(block2)
block2 = Activation('relu', name='conv1b_act')(block2)
block3 = MaxPooling2D((2, 2), name='conv1_pool')(block2)
block3 = Conv2D(128, (3, 3), padding='same', name='conv2a')(block3)
block3 = BatchNormalization(name='conv2a_bn')(block3)
block3 = Activation('relu', name='conv2a_act')(block3)
block3 = Conv2D(128, (3, 3), padding='same', name='conv2b')(block3)
block3 = BatchNormalization(name='conv2b_bn')(block3)
block3 = Activation('relu', name='conv2b_act')(block3)
block4 = MaxPooling2D((2, 2), name='conv2_pool')(block3)
block4 = Conv2D(256, (3, 3), padding='same', name='conv3a')(block4)
block4 = BatchNormalization(name='conv3a_bn')(block4)
block4 = Activation('relu', name='conv3a_act')(block4)
block4 = Conv2D(256, (3, 3), padding='same', name='conv3b')(block4)
block4 = BatchNormalization(name='conv3b_bn')(block4)
block4 = Activation('relu', name='conv3b_act')(block4)
block5 = MaxPooling2D((2, 2), name='conv3_pool')(block4)
block5 = Conv2D(512, (3, 3), padding='same', name='conv4a')(block5)
block5 = BatchNormalization(name='conv4a_bn')(block5)
block5 = Activation('relu', name='conv4a_act')(block5)
block5 = Conv2D(512, (3, 3), padding='same', name='conv4b')(block5)
block5 = BatchNormalization(name='conv4b_bn')(block5)
block5 = Activation('relu', name='conv4b_act')(block5)
feat_top = GlobalMaxPooling2D(name='avg_pool')(block5)
feat_top = Dense(1024, name='dense0', activation='elu')(feat_top)
feat_top = Dropout(0.4)(feat_top)
down_block5 = Conv2D(1, (1, 1), padding='same')(block5)
down_block5 = UpSampling2D((2,2))(down_block5)
lat_block4 = Conv2D(1, (1, 1), padding='same')(block4)
p_block4 = layers.add([lat_block4, down_block5])
down_block4 = UpSampling2D((2,2))(p_block4)
lat_block3 = Conv2D(1, (1, 1), padding='same')(block3)
p_block3 = layers.add([lat_block3, down_block4])
down_block3 = UpSampling2D((2,2))(p_block3)
lat_block2 = Conv2D(1, (1, 1), padding='same')(block2)
p_block2 = layers.add([lat_block2, down_block3])
feat_bot = Flatten()(p_block2)
feat_bot = Dense(1024, activation='elu')(feat_bot)
feat_bot = Dropout(0.4)(feat_bot)
all_feats = layers.add([feat_top, feat_bot])
all_feats = Dense(1024, activation='elu')(all_feats)
feat_bot = Dropout(0.4)(feat_bot)
skel_out = Dense(out_size[0]*out_size[1], activation='elu', name='skeleton')(all_feats)
skel_out = Reshape((out_size))(skel_out)
model = Model(img_input, skel_out)
#optimizer = Adam(lr=1e-2, decay=0.1)
optimizer = Adam(lr=1e-3, decay=0.1)
model.compile(loss='mean_absolute_error',
optimizer=optimizer,
metrics=['mean_absolute_error', 'mean_squared_error', 'mean_absolute_percentage_error'])
return model, 'pyramid_feat2_large'
def __create_dense_net(
nb_classes,
img_input,
include_top,
depth=40,
nb_dense_block=3,
growth_rate=12,
nb_filter=-1,
nb_layers_per_block=-1,
bottleneck=False,
reduction=0.0,
dropout_rate=None,
weight_decay=1e-4,
subsample_initial_block=False,
pooling=None,
activation="softmax",
transition_pooling="avg",
):
""" Build the DenseNet model
# Arguments
nb_classes: number of classes
img_input: tuple of shape (channels, rows, columns) or (rows, columns, channels)
include_top: flag to include the final Dense layer
depth: number or layers
nb_dense_block: number of dense blocks to add to end (generally = 3)
growth_rate: number of filters to add per dense block
nb_filter: initial number of filters. Default -1 indicates initial number
of filters is 2 * growth_rate
nb_layers_per_block: number of layers in each dense block.
Can be a -1, positive integer or a list.
If -1, calculates nb_layer_per_block from the depth of the network.
If positive integer, a set number of layers per dense block.
If list, nb_layer is used as provided. Note that list size must
be (nb_dense_block + 1)
bottleneck: add bottleneck blocks
reduction: reduction factor of transition blocks. Note : reduction value is
inverted to compute compression
dropout_rate: dropout rate
weight_decay: weight decay rate
subsample_initial_block: Changes model type to suit different datasets.
Should be set to True for ImageNet, and False for CIFAR datasets.
When set to True, the initial convolution will be strided and
adds a MaxPooling2D before the initial dense block.
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.
activation: Type of activation at the top layer. Can be one of 'softmax' or
'sigmoid'. Note that if sigmoid is used, classes must be 1.
transition_pooling: `avg` for avg pooling (default), `max` for max pooling,
None for no pooling during scale transition blocks. Please note that this
default differs from the DenseNetFCN paper in accordance with the DenseNet
paper.
# Returns
a keras tensor
# Raises
ValueError: in case of invalid argument for `reduction`
or `nb_dense_block`
"""
with K.name_scope("DenseNet"):
concat_axis = 1 if K.image_data_format() == "channels_first" else -1
if reduction != 0.0:
if not (reduction <= 1.0 and reduction > 0.0):
raise ValueError(
"`reduction` value must lie between 0.0 and 1.0")
# layers in each dense block
if type(nb_layers_per_block) is list or type(
nb_layers_per_block) is tuple:
nb_layers = list(nb_layers_per_block) # Convert tuple to list
if len(nb_layers) != nb_dense_block:
raise ValueError(
"If `nb_dense_block` is a list, its length must match "
"the number of layers provided by `nb_layers`.")
