def _preprocess_symbolic_input(x, data_format, mode):
"""Preprocesses a tensor encoding a batch of images.
Arguments:
x: Input tensor, 3D or 4D.
data_format: Data format of the image tensor.
mode: One of "caffe", "tf" or "torch".
- caffe: will convert the images from RGB to BGR,
then will zero-center each color channel with
respect to the ImageNet dataset,
without scaling.
- tf: will scale pixels between -1 and 1,
sample-wise.
- torch: will scale pixels between 0 and 1 and then
will normalize each channel with respect to the
ImageNet dataset.
Returns:
Preprocessed tensor.
"""
global _IMAGENET_MEAN
if mode == 'tf':
x /= 127.5
x -= 1.
return x
if mode == 'torch':
x /= 255.
mean = [0.485, 0.456, 0.406]
std = [0.229, 0.224, 0.225]
else:
if data_format == 'channels_first':
# 'RGB'->'BGR'
if K.ndim(x) == 3:
x = x[::-1, ...]
else:
x = x[:, ::-1, ...]
else:
# 'RGB'->'BGR'
x = x[..., ::-1]
mean = [103.939, 116.779, 123.68]
std = None
if _IMAGENET_MEAN is None:
_IMAGENET_MEAN = constant_op.constant(-np.array(mean),
dtype=K.floatx())
# Zero-center by mean pixel
if K.dtype(x) != K.dtype(_IMAGENET_MEAN):
x = K.bias_add(x, math_ops.cast(_IMAGENET_MEAN, K.dtype(x)),
data_format)
else:
x = K.bias_add(x, _IMAGENET_MEAN, data_format)
if std is not None:
x /= std
return x
def _preprocess_symbolic_input(x, data_format, mode):
"""Preprocesses a tensor encoding a batch of images.
Arguments:
x: Input tensor, 3D or 4D.
data_format: Data format of the image tensor.
mode: One of "caffe", "tf" or "torch".
- caffe: will convert the images from RGB to BGR,
then will zero-center each color channel with
respect to the ImageNet dataset,
without scaling.
- tf: will scale pixels between -1 and 1,
sample-wise.
- torch: will scale pixels between 0 and 1 and then
will normalize each channel with respect to the
ImageNet dataset.
Returns:
Preprocessed tensor.
"""
global _IMAGENET_MEAN
if mode == 'tf':
x /= 127.5
x -= 1.
return x
if mode == 'torch':
x /= 255.
mean = [0.485, 0.456, 0.406]
std = [0.229, 0.224, 0.225]
else:
if data_format == 'channels_first':
# 'RGB'->'BGR'
if K.ndim(x) == 3:
x = x[::-1, ...]
else:
x = x[:, ::-1, ...]
else:
# 'RGB'->'BGR'
x = x[..., ::-1]
mean = [103.939, 116.779, 123.68]
std = None
if _IMAGENET_MEAN is None:
_IMAGENET_MEAN = K.constant(-np.array(mean))
# Zero-center by mean pixel
if K.dtype(x) != K.dtype(_IMAGENET_MEAN):
x = K.bias_add(x, K.cast(_IMAGENET_MEAN, K.dtype(x)), data_format)
else:
x = K.bias_add(x, _IMAGENET_MEAN, data_format)
if std is not None:
x /= std
return x
def step(self, inputs, states):
h_tm1 = states[0] # previous memory
dp_mask = states[1] # dropout matrices for recurrent units
rec_dp_mask = states[2]
if self.implementation == 2:
matrix_x = K.dot(inputs * dp_mask[0], self.kernel)
if self.use_bias:
matrix_x = K.bias_add(matrix_x, self.bias)
matrix_inner = K.dot(h_tm1 * rec_dp_mask[0],
self.recurrent_kernel[:, :2 * self.units])
x_z = matrix_x[:, :self.units]
x_r = matrix_x[:, self.units:2 * self.units]
recurrent_z = matrix_inner[:, :self.units]
recurrent_r = matrix_inner[:, self.units:2 * self.units]
z = self.recurrent_activation(x_z + recurrent_z)
r = self.recurrent_activation(x_r + recurrent_r)
x_h = matrix_x[:, 2 * self.units:]
recurrent_h = K.dot(r * h_tm1 * rec_dp_mask[0],
self.recurrent_kernel[:, 2 * self.units:])
hh = self.activation(x_h + recurrent_h)
else:
if self.implementation == 0:
x_z = inputs[:, :self.units]
x_r = inputs[:, self.units:2 * self.units]
x_h = inputs[:, 2 * self.units:]
elif self.implementation == 1:
x_z = K.dot(inputs * dp_mask[0], self.kernel_z)
x_r = K.dot(inputs * dp_mask[1], self.kernel_r)
x_h = K.dot(inputs * dp_mask[2], self.kernel_h)
if self.use_bias:
x_z = K.bias_add(x_z, self.bias_z)
x_r = K.bias_add(x_r, self.bias_r)
x_h = K.bias_add(x_h, self.bias_h)
else:
raise ValueError('Unknown `implementation` mode.')
