def __init__(self,
kernel_size=(3, 3),
num_filters=None,
strides=1,
auto_pad=True,
padding_mode='zero',
activation=None,
normalization=None,
use_spectral=False,
use_bias=False,
dilation=1,
groups=1,
add_noise=False,
noise_intensity=0.005,
dropout_rate=0,
name=None,
depth_multiplier=None,
sequence_rank='cna',
**kwargs):
super(TransConv2d_Block, self).__init__(name=name)
if sequence_rank in ['cna', 'nac']:
self.sequence_rank = sequence_rank
self.kernel_size = kernel_size
self.num_filters = num_filters
self.strides = strides
self.auto_pad = auto_pad
self.use_bias = use_bias
self.dilation = dilation
self.groups = groups
self.add_noise = add_noise
self.noise_intensity = noise_intensity
self.dropout_rate = dropout_rate
self.depth_multiplier = depth_multiplier
self.use_spectral = use_spectral
if not self.use_spectral:
self.conv = TransConv2d(kernel_size=self.kernel_size,
num_filters=self.num_filters,
strides=self.strides,
auto_pad=self.auto_pad,
activation=None,
use_bias=self.use_bias,
dilation=self.dilation,
groups=self.groups,
name=self._name,
depth_multiplier=self.depth_multiplier)
self.norm = get_normalization(normalization)
self.activation = get_activation(activation)
self.droupout = None
def __init__(self,
kernel_size=(3, 3),
depth_multiplier=1,
strides=1,
auto_pad=True,
padding=None,
padding_mode='zero',
activation=None,
normalization=None,
use_spectral=False,
use_bias=False,
dilation=1,
groups=1,
add_noise=False,
noise_intensity=0.005,
dropout_rate=0,
name=None,
keep_output=False,
sequence_rank='cna',
**kwargs):
super(DepthwiseConv2d_Block, self).__init__(name=name)
if sequence_rank in ['cna', 'nac']:
self.sequence_rank = sequence_rank
self.kernel_size = kernel_size
self.depth_multiplier = depth_multiplier
self.strides = strides
self.auto_pad = auto_pad
self.padding = 0
self.padding_mode = padding_mode
# if self.auto_pad == False:
# self.padding = 0
# else:
# self.padding= tuple([n-2 for n in list(self.kernel_size)]) if hasattr(self.kernel_size,'__len__') else
# self.kernel_size-2
self.use_bias = use_bias
self.dilation = dilation
self.add_noise = add_noise
self.noise_intensity = noise_intensity
self.dropout_rate = dropout_rate
self.conv = None
self.norm = get_normalization(normalization)
self.use_spectral = use_spectral
self.activation = get_activation(activation)
self.droupout = None
self.keep_output = keep_output
self._name = name
def __init__(self,
kernel_size=(3, 3, 3),
num_filters=32,
strides=1,
input_shape=None,
auto_pad=True,
activation='leaky_relu',
normalization=None,
use_bias=False,
dilation=1,
groups=1,
add_noise=False,
noise_intensity=0.001,
dropout_rate=0,
name=None,
sequence_rank='cna',
**kwargs):
super(TransConv3d_Block, self).__init__(name=name)
if sequence_rank in ['cna', 'nac']:
self.sequence_rank = sequence_rank
if add_noise:
noise = tf.keras.layers.GaussianNoise(noise_intensity)
self.add(noise)
self._conv = TransConv3d(kernel_size=kernel_size,
num_filters=num_filters,
strides=strides,
input_shape=input_shape,
auto_pad=auto_pad,
activation=None,
use_bias=use_bias,
dilation=dilation,
groups=groups)
self.add(self._conv)
self.norm = get_normalization(normalization)
if self.norm is not None:
self.add(self.norm)
self.activation = get_activation(snake2camel(activation))
if self.activation is not None:
self.add(self.activation)
if dropout_rate > 0:
self.drop = Dropout(dropout_rate)
self.add(self.drop)
def __init__(self,
kernel_size=3,
num_filters=None,
strides=1,
auto_pad=True,
padding_mode='zero',
activation=None,
normalization=None,
use_spectral=False,
use_bias=False,
dilation=1,
groups=1,
add_noise=False,
noise_intensity=0.005,
dropout_rate=0,
name=None,
depth_multiplier=None,
keep_output=False,
sequence_rank='cna',
**kwargs):
super(Conv1d_Block, self).__init__(name=name, keep_output=keep_output)
if sequence_rank in ['cna', 'nac']:
self.sequence_rank = sequence_rank
else:
self.sequence_rank = 'cna'
self.kernel_size = kernel_size
self.num_filters = num_filters
self.strides = strides
self.auto_pad = auto_pad
self.padding = 0
self.padding_mode = padding_mode
# if self.auto_pad == False:
# self.padding = 0
# else:
# self.padding= tuple([n-2 for n in list(self.kernel_size)]) if hasattr(self.kernel_size,'__len__') else
# self.kernel_size-2
self.use_bias = use_bias
