def depth_to_space(input, scale, data_format=None):
''' Uses phase shift algorithm to convert channels/depth for spatial resolution '''
if data_format is None:
data_format = image_data_format()
data_format = data_format.lower()
input = _preprocess_conv2d_input(input, data_format)
out = tf.depth_to_space(input, scale)
out = _postprocess_conv2d_output(out, data_format)
return out
def _normalize_data_format(data_format):
if data_format is None:
data_format = image_data_format()
if data_format == 'channels_last':
return 'nxc'
if data_format == 'channels_first':
return 'ncx'
if data_format in ['nxc', 'ncx']:
return data_format
raise ValueError("Unrecognized data_format '{}'".format(data_format))
def deconv3d(x,
kernel,
output_shape,
strides=(1, 1, 1),
padding='valid',
data_format=None,
filter_shape=None):
'''3D deconvolution (transposed convolution).
# Arguments
kernel: kernel tensor.
output_shape: desired dimensions of output.
strides: strides tuple.
padding: string, "same" or "valid".
data_format: "channels_last" or "channels_first".
Whether to use Theano or TensorFlow dimension ordering
in inputs/kernels/ouputs.
'''
flip_filters = False
if data_format is None:
data_format = image_data_format()
if data_format not in {'channels_first', 'channels_last'}:
raise ValueError('Unknown data_format: ' + str(data_format))
if data_format == 'channels_last':
output_shape = (output_shape[0], output_shape[4], output_shape[1],
output_shape[2], output_shape[3])
x = _preprocess_conv3d_input(x, data_format)
kernel = _preprocess_conv3d_kernel(kernel, data_format)
kernel = kernel.dimshuffle((1, 0, 2, 3, 4))
th_padding = _preprocess_padding(padding)
if hasattr(kernel, '_keras_shape'):
kernel_shape = kernel._keras_shape
else:
# Will only work if `kernel` is a shared variable.
kernel_shape = kernel.eval().shape
filter_shape = _preprocess_conv3d_filter_shape(filter_shape, data_format)
filter_shape = tuple(filter_shape[i] for i in (1, 0, 2, 3, 4))
conv_out = T.nnet.abstract_conv.conv3d_grad_wrt_inputs(
x,
kernel,
output_shape,
filter_shape=filter_shape,
border_mode=th_padding,
subsample=strides,
filter_flip=not flip_filters)
conv_out = _postprocess_conv3d_output(conv_out, x, padding, kernel_shape,
strides, data_format)
return conv_out
def depth_to_space(input, scale, data_format=None):
''' Uses phase shift algorithm to convert channels/depth for spatial resolution '''
if data_format is None:
data_format = image_data_format()
if data_format == 'channels_first':
data_format = 'NCHW'
else:
data_format = 'NHWC'
data_format = data_format.lower()
out = tf.depth_to_space(input, scale, data_format=data_format)
return out
def depth_to_space(input, scale, data_format=None):
''' Uses phase shift algorithm to convert channels/depth for spatial resolution '''
if data_format is None:
data_format = image_data_format()
data_format = data_format.lower()
input = _preprocess_conv2d_input(input, data_format)
b, k, row, col = input.shape
out_channels = k // (scale**2)
x = T.reshape(input, (b, scale, scale, out_channels, row, col))
x = T.transpose(x, (0, 3, 4, 1, 5, 2))
out = T.reshape(x, (b, out_channels, row * scale, col * scale))
out = _postprocess_conv2d_output(out, input, None, None, None, data_format)
return out
def _bias_add(X, data_format):
x, bias = X
from keras.backend import image_data_format, ndim, reshape
if data_format is None:
data_format = image_data_format()
if data_format not in {'channels_first', 'channels_last'}:
raise ValueError('Unknown data_format ' + str(data_format))
if ndim(bias) != 1 and ndim(bias) != ndim(x) - 1:
raise ValueError('Unexpected bias dimensions %d, '
'expect to be 1 or %d dimensions'
% (ndim(bias), ndim(x) - 1))
bias_shape = tuple(bias.size())
ndim_x = len(x.size())
ndim_bias = len(bias_shape)
if ndim_x == 5:
if data_format == 'channels_first':
if ndim_bias == 1:
bias = reshape(bias, (1, bias_shape[0], 1, 1, 1))
else:
bias = reshape(bias, (1, bias_shape[3]) + bias_shape[:3])
elif data_format == 'channels_last':
if ndim_bias == 1:
bias = reshape(bias, (1, 1, 1, 1, bias_shape[0]))
else:
bias = reshape(bias, (1,) + bias_shape)
elif ndim_x == 4:
if data_format == 'channels_first':
if ndim_bias == 1:
bias = reshape(bias, (1, bias_shape[0], 1, 1))
else:
bias = reshape(bias, (1, bias_shape[2]) + bias_shape[:2])
elif data_format == 'channels_last':
if ndim_bias == 1:
bias = reshape(bias, (1, 1, 1, bias_shape[0]))
else:
bias = reshape(bias, (1,) + bias_shape)
elif ndim_x == 3:
if data_format == 'channels_first':
if ndim_bias == 1:
bias = reshape(bias, (1, bias_shape[0], 1))
else:
bias = reshape(bias, (1, bias_shape[1], bias_shape[0]))
elif data_format == 'channels_last':
if ndim_bias == 1:
bias = reshape(bias, (1, 1, bias_shape[0]))
else:
bias = reshape(bias, (1,) + bias_shape)
return x.add(bias.expand_as(x))
def deconv3d(x,
kernel,
output_shape,
strides=(1, 1, 1),
padding='valid',
data_format='default',
image_shape=None,
filter_shape=None):
'''3D deconvolution (i.e. transposed convolution).