final_nb_layer = nb_layers[-1]
nb_layers = nb_layers[:-1]
else:
if nb_layers_per_block == -1:
assert (depth - 4) % 3 == 0, ("Depth must be 3 N + 4 "
"if nb_layers_per_block == -1")
count = int((depth - 4) / 3)
if bottleneck:
count = count // 2
nb_layers = [count for _ in range(nb_dense_block)]
final_nb_layer = count
else:
final_nb_layer = nb_layers_per_block
nb_layers = [nb_layers_per_block] * nb_dense_block
# compute initial nb_filter if -1, else accept users initial nb_filter
if nb_filter <= 0:
nb_filter = 2 * growth_rate
# compute compression factor
compression = 1.0 - reduction
# Initial convolution
if subsample_initial_block:
initial_kernel = (7, 7)
initial_strides = (2, 2)
else:
initial_kernel = (3, 3)
initial_strides = (1, 1)
x = Conv2D(
nb_filter,
initial_kernel,
kernel_initializer="he_normal",
padding="same",
name="initial_conv2D",
strides=initial_strides,
use_bias=False,
kernel_regularizer=l2(weight_decay),
)(img_input)
if subsample_initial_block:
x = BatchNormalization(axis=concat_axis,
epsilon=1.1e-5,
name="initial_bn")(x)
x = Activation("relu")(x)
x = MaxPooling2D((3, 3), strides=(2, 2), padding="same")(x)
# Add dense blocks
for block_idx in range(nb_dense_block - 1):
x, nb_filter = __dense_block(
x,
nb_layers[block_idx],
nb_filter,
growth_rate,
bottleneck=bottleneck,
dropout_rate=dropout_rate,
weight_decay=weight_decay,
block_prefix="dense_%i" % block_idx,
)
# add transition_block
x = __transition_block(
x,
nb_filter,
compression=compression,
weight_decay=weight_decay,
block_prefix="tr_%i" % block_idx,
transition_pooling=transition_pooling,
)
nb_filter = int(nb_filter * compression)
# The last dense_block does not have a transition_block
x, nb_filter = __dense_block(
x,
final_nb_layer,
nb_filter,
growth_rate,
bottleneck=bottleneck,
dropout_rate=dropout_rate,
weight_decay=weight_decay,
block_prefix="dense_%i" % (nb_dense_block - 1),
)
x = BatchNormalization(axis=concat_axis,
epsilon=1.1e-5,
name="final_bn")(x)
x = Activation("relu")(x)
if include_top:
if pooling == "avg":
x = GlobalAveragePooling2D()(x)
elif pooling == "max":
x = GlobalMaxPooling2D()(x)
x = Dense(nb_classes, activation=activation)(x)
else:
if pooling == "avg":
x = GlobalAveragePooling2D()(x)
elif pooling == "max":
x = GlobalMaxPooling2D()(x)
return x
def VGG_CNN_F(arch_mode='full',
weights='imagenet',
input_shape=(224, 224, 3),
pooling=None,
classes=1000):
""" This function constructs VGG_CNN_F and is based on
https://github.com/keras-team/keras/blob/master/keras/applications/vgg16.py
Arguments:
arch_mode: {'full', 'rmlast', 'notop'}
"""
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 arch_mode == 'full' and classes != 1000:
raise ValueError('If using `weights` as imagenet with `arch_mode`'
' as `full`, `classes` should be 1000')
img_input = Input(shape=input_shape)
x = Conv2D(64, (11, 11), strides=(4, 4), activation='relu',
name='conv1')(img_input) # conv1
x = LRN(n=5, alpha=0.0005, beta=0.75, k=2, name='norm1')(x) # norm1
x = MaxPooling2D(pool_size=(3, 3),
strides=(2, 2),
padding='same',
name='pool1')(x) # pool1
x = ZeroPadding2D(padding=(2, 2))(x)
x = Conv2D(256, (5, 5), strides=(1, 1), activation='relu',
name='conv2')(x) # conv2
x = LRN(n=5, alpha=0.0005, beta=0.75, k=2, name='norm2')(x) # norm2
x = MaxPooling2D(pool_size=(3, 3), strides=(2, 2),
name='pool2')(x) # pool2
x = ZeroPadding2D(padding=(1, 1))(x)
x = Conv2D(256, (3, 3), strides=(1, 1), activation='relu',
name='conv3')(x) # conv3
x = ZeroPadding2D(padding=(1, 1))(x)
x = Conv2D(256, (3, 3), strides=(1, 1), activation='relu',
name='conv4')(x) # conv4
x = ZeroPadding2D(padding=(1, 1))(x)
x = Conv2D(256, (3, 3), strides=(1, 1), activation='relu',
name='conv5')(x) # conv5
x = MaxPooling2D(pool_size=(3, 3), strides=(2, 2),
name='pool5')(x) # pool5
if arch_mode == 'notop':
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
else:
x = Flatten()(x)
x = Dense(4096, activation='relu', name='fc6')(x) # fc6
x = Dropout(0.5)(x) # drop6
x = Dense(4096, activation='relu', name='fc7')(x) # fc7
x = Dropout(0.5)(x) # drop7
if arch_mode == 'full':
x = Dense(classes, name='fc8')(x) # fc8
x = Activation('softmax', name='prob')(x) # prob
elif arch_mode != 'rmlast':
raise ValueError(
'arch_mode: {} is not supported.'.format(arch_mode))
inputs = img_input
# Create model.
model = Model(inputs, x, name='vgg_cnn_f')
# load weights
if weights == 'imagenet':
if arch_mode == 'full':
weights_path = 'init_weights/vgg_cnn_f.h5'
elif arch_mode == 'rmlast':
weights_path = 'init_weights/vgg_cnn_f_rmlast.h5'
elif arch_mode == 'notop':
weights_path = 'init_weights/vgg_cnn_f_notop.h5'
model.load_weights(weights_path)
if K.backend() == 'theano':
raise NotImplemented('Support for loading imagenet weights with '
'theano backend is not implemented yet!')
if K.image_data_format() == 'channels_first':
raise NotImplemented(
'Support for loading imagenet weights with '
'channels_first image data format is not implemented yet!')
elif weights is not None:
model.load_weights(weights)
return model
def SENET50(include_top=True, weights='vggface',
input_tensor=None, input_shape=None,
pooling=None,
classes=8631):
input_shape = _obtain_input_shape(input_shape,
default_size=224,
min_size=197,
data_format=K.image_data_format(),
require_flatten=include_top,
weights=weights)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
if K.image_data_format() == 'channels_last':
bn_axis = 3
else:
bn_axis = 1
x = Conv2D(
64, (7, 7), use_bias=False, strides=(2, 2), padding='same',
name='conv1/7x7_s2')(img_input)
x = BatchNormalization(axis=bn_axis, name='conv1/7x7_s2/bn')(x)
x = Activation('relu')(x)
x = MaxPooling2D((3, 3), strides=(2, 2))(x)
x = senet_conv_block(x, 3, [64, 64, 256], stage=2, block=1, strides=(1, 1))
x = senet_identity_block(x, 3, [64, 64, 256], stage=2, block=2)
x = senet_identity_block(x, 3, [64, 64, 256], stage=2, block=3)
x = senet_conv_block(x, 3, [128, 128, 512], stage=3, block=1)
x = senet_identity_block(x, 3, [128, 128, 512], stage=3, block=2)
x = senet_identity_block(x, 3, [128, 128, 512], stage=3, block=3)
x = senet_identity_block(x, 3, [128, 128, 512], stage=3, block=4)
x = senet_conv_block(x, 3, [256, 256, 1024], stage=4, block=1)
x = senet_identity_block(x, 3, [256, 256, 1024], stage=4, block=2)
x = senet_identity_block(x, 3, [256, 256, 1024], stage=4, block=3)
x = senet_identity_block(x, 3, [256, 256, 1024], stage=4, block=4)
x = senet_identity_block(x, 3, [256, 256, 1024], stage=4, block=5)
x = senet_identity_block(x, 3, [256, 256, 1024], stage=4, block=6)
x = senet_conv_block(x, 3, [512, 512, 2048], stage=5, block=1)
x = senet_identity_block(x, 3, [512, 512, 2048], stage=5, block=2)
x = senet_identity_block(x, 3, [512, 512, 2048], stage=5, block=3)
x = AveragePooling2D((7, 7), name='avg_pool')(x)
if include_top:
x = Flatten()(x)
x = Dense(classes, activation='softmax', name='classifier')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model.
