#include <array_ops.h>
SpaceToBatch for N-D tensors of type T.
This operation divides "spatial" dimensions [1, ..., M]
of the input into a grid of blocks of shape block_shape
, and interleaves these blocks with the "batch" dimension (0) such that in the output, the spatial dimensions [1, ..., M]
correspond to the position within the grid, and the batch dimension combines both the position within a spatial block and the original batch position. Prior to division into blocks, the spatial dimensions of the input are optionally zero padded according to paddings
. See below for a precise description.
Arguments:
input_shape = [batch] + spatial_shape + remaining_shape
, where spatial_shape has M
dimensions.[M]
, all values must be >= 1.[M, 2]
, all values must be >= 0. paddings[i] = [pad_start, pad_end]
specifies the padding for input dimension i + 1
, which corresponds to spatial dimension i
. It is required that block_shape[i]
divides input_shape[i + 1] + pad_start + pad_end
.This operation is equivalent to the following steps:
[1, ..., M]
of the input according to paddings
to produce padded
of shape padded_shape
.padded
to reshaped_padded
of shape:[batch] + [padded_shape[1] / block_shape[0], block_shape[0], ..., padded_shape[M] / block_shape[M-1], block_shape[M-1]] + remaining_shapereshaped_padded
to produce permuted_reshaped_padded
of shape:block_shape + [batch] + [padded_shape[1] / block_shape[0], ..., padded_shape[M] / block_shape[M-1]] + remaining_shapepermuted_reshaped_padded
to flatten block_shape
into the batch dimension, producing an output tensor of shape:[batch * prod(block_shape)] + [padded_shape[1] / block_shape[0], ..., padded_shape[M] / block_shape[M-1]] + remaining_shapeSome examples:
(1) For the following input of shape [1, 2, 2, 1]
, block_shape = [2, 2]
, and paddings = [[0, 0], [0, 0]]
:
x = [[[[1], [2]], [[3], [4]]]]
The output tensor has shape [4, 1, 1, 1]
and value:
[[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
(2) For the following input of shape [1, 2, 2, 3]
, block_shape = [2, 2]
, and paddings = [[0, 0], [0, 0]]
:
x = [[[[1, 2, 3], [4, 5, 6]], [[7, 8, 9], [10, 11, 12]]]]
The output tensor has shape [4, 1, 1, 3]
and value:
[[[1, 2, 3]], [[4, 5, 6]], [[7, 8, 9]], [[10, 11, 12]]]
(3) For the following input of shape [1, 4, 4, 1]
, block_shape = [2, 2]
, and paddings = [[0, 0], [0, 0]]
:
x = [[[[1], [2], [3], [4]], [[5], [6], [7], [8]], [[9], [10], [11], [12]], [[13], [14], [15], [16]]]]
The output tensor has shape [4, 2, 2, 1]
and value:
x = [[[[1], [3]], [[9], [11]]], [[[2], [4]], [[10], [12]]], [[[5], [7]], [[13], [15]]], [[[6], [8]], [[14], [16]]]]
(4) For the following input of shape [2, 2, 4, 1]
, block_shape = [2, 2]
, and paddings = [[0, 0], [2, 0]]
:
x = [[[[1], [2], [3], [4]], [[5], [6], [7], [8]]], [[[9], [10], [11], [12]], [[13], [14], [15], [16]]]]
The output tensor has shape [8, 1, 3, 1]
and value:
x = [[[[0], [1], [3]]], [[[0], [9], [11]]], [[[0], [2], [4]]], [[[0], [10], [12]]], [[[0], [5], [7]]], [[[0], [13], [15]]], [[[0], [6], [8]]], [[[0], [14], [16]]]]
Among others, this operation is useful for reducing atrous convolution into regular convolution.
Returns:
Output
: The output tensor. Constructors and Destructors | |
---|---|
SpaceToBatchND(const ::tensorflow::Scope & scope, ::tensorflow::Input input, ::tensorflow::Input block_shape, ::tensorflow::Input paddings) |
Public attributes | |
---|---|
output |
Public functions | |
---|---|
node() const | ::tensorflow::Node * |
operator::tensorflow::Input() const | |
operator::tensorflow::Output() const |
::tensorflow::Output output
SpaceToBatchND( const ::tensorflow::Scope & scope, ::tensorflow::Input input, ::tensorflow::Input block_shape, ::tensorflow::Input paddings )
::tensorflow::Node * node() const
operator::tensorflow::Input() const
operator::tensorflow::Output() const
© 2018 The TensorFlow Authors. All rights reserved.
Licensed under the Creative Commons Attribution License 3.0.
Code samples licensed under the Apache 2.0 License.
https://www.tensorflow.org/api_docs/cc/class/tensorflow/ops/space-to-batch-n-d.html