# Arguments
x: input tensor.
kernel: kernel tensor.
output_shape: 1D int tensor for the output shape.
strides: strides tuple.
padding: string, "same" or "valid".
data_format: "tf" or "th".
Whether to use Theano or TensorFlow dimension ordering
for inputs/kernels/ouputs.
# Returns
A tensor, result of transposed 3D convolution.
# Raises
ValueError: if `data_format` is neither `tf` or `th`.
'''
if data_format == 'default':
data_format = image_data_format()
if data_format not in {'channels_first', 'channels_last'}:
raise ValueError('Unknown data_format ' + str(data_format))
x = _preprocess_conv3d_input(x, data_format)
output_shape = _preprocess_deconv_output_shape(x, output_shape,
data_format)
kernel = _preprocess_conv3d_kernel(kernel, data_format)
kernel = tf.transpose(kernel, (0, 1, 2, 4, 3))
padding = _preprocess_padding(padding)
strides = (1, ) + strides + (1, )
x = tf.nn.conv3d_transpose(x,
kernel,
output_shape,
strides,
padding=padding)
return _postprocess_conv3d_output(x, data_format)
def depth_to_space(input, scale, data_format=None):
''' Uses phase shift algorithm to convert channels/depth for spatial resolution '''
if data_format is None:
data_format = image_data_format()
data_format = data_format.lower()
input = _preprocess_conv2d_input(input, data_format)
b, k, row, col = input.shape
output_shape = (b, k // (scale ** 2), row * scale, col * scale)
out = T.zeros(output_shape)
r = scale
for y, x in itertools.product(range(scale), repeat=2):
out = T.inc_subtensor(out[:, :, y::r, x::r], input[:, r * y + x:: r * r, :, :])
out = _postprocess_conv2d_output(out, input, None, None, None, data_format)
return out
def bias_add(x, bias, data_format=None):
if data_format is None:
data_format = image_data_format()
if data_format not in _CONV_DATA_FORMAT:
raise PlaidMLKerasException(
'Unrecognized data_format given to bias_add: "{}"; '.format(data_format) +
'only "channels_first" and "channels_last" recognized.')
if ndim(x) > 2:
if data_format == 'channels_first':
try:
bias_dims = bias.tensor.shape.dims
except AttributeError:
bias_dims = bias.shape
x += reshape(bias, (1, bias_dims[0]) + (1,) * (ndim(x) - 2))
elif data_format == 'channels_last':
x += bias
else:
x += bias
return x
def bias_add(x, bias, data_format=None):
logger.debug('bias_add(x: {}, bias: {}, data_format: {})'.format(x, bias, data_format))
if data_format is None:
data_format = image_data_format()
if data_format not in _CONV_DATA_FORMAT:
raise PlaidMLKerasException(
"Unrecognized data_format given to bias_add: '{}'; ".format(data_format) +
"only 'channels_first' and 'channels_last' recognized.")
if ndim(x) > 2:
if data_format == 'channels_first':
try:
bias_dims = bias.tensor.shape.dims
except AttributeError:
bias_dims = bias.shape
x += reshape(bias, (1, bias_dims[0]) + (1,) * (ndim(x) - 2))
elif data_format == 'channels_last':
x += bias
else:
x += bias
return x
def depth_to_space(input, scale, data_format=None):
""" Uses phase shift algorithm to convert channels/depth for spatial resolution.
# Arguments
input: Input tensor
scale: n `int` that is `>= 2`. The size of the spatial block.
data_format: 'channels_first' or 'channels_last'.
Whether to use Theano or TensorFlow dimension
ordering in inputs/kernels/ouputs.
# Returns
TODO (PR welcome): Filling this section.
"""
if data_format is None:
data_format = image_data_format()
data_format = data_format.lower()
input = _preprocess_conv2d_input(input, data_format)
out = tf.depth_to_space(input, scale)
out = _postprocess_conv2d_output(out, data_format)
return out
def spatial_reflection_2d_padding(x,
padding=((1, 1), (1, 1)),
data_format=None):
"""
Pad the 2nd and 3rd dimensions of a 4D tensor.
:param x: Input tensor
:param padding: Shape of padding to use
:param data_format: Tensorflow vs Theano convention ('channels_last', 'channels_first')
:return: Tensorflow tensor
"""
assert len(padding) == 2
assert len(padding[0]) == 2
assert len(padding[1]) == 2
if data_format is None:
data_format = image_data_format()
if data_format not in {'channels_first', 'channels_last'}:
raise ValueError('Unknown data_format ' + str(data_format))
if data_format == 'channels_first':
pattern = [[0, 0], [0, 0], list(padding[0]), list(padding[1])]
else:
pattern = [[0, 0], list(padding[0]), list(padding[1]), [0, 0]]
return tf.pad(x, pattern, "REFLECT")