model = Model(inputs, x, name='vggface_senet50')
# load weights
if weights == 'vggface':
if include_top:
weights_path = get_file('rcmalli_vggface_tf_senet50.h5',
utils.SENET50_WEIGHTS_PATH,
cache_subdir=utils.VGGFACE_DIR)
else:
weights_path = get_file('rcmalli_vggface_tf_notop_senet50.h5',
utils.SENET50_WEIGHTS_PATH_NO_TOP,
cache_subdir=utils.VGGFACE_DIR)
model.load_weights(weights_path)
if K.backend() == 'theano':
layer_utils.convert_all_kernels_in_model(model)
if include_top:
maxpool = model.get_layer(name='avg_pool')
shape = maxpool.output_shape[1:]
dense = model.get_layer(name='classifier')
layer_utils.convert_dense_weights_data_format(dense, shape,
'channels_first')
if K.image_data_format() == 'channels_first' and K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
elif weights is not None:
model.load_weights(weights)
return model
def test_keras_2_image_bias(self):
#define Keras model and get prediction
input_shape1 = (100, 60, 3)
input_shape2 = (23, 45, 3)
data1 = Input(shape=input_shape1)
data2 = Input(shape=input_shape2)
a_pool = GlobalMaxPooling2D()(data1)
b_pool = GlobalMaxPooling2D()(data2)
output = keras.layers.add([a_pool, b_pool])
model = Model(inputs=[data1, data2], outputs=output)
data1 = np.ones(input_shape1)
data2 = np.ones(input_shape2)
keras_input1 = np.ones(input_shape1)
keras_input2 = np.ones(input_shape2)
data1[:, :, 0] = 100.0
data1[:, :, 1] = 79.0
data1[:, :, 2] = 194.0
data2[:, :, 0] = 130.0
data2[:, :, 1] = 91.0
data2[:, :, 2] = 11.0
red_bias1 = -88.0
green_bias1 = -2
blue_bias1 = -40
red_bias2 = -100.0
green_bias2 = -29
blue_bias2 = -15
keras_input1[:, :, 0] = data1[:, :, 2] + blue_bias1
keras_input1[:, :, 1] = data1[:, :, 1] + green_bias1
keras_input1[:, :, 2] = data1[:, :, 0] + red_bias1
keras_input2[:, :, 0] = data2[:, :, 0] + red_bias2
keras_input2[:, :, 1] = data2[:, :, 1] + green_bias2
keras_input2[:, :, 2] = data2[:, :, 2] + blue_bias2
keras_preds = model.predict([
np.expand_dims(keras_input1, axis=0),
np.expand_dims(keras_input2, axis=0)
])
keras_preds = keras_preds.flatten()
#convert to coreml and get predictions
model_dir = tempfile.mkdtemp()
model_path = os.path.join(model_dir, 'keras.mlmodel')
from coremltools.converters import keras as keras_converter
coreml_model = keras_converter.convert(
model,
input_names=['data1', 'data2'],
output_names=['output'],
image_input_names=['data1', 'data2'],
red_bias={
'data1': red_bias1,
'data2': red_bias2
},
green_bias={
'data1': green_bias1,
'data2': green_bias2
},
blue_bias={
'data1': blue_bias1,
'data2': blue_bias2
},
is_bgr={
'data1': True,
'data2': False
})
#coreml_model.save(model_path)
#coreml_model = coremltools.models.MLModel(model_path)
coreml_input_dict = dict()
coreml_input_dict["data1"] = PIL.Image.fromarray(data1.astype(
np.uint8))
coreml_input_dict["data2"] = PIL.Image.fromarray(data2.astype(
np.uint8))
coreml_preds = coreml_model.predict(
coreml_input_dict)['output'].flatten()
#compare
self.assertEquals(len(keras_preds), len(coreml_preds))
max_relative_error = compare_models(keras_preds, coreml_preds)
self.assertAlmostEquals(max(max_relative_error, .001),
.001,
delta=1e-6)
if os.path.exists(model_dir):
shutil.rmtree(model_dir)
def VGG16(include_top=True, weights='imagenet',
input_tensor=None, input_shape=None,
pooling=None,
classes=1000):
"""Instantiates the VGG16 architecture.
Optionally loads weights pre-trained
on ImageNet. Note that when using TensorFlow,
for best performance you should set
`image_data_format='channels_last'` in your Keras config
at ~/.keras/keras.json.
The model and the weights are compatible with both
TensorFlow and Theano. The data format
convention used by the model is the one
specified in your Keras config file.
# Arguments
include_top: whether to include the 3 fully-connected
layers at the top of the network.
weights: one of `None` (random initialization),
'imagenet' (pre-training on ImageNet),
or the path to the weights file to be loaded.
input_tensor: optional Keras tensor (i.e. output of `layers.Input()`)
to use as image input for the model.
input_shape: optional shape tuple, only to be specified
if `include_top` is False (otherwise the input shape
has to be `(224, 224, 3)` (with `channels_last` data format)
or `(3, 224, 224)` (with `channels_first` data format).
It should have exactly 3 input channels,
and width and height should be no smaller than 48.
E.g. `(200, 200, 3)` would be one valid value.
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.
"""
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')
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=224,
min_size=48,
data_format=K.image_data_format(),
require_flatten=include_top,
weights=weights)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
# Block 1
x = Conv2D(64, (3, 3), activation='relu', padding='same', name='block1_conv1')(img_input)
x = Conv2D(64, (3, 3), activation='relu', padding='same', name='block1_conv2')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block1_pool')(x)
# Block 2
x = Conv2D(128, (3, 3), activation='relu', padding='same', name='block2_conv1')(x)
x = Conv2D(128, (3, 3), activation='relu', padding='same', name='block2_conv2')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block2_pool')(x)
# Block 3
x = Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv1')(x)
x = Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv2')(x)
x = Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv3')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block3_pool')(x)
# Block 4
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv1')(x)
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv2')(x)
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv3')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block4_pool')(x)
# Block 5
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv1')(x)
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv2')(x)
x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv3')(x)
x = MaxPooling2D((2, 2), strides=(2, 2), name='block5_pool')(x)
if include_top:
# Classification block
x = Flatten(name='flatten')(x)
x = Dense(4096, activation='relu', name='fc1')(x)
x = Dense(4096, activation='relu', name='fc2')(x)
x = Dense(classes, activation='softmax', name='predictions')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model.