z = self.recurrent_activation(x_z + K.dot(h_tm1 * rec_dp_mask[0],
self.recurrent_kernel_z))
r = self.recurrent_activation(x_r + K.dot(h_tm1 * rec_dp_mask[1],
self.recurrent_kernel_r))
hh = self.activation(x_h + K.dot(r * h_tm1 * rec_dp_mask[2],
self.recurrent_kernel_h))
h = z * h_tm1 + (1 - z) * hh
if 0 < self.dropout + self.recurrent_dropout:
h._uses_learning_phase = True
return h, [h]
def call(self, inputs):
output = K.local_conv1d(inputs, self.kernel, self.kernel_size, self.strides)
if self.use_bias:
output = K.bias_add(output, self.bias)
if self.activation is not None:
output = self.activation(output)
return output
def input_conv(self, x, w, b=None, padding='valid'):
conv_out = K.conv2d(x, w, strides=self.strides,
padding=padding,
data_format=self.data_format,
dilation_rate=self.dilation_rate)
if b is not None:
conv_out = K.bias_add(conv_out, b,
data_format=self.data_format)
return conv_out
def input_conv(self, x, w, b=None, padding='valid'):
conv_out = K.conv2d(x,
w,
strides=self.strides,
padding=padding,
data_format=self.data_format,
dilation_rate=self.dilation_rate)
if b is not None:
conv_out = K.bias_add(conv_out, b, data_format=self.data_format)
return conv_out
def call(self, inputs):
output = K.local_conv2d(inputs, self.kernel, self.kernel_size, self.strides,
(self.output_row, self.output_col),
self.data_format)
if self.use_bias:
output = K.bias_add(output, self.bias, data_format=self.data_format)
output = self.activation(output)
return output
def _linear(self, x, kernel, bias):
in_shape = x.shape
if len(in_shape) > 2:
x_shape = [int(dim) for dim in x.shape]
x = K.reshape(x, (x_shape[0] * x_shape[1], x_shape[2]))
x = K.dot(x, kernel)
x = K.bias_add(x, bias)
if len(in_shape) > 2:
x = K.reshape(x, (x_shape[0], x_shape[1], int(kernel.shape[1])))
return x
def _preprocess_symbolic_input(x, data_format, mode):
"""Preprocesses a symbolic image tensor.
Arguments:
x: symoblic tensor, 3D or 4D.
data_format: data format of the image tensor.
mode: One of "caffe", "tf".
- caffe: will convert the images from RGB to BGR,
then will zero-center each color channel with
respect to the ImageNet dataset,
without scaling.
- tf: will scale pixels between -1 and 1,
sample-wise.
Returns:
Preprocessed tensor.
"""
global _IMAGENET_MEAN
if mode == 'tf':
x /= 127.5
x -= 1.
return x
if data_format == 'channels_first':
# 'RGB'->'BGR'
if K.ndim(x) == 3:
x = x[::-1, ...]
else:
x = x[:, ::-1, ...]
else:
# 'RGB'->'BGR'
x = x[..., ::-1]
if _IMAGENET_MEAN is None:
_IMAGENET_MEAN = K.constant(-np.array([103.939, 116.779, 123.68]))
# Zero-center by mean pixel
if K.dtype(x) != K.dtype(_IMAGENET_MEAN):
x = K.bias_add(x, K.cast(_IMAGENET_MEAN, K.dtype(x)), data_format)
else:
x = K.bias_add(x, _IMAGENET_MEAN, data_format)
return x
def _time_distributed_dense(x,
w,
b=None,
dropout=None,
input_dim=None,
output_dim=None,
timesteps=None,
training=None):
"""Apply `y . w + b` for every temporal slice y of x.
Arguments:
x: input tensor.
w: weight matrix.
b: optional bias vector.
dropout: whether to apply dropout (same dropout mask
for every temporal slice of the input).
input_dim: integer; optional dimensionality of the input.
output_dim: integer; optional dimensionality of the output.
timesteps: integer; optional number of timesteps.
training: training phase tensor or boolean.
Returns:
Output tensor.
