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.
Constants
String | OP_NAME | The name of this op, as known by TensorFlow core engine |
Public Methods
Output<T> |
asOutput()
Returns the symbolic handle of the tensor.
|
static <T extends TType> SpaceToBatchNd<T> | |
Output<T> |
output()
|
Inherited Methods
Constants
public static final String OP_NAME
The name of this op, as known by TensorFlow core engine
Public Methods
public Output<T> asOutput ()
Returns the symbolic handle of the tensor.
Inputs to TensorFlow operations are outputs of another TensorFlow operation. This method is used to obtain a symbolic handle that represents the computation of the input.
public static SpaceToBatchNd<T> create (Scope scope, Operand<T> input, Operand<? extends TNumber> blockShape, Operand<? extends TNumber> paddings)
Factory method to create a class wrapping a new SpaceToBatchNd operation.
Parameters
scope | current scope |
---|---|
input | N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`, where spatial_shape has `M` dimensions. |
blockShape | 1-D with shape `[M]`, all values must be >= 1. |
paddings | 2-D with shape `[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. Zero-pad the start and end of dimensions `[1, ..., M]` of the input according to `paddings` to produce `padded` of shape `padded_shape`. 2. Reshape `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_shape 3. Permute dimensions of `reshaped_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_shape 4. Reshape `permuted_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_shape Some examples: (1) For the following input of shape `[1, 2, 2, 1]`, `block_shape = [2, 2]`, and `paddings = [[0, 0], [0, 0]]`:
The output tensor has shape `[4, 1, 1, 1]` and value:
(2) For the following input of shape `[1, 2, 2, 3]`, `block_shape = [2, 2]`, and
`paddings = [[0, 0], [0, 0]]`:
The output tensor has shape `[4, 1, 1, 3]` and value:
(3) For the following input of shape `[1, 4, 4, 1]`, `block_shape = [2, 2]`, and
`paddings = [[0, 0], [0, 0]]`:
The output tensor has shape `[4, 2, 2, 1]` and value:
(4) For the following input of shape `[2, 2, 4, 1]`, block_shape = `[2, 2]`, and
paddings = `[[0, 0], [2, 0]]`:
The output tensor has shape `[8, 1, 3, 1]` and value:
Among others, this operation is useful for reducing atrous convolution into
regular convolution. |
Returns
- a new instance of SpaceToBatchNd