model = Model(inputs, x, name='vgg16')
# load weights
if weights == 'imagenet':
if include_top:
weights_path = WEIGHTS_PATH
else:
weights_path = WEIGHTS_PATH_NO_TOP
model.load_weights(weights_path)
if K.backend() == 'theano':
layer_utils.convert_all_kernels_in_model(model)
if K.image_data_format() == 'channels_first':
if include_top:
maxpool = model.get_layer(name='block5_pool')
shape = maxpool.output_shape[1:]
dense = model.get_layer(name='fc1')
layer_utils.convert_dense_weights_data_format(dense, shape, 'channels_first')
if K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
elif weights is not None:
model.load_weights(weights)
return model
def build_model_up(input_shape, n_label):
input_img = Input(shape=input_shape)
# segmentation network
seg_conv1 = conv(input_img, 32, 5, 'seg_conv1_01')
seg_conv1 = conv(seg_conv1, 32, 5, 'seg_conv1_02') # 500
seg_maxp1 = MaxPooling2D(pool_size=(2, 2))(seg_conv1) # 250
seg_conv2 = conv(seg_maxp1, 64, 5, 'seg_conv2_01') # 250
seg_conv2 = conv(seg_conv2, 64, 5, 'seg_conv2_02')
seg_conv2 = conv(seg_conv2, 64, 5, 'seg_conv2_03') # 250
seg_maxp2 = MaxPooling2D(pool_size=(2, 2))(seg_conv2) # 125
seg_conv3 = conv(seg_maxp2, 64, 5, 'seg_conv3_01') # 125
seg_conv3 = conv(seg_conv3, 64, 5, 'seg_conv3_02')
seg_conv3 = conv(seg_conv3, 64, 5, 'seg_conv3_03')
seg_conv3 = conv(seg_conv3, 64, 5, 'seg_conv3_04') # 125
seg_maxp3 = MaxPooling2D(pool_size=(2, 2))(seg_conv3) # 62
seg_conv4 = conv(seg_maxp3, 1024, 5, 'seg_conv4_01') # 62
seg_mask = Conv2D(n_label,
1,
strides=(1, 1),
padding='same',
kernel_initializer='he_normal',
activation='softmax',
name='seg_mask')(seg_conv4)
seg_upsp5 = UpSampling2D()(seg_conv4) # 124
seg_upsp5 = ZeroPadding2D(padding=((0, 1), (0, 1)))(seg_upsp5) # 125
seg_upsp5 = concatenate([seg_upsp5, seg_conv3], axis=3)
seg_conv5 = conv(seg_upsp5, 64, 5, 'seg_conv5_01')
#seg_conv5 = conv(seg_upsp5, 64, 5, 'seg_conv5_02')
#seg_conv5 = conv(seg_upsp5, 64, 5, 'seg_conv5_03')
#seg_conv5 = conv(seg_upsp5, 64, 5, 'seg_conv5_04') # 125
seg_upsp6 = UpSampling2D()(seg_conv5) # 256
seg_upsp6 = concatenate([seg_upsp6, seg_conv2], axis=3)
seg_conv6 = conv(seg_upsp6, 64, 5, 'seg_conv6_01')
#seg_conv6 = conv(seg_upsp6, 64, 5, 'seg_conv6_02')
#seg_conv6 = conv(seg_upsp6, 64, 5, 'seg_conv6_03') # 250
seg_upsp7 = UpSampling2D()(seg_conv6) # 512
seg_upsp7 = concatenate([seg_upsp7, seg_conv1], axis=3)
seg_conv7 = conv(seg_upsp7, 32, 5, 'seg_conv7_01')
#seg_conv7 = conv(seg_upsp7, 32, 5, 'seg_conv7_02') # 500
seg_output = Conv2D(n_label,
1,
strides=(1, 1),
padding='same',
kernel_initializer='he_normal',
activation='softmax',
name='seg_output')(seg_conv7)
# decision network
dec_conv1 = concatenate([seg_conv4, seg_mask], axis=3)
dec_conv1 = MaxPooling2D(pool_size=(2, 2))(dec_conv1)
dec_conv2 = conv(dec_conv1, 16, 5, 'dec_conv2_01')
dec_conv2 = MaxPooling2D(pool_size=(2, 2))(dec_conv2)
dec_conv3 = conv(dec_conv2, 32, 5, 'dec_conv3_01')
dec_conv3 = MaxPooling2D(pool_size=(2, 2))(dec_conv3)
dec_conv4 = conv(dec_conv3, 64, 5, 'dec_conv4_01')
# classification
class1 = GlobalMaxPooling2D()(dec_conv4)
class2 = GlobalAveragePooling2D()(dec_conv4)
class3 = GlobalAveragePooling2D()(seg_mask)
class4 = GlobalMaxPooling2D()(seg_mask)
classes = concatenate([class1, class2, class3, class4], axis=1)
class_output = Dense(n_label, activation='softmax',
name='class_output')(classes)
model = Model(inputs=input_img, outputs=[seg_output, class_output])
return model
def ResNet50(include_top=True,
weights='imagenet',
input_tensor=None,
input_shape=(224, 224, 3),
pooling="max",
classes=2):
if weights not in {'imagenet', None}:
raise ValueError('The `weights` argument should be either '
'`None` (random initialization) or `imagenet` '
'(pre-training on ImageNet).')
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')
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=224,
min_size=197,
data_format=K.image_data_format(),
require_flatten=include_top)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
if K.image_data_format() == 'channels_last':
bn_axis = 3
else:
bn_axis = 1
x = ZeroPadding2D((3, 3))(img_input)
x = Conv2D(64, (7, 7), strides=(2, 2), name='conv1')(x)
x = BatchNormalization(axis=bn_axis, name='bn_conv1')(x)
x = Activation('relu')(x)
x = MaxPooling2D((3, 3), strides=(2, 2))(x)
x = conv_block(x, 3, [64, 64, 256], stage=2, block='a', strides=(1, 1))
x = identity_block(x, 3, [64, 64, 256], stage=2, block='b')
x = identity_block(x, 3, [64, 64, 256], stage=2, block='c')
x = conv_block(x, 3, [128, 128, 512], stage=3, block='a')
x = identity_block(x, 3, [128, 128, 512], stage=3, block='b')
x = identity_block(x, 3, [128, 128, 512], stage=3, block='c')
x = identity_block(x, 3, [128, 128, 512], stage=3, block='d')
x = conv_block(x, 3, [256, 256, 1024], stage=4, block='a')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='b')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='c')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='d')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='e')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='f')
x = conv_block(x, 3, [512, 512, 2048], stage=5, block='a')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='b')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='c')
x = AveragePooling2D((7, 7), name='avg_pool')(x)
if include_top:
x = Flatten()(x)
x = Dense(classes, activation='softmax', name='fc1000')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model.