"""
if not input_dim:
input_dim = K.shape(x)[2]
if not timesteps:
timesteps = K.shape(x)[1]
if not output_dim:
output_dim = K.shape(w)[1]
if dropout is not None and 0. < dropout < 1.:
# apply the same dropout pattern at every timestep
ones = K.ones_like(K.reshape(x[:, 0, :], (-1, input_dim)))
dropout_matrix = K.dropout(ones, dropout)
expanded_dropout_matrix = K.repeat(dropout_matrix, timesteps)
x = K.in_train_phase(x * expanded_dropout_matrix, x, training=training)
# collapse time dimension and batch dimension together
x = K.reshape(x, (-1, input_dim))
x = K.dot(x, w)
if b is not None:
x = K.bias_add(x, b)
# reshape to 3D tensor
if K.backend() == 'tensorflow':
x = K.reshape(x, K.stack([-1, timesteps, output_dim]))
x.set_shape([None, None, output_dim])
else:
x = K.reshape(x, (-1, timesteps, output_dim))
return x
def call(self, inputs, training=None):
outputs = K.depthwise_conv2d(
inputs,
self.depthwise_kernel,
strides=self.strides,
padding=self.padding,
dilation_rate=self.dilation_rate,
data_format=self.data_format)
if self.bias:
outputs = K.bias_add(outputs, self.bias, data_format=self.data_format)
if self.activation is not None:
return self.activation(outputs)
return outputs
def step(self, inputs, states):
h_tm1 = states[0]
c_tm1 = states[1]
dp_mask = states[2]
rec_dp_mask = states[3]
if self.implementation == 2:
z = K.dot(inputs * dp_mask[0], self.kernel)
z += K.dot(h_tm1 * rec_dp_mask[0], self.recurrent_kernel)
if self.use_bias:
z = K.bias_add(z, self.bias)
z0 = z[:, :self.units]
z1 = z[:, self.units:2 * self.units]
z2 = z[:, 2 * self.units:3 * self.units]
z3 = z[:, 3 * self.units:]
i = self.recurrent_activation(z0)
f = self.recurrent_activation(z1)
c = f * c_tm1 + i * self.activation(z2)
o = self.recurrent_activation(z3)
else:
if self.implementation == 0:
x_i = inputs[:, :self.units]
x_f = inputs[:, self.units:2 * self.units]
x_c = inputs[:, 2 * self.units:3 * self.units]
x_o = inputs[:, 3 * self.units:]
elif self.implementation == 1:
x_i = K.dot(inputs * dp_mask[0], self.kernel_i) + self.bias_i
x_f = K.dot(inputs * dp_mask[1], self.kernel_f) + self.bias_f
x_c = K.dot(inputs * dp_mask[2], self.kernel_c) + self.bias_c
x_o = K.dot(inputs * dp_mask[3], self.kernel_o) + self.bias_o
else:
raise ValueError('Unknown `implementation` mode.')
i = self.recurrent_activation(x_i + K.dot(h_tm1 * rec_dp_mask[0],
self.recurrent_kernel_i))
f = self.recurrent_activation(x_f + K.dot(h_tm1 * rec_dp_mask[1],
self.recurrent_kernel_f))
c = f * c_tm1 + i * self.activation(
x_c + K.dot(h_tm1 * rec_dp_mask[2], self.recurrent_kernel_c))
o = self.recurrent_activation(x_o + K.dot(h_tm1 * rec_dp_mask[3],
self.recurrent_kernel_o))
h = o * self.activation(c)
if 0 < self.dropout + self.recurrent_dropout:
h._uses_learning_phase = True
return h, [h, c]
def call(self, inputs, training=None):
outputs = K.depthwise_conv2d(inputs,
self.depthwise_kernel,
strides=self.strides,
padding=self.padding,
dilation_rate=self.dilation_rate,
data_format=self.data_format)
if self.bias:
outputs = K.bias_add(outputs,
self.bias,
data_format=self.data_format)
if self.activation is not None:
return self.activation(outputs)
return outputs
def call(self, inputs, training=None):
outputs = K.conv2d(inputs,
self.compute_spectral_normal(training),
strides=self.strides,
padding=self.padding,
data_format=self.data_format,
dilation_rate=self.dilation_rate)
if self.bias is not None:
outputs = K.bias_add(outputs,
self.bias,
data_format=self.data_format)
if self.activation is not None:
return self.activation(outputs)
return outputs
def call(self, inputs):
frep_parts = tf.split(inputs, self.splitaxis, -1)
convs = []
for i, frep_part in enumerate(frep_parts):
individual_channels = tf.split(frep_part, frep_part.shape[-1], -1)
for ind_ch in individual_channels:
conv = K.local_conv2d(ind_ch, self.kernels[i],
self.kernel_size, self.strides,
(self.output_row, self.output_col),
self.data_format)
convs.append(conv)
outputs = tf.concat(convs, -1)
if self.use_bias:
outputs = K.bias_add(outputs,
self.bias,
data_format=self.data_format)
outputs = self.activation(outputs)
return outputs
def step(self, inputs, states):
if self.implementation == 0:
h = inputs
else:
if 0 < self.dropout < 1:
h = K.dot(inputs * states[1], self.kernel)
else:
h = K.dot(inputs, self.kernel)
if self.bias is not None:
h = K.bias_add(h, self.bias)
prev_output = states[0]
if 0 < self.recurrent_dropout < 1:
prev_output *= states[2]
output = h + K.dot(prev_output, self.recurrent_kernel)
if self.activation is not None:
output = self.activation(output)
# Properly set learning phase on output tensor.
if 0 < self.dropout + self.recurrent_dropout:
output._uses_learning_phase = True
return output, [output]
