mindspore.ops.SpaceToBatch

class mindspore.ops.SpaceToBatch(block_size, paddings)[source]

Divides spatial dimensions into blocks and combines the block size with the original batch.

This operation will divide spatial dimensions (H, W) into blocks with block_size, the output tensor’s H and W dimension is the corresponding number of blocks after division. The output tensor’s batch dimension is the product of the original batch and the square of block_size. Before division, the spatial dimensions of the input are zero padded according to paddings if necessary.

Parameters

  • block_size (int) – The block size of dividing blocks with value greater than or euqual to 2.

  • paddings (Union[tuple, list]) – The padding values for H and W dimension, containing 2 subtraction lists. Each subtraction list contains 2 integer value. All values must be greater than 0. paddings[i] specifies the paddings for the spatial dimension i, which corresponds to the input dimension i+2. It is required that input_shape[i+2]+paddings[i][0]+paddings[i][1] is divisible by block_size.

Inputs:

  • input_x (Tensor) - The input tensor. It must be a 4-D tensor. The data type is Number.

Outputs:

Tensor, the output tensor with the same data type as input. Assume input shape is \((n, c, h, w)\) with \(block\_size\) and \(paddings\). The shape of the output tensor will be \((n', c', h', w')\), where

\(n' = n*(block\_size*block\_size)\)

\(c' = c\)

\(h' = (h+paddings[0][0]+paddings[0][1])//block\_size\)

\(w' = (w+paddings[1][0]+paddings[1][1])//block\_size\)

Raises
Supported Platforms:

Ascend GPU

Examples

>>> block_size = 2
>>> paddings = [[0, 0], [0, 0]]
>>> space_to_batch = ops.SpaceToBatch(block_size, paddings)
>>> input_x = Tensor(np.array([[[[1, 2], [3, 4]]]]), mindspore.float32)
>>> output = space_to_batch(input_x)
>>> print(output)
[[[[1.]]]
 [[[2.]]]
 [[[3.]]]
 [[[4.]]]]