model = Model(inputs, x, name='resnet50')
# load weights
if weights == 'imagenet':
if include_top:
weights_path = get_file(
'resnet50_weights_tf_dim_ordering_tf_kernels.h5',
WEIGHTS_PATH,
cache_subdir='models',
md5_hash='a7b3fe01876f51b976af0dea6bc144eb')
else:
weights_path = get_file(
'resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5',
WEIGHTS_PATH_NO_TOP,
cache_subdir='models',
md5_hash='a268eb855778b3df3c7506639542a6af')
model.load_weights(weights_path)
if K.image_data_format() == 'channels_first':
if include_top:
maxpool = model.get_layer(name='avg_pool')
shape = maxpool.output_shape[1:]
dense = model.get_layer(name='fc1000')
layer_utils.convert_dense_weights_data_format(
dense, shape, 'channels_first')
if K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
return model
def simple_resnet_1113(
input_shape=(75, 75, 3), KernelSize=(5, 5), Momentum=0.99):
X_input = Input(input_shape)
#input_CNN = ZeroPadding2D((0, 0))(X_input)
input_CNN = BatchNormalization(momentum=Momentum)(X_input)
## Input Layer
input_CNN = Conv2D(32, kernel_size=KernelSize, padding='same',
name='c11')(input_CNN)
input_CNN = BatchNormalization(momentum=Momentum, name='b11')(input_CNN)
input_CNN = Activation('elu')(input_CNN)
input_CNN = MaxPooling2D((2, 2), strides=(2, 2), name='m11')(input_CNN)
#input_CNN = Dropout(0.25)(input_CNN)
input_CNN = Conv2D(64, kernel_size=KernelSize, padding='same',
name='c12')(input_CNN)
input_CNN = BatchNormalization(momentum=Momentum, name='b12')(input_CNN)
input_CNN = Activation('elu')(input_CNN)
input_CNN = MaxPooling2D((2, 2), strides=(2, 2), name='m12')(input_CNN)
#input_CNN = Dropout(0.25)(input_CNN)
## First Residual
input_CNN_residual = BatchNormalization(momentum=Momentum)(input_CNN)
input_CNN_residual = Conv2D(128, kernel_size=KernelSize,
padding='same')(input_CNN_residual)
input_CNN_residual = BatchNormalization(
momentum=Momentum)(input_CNN_residual)
input_CNN_residual = Activation('elu')(input_CNN_residual)
input_CNN_residual = Dropout(0.25)(input_CNN_residual)
input_CNN_residual = Conv2D(64, kernel_size=KernelSize,
padding='same')(input_CNN_residual)
input_CNN_residual = BatchNormalization(
momentum=Momentum)(input_CNN_residual)
input_CNN_residual = Activation('elu')(input_CNN_residual)
input_CNN_residual = Dropout(0.25)(input_CNN_residual)
input_CNN_residual = Add()([input_CNN_residual, input_CNN])
## Top CNN
top_CNN = Conv2D(128, kernel_size=KernelSize,
padding='same')(input_CNN_residual)
top_CNN = BatchNormalization(momentum=Momentum)(top_CNN)
top_CNN = Activation('elu')(top_CNN)
top_CNN = MaxPooling2D((2, 2), strides=(2, 2))(top_CNN)
top_CNN = Conv2D(256, kernel_size=KernelSize, padding='same')(top_CNN)
top_CNN = BatchNormalization(momentum=Momentum)(top_CNN)
top_CNN = Activation('elu')(top_CNN)
top_CNN = Dropout(0.25)(top_CNN)
top_CNN = MaxPooling2D((2, 2), strides=(2, 2))(top_CNN)
top_CNN = Conv2D(512, kernel_size=KernelSize, padding='same')(top_CNN)
top_CNN = BatchNormalization(momentum=Momentum)(top_CNN)
top_CNN = Activation('elu')(top_CNN)
top_CNN = Dropout(0.25)(top_CNN)
top_CNN = MaxPooling2D((2, 2), strides=(2, 2))(top_CNN)
top_CNN = GlobalMaxPooling2D()(top_CNN)
#Dense Layers
#X = Flatten()(top_CNN)
X = Dense(512)(top_CNN)
X = BatchNormalization(momentum=Momentum)(X)
X = Activation('elu')(X)
X = Dropout(0.25)(X)
X = Dense(256)(X)
X = BatchNormalization(momentum=Momentum)(X)
X = Activation('elu')(X)
X = Dropout(0.25)(X)
X = Dense(1, activation='sigmoid')(X)
model = Model(inputs=X_input, outputs=X, name='simple_resnet')
return model
def NASNet(input_shape=None,
penultimate_filters=4032,
nb_blocks=6,
stem_filters=96,
skip_reduction=True,
use_auxiliary_branch=False,
filters_multiplier=2,
dropout=0.5,
weight_decay=5e-5,
include_top=True,
weights=None,
input_tensor=None,
pooling=None,
classes=1000,
default_size=None):
"""Instantiates a NASNet architecture.
Note that only TensorFlow is supported for now,
therefore it only works with the data format
`image_data_format='channels_last'` 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 or
`(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.
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
nb_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_filters: number of filters in the initial stem block
skip_reduction: Whether to skip the reduction step at the tail
end of the network. Set to `False` for CIFAR models.
use_auxiliary_branch: Whether to use the auxiliary branch during
training or evaluation.
filters_multiplier: controls the width of the network.
- If `filters_multiplier` < 1.0, proportionally decreases the number
of filters in each layer.
- If `filters_multiplier` > 1.0, proportionally increases the number
of filters in each layer.
- If `filters_multiplier` = 1, default number of filters from the paper
are used at each layer.
dropout: dropout rate
weight_decay: l2 regularization weight
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.
default_size: specifies the default image size of the model
# 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.
"""
if K.backend() != 'tensorflow':
raise RuntimeError('Only Tensorflow backend is currently supported, '
'as other backends do not support '
'separable convolution.')
if weights not in {'imagenet', None}:
raise ValueError('The `weights` argument should be either '
'`None` (random initialization) or `imagenet` '
'(pre-training on ImageNet).')
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 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=K.image_data_format(),
require_flatten=include_top,
weights=weights)
if K.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.')