self.dilation = dilation
self.groups = groups
self.depth_multiplier = depth_multiplier
self.add_noise = add_noise
self.noise_intensity = noise_intensity
self.dropout_rate = dropout_rate
norm = get_normalization(normalization)
conv = Conv1d(kernel_size=self.kernel_size,
num_filters=self.num_filters,
strides=self.strides,
auto_pad=self.auto_pad,
padding_mode=self.padding_mode,
activation=None,
use_bias=self.use_bias,
dilation=self.dilation,
groups=self.groups,
name=self._name,
depth_multiplier=self.depth_multiplier)
self.use_spectral = use_spectral
# if isinstance(norm, SpectralNorm):
# self.use_spectral = True
# norm = None
# conv= nn.utils.spectral_norm(conv)
if (hasattr(self, 'sequence_rank') and self.sequence_rank
== 'cna') or not hasattr(self, 'sequence_rank'):
self.conv = conv
self.norm = norm
self.activation = get_activation(activation)
elif self.sequence_rank == 'nac':
self.norm = norm
self.activation = get_activation(activation)
self.conv = conv
def __init__(self,
kernel_size=(3, 3),
num_filters=None,
strides=1,
auto_pad=True,
padding_mode='zero',
activation=None,
normalization=None,
use_spectral=False,
use_bias=False,
dilation=1,
groups=1,
add_noise=False,
noise_intensity=0.005,
dropout_rate=0,
name=None,
depth_multiplier=None,
keep_output=False,
sequence_rank='cna',
**kwargs):
super(Conv2d_Block, self).__init__(name=name)
if sequence_rank in ['cna', 'nac']:
self.sequence_rank = sequence_rank
self.kernel_size = kernel_size
self.num_filters = num_filters
self.strides = strides
self.keep_output = keep_output
padding = kwargs.get('padding', None)
if 'padding' in kwargs:
kwargs.pop('padding')
if isinstance(padding, str) and auto_pad == False:
auto_pad = (padding.lower() == 'same')
elif isinstance(padding, int) and padding > 0:
padding = _pair(padding)
auto_pad = False
elif isinstance(padding, tuple):
auto_pad = False
pass
self.auto_pad = auto_pad
self.padding = padding
self.use_bias = use_bias
self.dilation = dilation
self.groups = groups
self.add_noise = add_noise
self.noise_intensity = noise_intensity
self.dropout_rate = dropout_rate
self.depth_multiplier = depth_multiplier
self.use_spectral = use_spectral
if not self.use_spectral:
self.conv = Conv2d(kernel_size=self.kernel_size,
num_filters=self.num_filters,
strides=self.strides,
auto_pad=self.auto_pad,
activation=None,
use_bias=self.use_bias,
dilation=self.dilation,
groups=self.groups,
depth_multiplier=self.depth_multiplier,
padding=self.padding,
**kwargs)
self.norm = get_normalization(normalization)
self.activation = get_activation(activation)
self.droupout = None
def __init__(self,
num_filters=None,
activation=None,
normalization=None,
use_spectral=False,
use_bias=False,
add_noise=False,
noise_intensity=0.005,
dropout_rate=0,
name=None,
depth_multiplier=None,
keep_output=False,
sequence_rank='cna',
**kwargs):
super(FullConnect_Block, self).__init__(name=name,
keep_output=keep_output)
if sequence_rank in ['fna', 'naf', 'afn']:
self.sequence_rank = sequence_rank
else:
self.sequence_rank = 'fna'
self.num_filters = num_filters
self.use_bias = use_bias
self.add_noise = add_noise
self.noise_intensity = noise_intensity
self.dropout_rate = dropout_rate
self.droupout = None
self.depth_multiplier = depth_multiplier
self.keep_output = keep_output
norm = get_normalization(normalization)
fc = Dense(num_filters=self.num_filters,
activation=None,
use_bias=self.use_bias,
depth_multiplier=self.depth_multiplier).to(self.device)
self.use_spectral = use_spectral
if isinstance(norm, SpectralNorm):
self.use_spectral = True
norm = None
fc = SpectralNorm(module=fc)
if (hasattr(self, 'sequence_rank') and self.sequence_rank
== 'fna') or not hasattr(self, 'sequence_rank'):
self.add_module('fc', fc)
self.add_module('norm', norm)
self.add_module('activation',
get_activation(activation, only_layer=True))
elif self.sequence_rank == 'naf':
self.add_module('norm', norm)
self.add_module('activation',
get_activation(activation, only_layer=True))
self.add_module('fc', fc)
elif self.sequence_rank == 'afn':
self.add_module('activation',
get_activation(activation, only_layer=True))
self.add_module('fc', fc)
self.add_module('norm', norm)
self._name = name