K.set_image_data_format('channels_last')
old_data_format = 'channels_first'
else:
old_data_format = None
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
assert penultimate_filters % 24 == 0, "`penultimate_filters` needs to be divisible " \
"by 24."
channel_dim = 1 if K.image_data_format() == 'channels_first' else -1
filters = penultimate_filters // 24
if not skip_reduction:
x = Conv2D(stem_filters, (3, 3), strides=(2, 2), padding='valid', use_bias=False, name='stem_conv1',
kernel_initializer='he_normal', kernel_regularizer=l2(weight_decay))(img_input)
else:
x = Conv2D(stem_filters, (3, 3), strides=(1, 1), padding='same', use_bias=False, name='stem_conv1',
kernel_initializer='he_normal', kernel_regularizer=l2(weight_decay))(img_input)
x = BatchNormalization(axis=channel_dim, momentum=_BN_DECAY, epsilon=_BN_EPSILON,
name='stem_bn1')(x)
p = None
if not skip_reduction: # imagenet / mobile mode
x, p = _reduction_A(x, p, filters // (filters_multiplier ** 2), weight_decay, id='stem_1')
x, p = _reduction_A(x, p, filters // filters_multiplier, weight_decay, id='stem_2')
for i in range(nb_blocks):
x, p = _normal_A(x, p, filters, weight_decay, id='%d' % (i))
x, p0 = _reduction_A(x, p, filters * filters_multiplier, weight_decay, id='reduce_%d' % (nb_blocks))
p = p0 if not skip_reduction else p
for i in range(nb_blocks):
x, p = _normal_A(x, p, filters * filters_multiplier, weight_decay, id='%d' % (nb_blocks + i + 1))
auxiliary_x = None
if not skip_reduction: # imagenet / mobile mode
if use_auxiliary_branch:
auxiliary_x = _add_auxiliary_head(x, classes, weight_decay)
x, p0 = _reduction_A(x, p, filters * filters_multiplier ** 2, weight_decay, id='reduce_%d' % (2 * nb_blocks))
if skip_reduction: # CIFAR mode
if use_auxiliary_branch:
auxiliary_x = _add_auxiliary_head(x, classes, weight_decay)
p = p0 if not skip_reduction else p
for i in range(nb_blocks):
x, p = _normal_A(x, p, filters * filters_multiplier ** 2, weight_decay, id='%d' % (2 * nb_blocks + i + 1))
x = Activation('relu')(x)
if include_top:
x = GlobalAveragePooling2D()(x)
x = Dropout(dropout)(x)
x = Dense(classes, activation='softmax', kernel_regularizer=l2(weight_decay), name='predictions')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model.
if use_auxiliary_branch:
model = Model(inputs, [x, auxiliary_x], name='NASNet_with_auxiliary')
else:
model = Model(inputs, x, name='NASNet')
# load weights
if weights == 'imagenet':
if default_size == 224: # mobile version
if include_top:
if use_auxiliary_branch:
weight_path = NASNET_MOBILE_WEIGHT_PATH_WITH_AUXULARY
model_name = 'nasnet_mobile_with_aux.h5'
else:
weight_path = NASNET_MOBILE_WEIGHT_PATH
model_name = 'nasnet_mobile.h5'
else:
if use_auxiliary_branch:
weight_path = NASNET_MOBILE_WEIGHT_PATH_WITH_AUXULARY_NO_TOP
model_name = 'nasnet_mobile_with_aux_no_top.h5'
else:
weight_path = NASNET_MOBILE_WEIGHT_PATH_NO_TOP
model_name = 'nasnet_mobile_no_top.h5'
weights_file = get_file(model_name, weight_path, cache_subdir='models')
model.load_weights(weights_file, by_name=True)
elif default_size == 331: # large version
if include_top:
if use_auxiliary_branch:
weight_path = NASNET_LARGE_WEIGHT_PATH_WITH_auxiliary
model_name = 'nasnet_large_with_aux.h5'
else:
weight_path = NASNET_LARGE_WEIGHT_PATH
model_name = 'nasnet_large.h5'
else:
if use_auxiliary_branch:
weight_path = NASNET_LARGE_WEIGHT_PATH_WITH_auxiliary_NO_TOP
model_name = 'nasnet_large_with_aux_no_top.h5'
else:
weight_path = NASNET_LARGE_WEIGHT_PATH_NO_TOP
model_name = 'nasnet_large_no_top.h5'
weights_file = get_file(model_name, weight_path, cache_subdir='models')
model.load_weights(weights_file, by_name=True)
else:
raise ValueError('ImageNet weights can only be loaded on NASNetLarge or NASNetMobile')
if old_data_format:
K.set_image_data_format(old_data_format)
return model
def ResNet50(include_top=True,
weights='imagenet',
input_tensor=None,
input_shape=None,
pooling=None,
classes=1000):
"""Instantiates the ResNet50 architecture.
Optionally loads weights pre-trained
on ImageNet. Note that when using TensorFlow,
for best performance you should set
`image_data_format="channels_last"` in your Keras config
at ~/.keras/keras.json.
The model and the weights are compatible with both
TensorFlow and Theano. The data format
convention used by the model is the one
specified in your Keras config file.
# Arguments
include_top: whether to include the fully-connected
layer at the top of the network.
weights: one of `None` (random initialization)
or "imagenet" (pre-training on ImageNet).
input_tensor: optional Keras tensor (i.e. output of `layers.Input()`)
to use as image input for the model.
input_shape: optional shape tuple, only to be specified
if `include_top` is False (otherwise the input shape
has to be `(224, 224, 3)` (with `channels_last` data format)
or `(3, 224, 244)` (with `channels_first` data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 197.
E.g. `(200, 200, 3)` would be one valid value.
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.
"""
if weights not in {'imagenet', None}:
raise ValueError('The `weights` argument should be either '
'`None` (random initialization) or `imagenet` '
'(pre-training on ImageNet).')
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')
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=224,
min_size=197,
data_format=K.image_data_format(),
require_flatten=include_top)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
if K.image_data_format() == 'channels_last':
bn_axis = 3
else:
bn_axis = 1
x = ZeroPadding2D((3, 3))(img_input)
x = Conv2D(64, (7, 7), strides=(2, 2), name='conv1')(x)
x = BatchNormalization(axis=bn_axis, name='bn_conv1')(x)
x = Activation('relu')(x)
x = MaxPooling2D((3, 3), strides=(2, 2))(x)
x = conv_block(x, 3, [64, 64, 256], stage=2, block='a', strides=(1, 1))
x = identity_block(x, 3, [64, 64, 256], stage=2, block='b')
x = identity_block(x, 3, [64, 64, 256], stage=2, block='c')
x = conv_block(x, 3, [128, 128, 512], stage=3, block='a')
x = identity_block(x, 3, [128, 128, 512], stage=3, block='b')
x = identity_block(x, 3, [128, 128, 512], stage=3, block='c')
x = identity_block(x, 3, [128, 128, 512], stage=3, block='d')
x = conv_block(x, 3, [256, 256, 1024], stage=4, block='a')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='b')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='c')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='d')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='e')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='f')
x = conv_block(x, 3, [512, 512, 2048], stage=5, block='a')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='b')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='c')
x = AveragePooling2D((7, 7), name='avg_pool')(x)
if include_top:
x = Flatten()(x)
x = Dense(classes, activation='softmax', name='fc1000')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model.
model = Model(inputs, x, name='resnet50')
# load weights
if weights == 'imagenet':
if include_top:
weights_path = get_file(
'resnet50_weights_tf_dim_ordering_tf_kernels.h5',
WEIGHTS_PATH,
cache_subdir='models',
md5_hash='a7b3fe01876f51b976af0dea6bc144eb')
else:
weights_path = get_file(
'resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5',
WEIGHTS_PATH_NO_TOP,
cache_subdir='models',
md5_hash='a268eb855778b3df3c7506639542a6af')
model.load_weights(weights_path)
if K.backend() == 'theano':
layer_utils.convert_all_kernels_in_model(model)
if K.image_data_format() == 'channels_first':
if include_top:
maxpool = model.get_layer(name='avg_pool')
shape = maxpool.output_shape[1:]
dense = model.get_layer(name='fc1000')
layer_utils.convert_dense_weights_data_format(
dense, shape, 'channels_first')
if K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
return model
def SqueezeNet(include_top=True,
weights="imagenet",
input_tensor=None,
input_shape=None,
pooling=None,
classes=1000):
if weights not in {'imagenet', None}:
raise ValueError('The `weights` argument should be either '
'`None` (random initialization) or `imagenet` '
'(pre-training on ImageNet).')
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')
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=224,
min_size=48,
data_format=K.image_data_format(),
require_flatten=include_top)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
x = Convolution2D(64,
kernel_size=(3, 3),
strides=(2, 2),
padding="same",
activation="relu",
name='conv1')(img_input)
x = MaxPooling2D(pool_size=(3, 3),
strides=(2, 2),
name='maxpool1',
padding="valid")(x)
x = _fire(x, (16, 64, 64), name="fire2")
x = _fire(x, (16, 64, 64), name="fire3")
x = MaxPooling2D(pool_size=(3, 3),
strides=(2, 2),
name='maxpool3',
padding="valid")(x)
x = _fire(x, (32, 128, 128), name="fire4")
x = _fire(x, (32, 128, 128), name="fire5")
x = MaxPooling2D(pool_size=(3, 3),
strides=(2, 2),
name='maxpool5',
padding="valid")(x)
x = _fire(x, (48, 192, 192), name="fire6")
x = _fire(x, (48, 192, 192), name="fire7")
x = _fire(x, (64, 256, 256), name="fire8")
x = _fire(x, (64, 256, 256), name="fire9")
if include_top:
x = Dropout(0.5, name='dropout9')(x)
x = Convolution2D(classes, (1, 1), padding='valid', name='conv10')(x)
x = AveragePooling2D(pool_size=(13, 13), name='avgpool10')(x)
x = Flatten(name='flatten10')(x)
x = Activation("softmax", name='softmax')(x)
else:
if pooling == "avg":
x = GlobalAveragePooling2D(name="avgpool10")(x)
else:
x = GlobalMaxPooling2D(name="maxpool10")(x)
model = Model(img_input, x, name="squeezenet")
if weights == 'imagenet':
weights_path = get_file('squeezenet_weights.h5',
WEIGHTS_PATH,
cache_subdir='models')
model.load_weights(weights_path)
return model
def build_model(lr, l2, activation='sigmoid'):
##############
# BRANCH MODEL
##############
regul = regularizers.l2(l2)
optim = Adam(lr=lr)
kwargs = {'padding': 'same', 'kernel_regularizer': regul}
inp = Input(shape=img_shape) # 384x384x1
x = Conv2D(64, (9, 9), strides=2, activation='relu', **kwargs)(inp)
x = MaxPooling2D((2, 2), strides=(2, 2), name='maxpool1')(x) # 96x96x64
for i in range(2):
x = BatchNormalization(name=f'bn{i+1}')(x)
x = Conv2D(64, (3, 3), activation='relu', **kwargs)(x)
x = MaxPooling2D((2, 2), strides=(2, 2),
name='layer1.maxpool')(x) # 48x48x64
x = BatchNormalization()(x)
x = Conv2D(128, (1, 1), activation='relu', **kwargs)(x) # 48x48x128
for _ in range(4):
x = subblock(x, 64, **kwargs)
x = MaxPooling2D((2, 2), strides=(2, 2),
name='layer2.maxpool')(x) # 24x24x128
x = BatchNormalization()(x)
x = Conv2D(256, (1, 1), activation='relu', **kwargs)(x) # 24x24x256
for _ in range(4):
x = subblock(x, 64, **kwargs)
x = MaxPooling2D((2, 2), strides=(2, 2),
name='layer3.maxpool')(x) # 12x12x256
x = BatchNormalization()(x)
x = Conv2D(384, (1, 1), activation='relu', **kwargs)(x) # 12x12x384
for _ in range(4):
x = subblock(x, 96, **kwargs)
x = MaxPooling2D((2, 2), strides=(2, 2),
name='layer4.maxpool')(x) # 6x6x384
x = BatchNormalization()(x)
x = Conv2D(512, (1, 1), activation='relu', **kwargs)(x) # 6x6x512
for _ in range(4):
x = subblock(x, 128, **kwargs)
x = GlobalMaxPooling2D()(x) # 512
branch_model = Model(inp, x)
############
# HEAD MODEL
############
mid = 32
xa_inp = Input(shape=branch_model.output_shape[1:])
xb_inp = Input(shape=branch_model.output_shape[1:])
x1 = Lambda(lambda x: x[0] * x[1])([xa_inp, xb_inp])
x2 = Lambda(lambda x: x[0] + x[1])([xa_inp, xb_inp])
x3 = Lambda(lambda x: K.abs(x[0] - x[1]))([xa_inp, xb_inp])
x4 = Lambda(lambda x: K.square(x))(x3)
x = Concatenate()([x1, x2, x3, x4])
x = Reshape((4, branch_model.output_shape[1], 1), name='reshape1')(x)
# Per feature NN with shared weight is implemented using CONV2D with appropriate stride.
x = Conv2D(mid, (4, 1), activation='relu', padding='valid')(x)
x = Reshape((branch_model.output_shape[1], mid, 1))(x)
x = Conv2D(1, (1, mid), activation='linear', padding='valid')(x)
x = Flatten(name='flatten')(x)
# Weighted sum implemented as a Dense layer.
x = Dense(1, use_bias=True, activation=activation,
name='weighted-average')(x)
head_model = Model([xa_inp, xb_inp], x, name='head')
########################
# SIAMESE NEURAL NETWORK
########################
# Complete model is constructed by calling the branch model on each input image,
# and then the head model on the resulting 512-vectors.
img_a = Input(shape=img_shape)
img_b = Input(shape=img_shape)
xa = branch_model(img_a)
xb = branch_model(img_b)
x = head_model([xa, xb])
model = Model([img_a, img_b], x)
model.compile(optim,
loss='binary_crossentropy',
metrics=['binary_crossentropy', 'acc'])
return model, branch_model, head_model
def ResNet152(include_top=True, weights=None,
input_tensor=None, input_shape=None,
large_input=False, pooling=None,
classes=1000):
"""Instantiate the ResNet152 architecture.
Keyword arguments:
include_top -- whether to include the fully-connected layer at the
top of the network. (default True)
weights -- one of `None` (random initialization) or "imagenet"
(pre-training on ImageNet). (default None)
input_tensor -- optional Keras tensor (i.e. output of `layers.Input()`)
to use as image input for the model.(default None)
input_shape -- optional shape tuple, only to be specified if
`include_top` is False (otherwise the input shape has to be
`(224, 224, 3)` (with `channels_last` data format) or
`(3, 224, 224)` (with `channels_first` data format). It should
have exactly 3 inputs channels, and width and height should be
no smaller than 197. E.g. `(200, 200, 3)` would be one valid value.
(default None)
large_input -- if True, then the input shape expected will be
`(448, 448, 3)` (with `channels_last` data format) or
`(3, 448, 448)` (with `channels_first` data format). (default False)
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.
(default None)
classes -- optional number of classes to classify image into, only
to be specified if `include_top` is True, and if no `weights`
argument is specified. (default 1000)
Returns:
A Keras model instance.
Raises:
ValueError: in case of invalid argument for `weights`,
or invalid input shape.
"""
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')
eps = 1.1e-5
if large_input:
img_size = 448
else:
img_size = 224
# Determine proper input shape
input_shape = _obtain_input_shape(input_shape,
default_size=img_size,
min_size=197,
data_format=K.image_data_format(),
require_flatten=include_top,
weights=weights)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
# handle dimension ordering for different backends
if K.image_dim_ordering() == 'tf':
bn_axis = 3
else:
bn_axis = 1
x = ZeroPadding2D((3, 3), name='conv1_zeropadding')(img_input)
x = Conv2D(64, (7, 7), strides=(2, 2), name='conv1', use_bias=False)(x)
x = BatchNormalization(epsilon=eps, axis=bn_axis, name='bn_conv1')(x)
x = Scale(axis=bn_axis, name='scale_conv1')(x)
x = Activation('relu', name='conv1_relu')(x)
x = MaxPooling2D((3, 3), strides=(2, 2), name='pool1')(x)
x = conv_block(x, 3, [64, 64, 256], stage=2, block='a', strides=(1, 1))
x = identity_block(x, 3, [64, 64, 256], stage=2, block='b')
x = identity_block(x, 3, [64, 64, 256], stage=2, block='c')
x = conv_block(x, 3, [128, 128, 512], stage=3, block='a')
for i in range(1, 8):
x = identity_block(x, 3, [128, 128, 512], stage=3, block='b' + str(i))
x = conv_block(x, 3, [256, 256, 1024], stage=4, block='a')
for i in range(1, 36):
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='b' + str(i))
x = conv_block(x, 3, [512, 512, 2048], stage=5, block='a')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='b')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='c')
if large_input:
x = AveragePooling2D((14, 14), name='avg_pool')(x)
else:
x = AveragePooling2D((7, 7), name='avg_pool')(x)
# include classification layer by default, not included for feature extraction
if include_top:
x = Flatten()(x)
x = Dense(classes, activation='softmax', name='fc1000')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling == 'max':
x = GlobalMaxPooling2D()(x)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
# Create model.
model = Model(inputs, x, name='resnet152')
# load weights
if weights == 'imagenet':
if include_top:
weights_path = get_file('resnet152_weights_tf.h5',
WEIGHTS_PATH,
cache_subdir='models',
md5_hash='cdb18a2158b88e392c0905d47dcef965')
else:
weights_path = get_file('resnet152_weights_tf_notop.h5',
WEIGHTS_PATH_NO_TOP,
cache_subdir='models',
md5_hash='4a90dcdafacbd17d772af1fb44fc2660')
model.load_weights(weights_path, by_name=True)
if K.backend() == 'theano':
layer_utils.convert_all_kernels_in_model(model)
if include_top:
maxpool = model.get_layer(name='avg_pool')
shape = maxpool.output_shape[1:]
dense = model.get_layer(name='fc1000')
layer_utils.convert_dense_weights_data_format(dense, shape, 'channels_first')
if K.image_data_format() == 'channels_first' and K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
elif weights is not None:
model.load_weights(weights)
return model
def SqueezeNet(include_top=True, weights='imagenet',
input_tensor=None, input_shape=None,
pooling=None,
classes=1000):
"""Instantiates the SqueezeNet architecture.
"""
if weights not in {'imagenet', None}:
raise ValueError('The `weights` argument should be either '
'`None` (random initialization) or `imagenet` '
'(pre-training on ImageNet).')
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')
input_shape = _obtain_input_shape(input_shape,
default_size=227,
min_size=48,
data_format=K.image_data_format(),
require_flatten=include_top)
if input_tensor is None:
img_input = Input(shape=input_shape)
else:
if not K.is_keras_tensor(input_tensor):
img_input = Input(tensor=input_tensor, shape=input_shape)
else:
img_input = input_tensor
x = Convolution2D(64, (3, 3), strides=(2, 2), padding='valid', name='conv1')(img_input)
x = Activation('relu', name='relu_conv1')(x)
x = MaxPooling2D(pool_size=(3, 3), strides=(2, 2), name='pool1')(x)
x = fire_module(x, fire_id=2, squeeze=16, expand=64)
x = fire_module(x, fire_id=3, squeeze=16, expand=64)
x = MaxPooling2D(pool_size=(3, 3), strides=(2, 2), name='pool3')(x)
x = fire_module(x, fire_id=4, squeeze=32, expand=128)
x = fire_module(x, fire_id=5, squeeze=32, expand=128)
x = MaxPooling2D(pool_size=(3, 3), strides=(2, 2), name='pool5')(x)
x = fire_module(x, fire_id=6, squeeze=48, expand=192)
x = fire_module(x, fire_id=7, squeeze=48, expand=192)
x = fire_module(x, fire_id=8, squeeze=64, expand=256)
x = fire_module(x, fire_id=9, squeeze=64, expand=256)
if include_top:
# It's not obvious where to cut the network...
# Could do the 8th or 9th layer... some work recommends cutting earlier layers.
x = Dropout(0.5, name='drop9')(x)
x = Convolution2D(classes, (1, 1), padding='valid', name='conv10')(x)
x = Activation('relu', name='relu_conv10')(x)
x = GlobalAveragePooling2D()(x)
x = Activation('softmax', name='loss')(x)
else:
if pooling == 'avg':
x = GlobalAveragePooling2D()(x)
elif pooling=='max':
x = GlobalMaxPooling2D()(x)
elif pooling==None:
pass
else:
raise ValueError("Unknown argument for 'pooling'=" + pooling)
# Ensure that the model takes into account
# any potential predecessors of `input_tensor`.
if input_tensor is not None:
inputs = get_source_inputs(input_tensor)
else:
inputs = img_input
model = Model(inputs, x, name='squeezenet')
# load weights
if weights == 'imagenet':
if include_top:
weights_path = get_file('squeezenet_weights_tf_dim_ordering_tf_kernels.h5',
WEIGHTS_PATH,
cache_subdir='models')
else:
weights_path = get_file('squeezenet_weights_tf_dim_ordering_tf_kernels_notop.h5',
WEIGHTS_PATH_NO_TOP,
cache_subdir='models')
model.load_weights(weights_path, by_name=True)
if K.backend() == 'theano':
layer_utils.convert_all_kernels_in_model(model)
if K.image_data_format() == 'channels_first':
if K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
return model
