tf.compat.v1.batch_to_space_nd  |  TensorFlow v2.16.1 (original) (raw)

crops

A Tensor. Must be one of the following types: int32, int64. 2-D with shape [M, 2], all values must be >= 0.crops[i] = [crop_start, crop_end] specifies the amount to crop from input dimension i + 1, which corresponds to spatial dimension i. It is required thatcrop_start[i] + crop_end[i] <= block_shape[i] * input_shape[i + 1].

This operation is equivalent to the following steps:

1. Reshape input to reshaped of shape: [block_shape[0], ..., block_shape[M-1], batch / prod(block_shape), input_shape[1], ..., input_shape[N-1]]
2. Permute dimensions of reshaped to produce permuted of shape [batch / prod(block_shape),
   input_shape[1], block_shape[0], ..., input_shape[M], block_shape[M-1],
   input_shape[M+1], ..., input_shape[N-1]]
3. Reshape permuted to produce reshaped_permuted of shape [batch / prod(block_shape),
   input_shape[1] * block_shape[0], ..., input_shape[M] * block_shape[M-1],
   input_shape[M+1], ..., input_shape[N-1]]
4. Crop the start and end of dimensions [1, ..., M] ofreshaped_permuted according to crops to produce the output of shape: [batch / prod(block_shape),
   input_shape[1] * block_shape[0] - crops[0,0] - crops[0,1], ..., input_shape[M] * block_shape[M-1] - crops[M-1,0] - crops[M-1,1],
   input_shape[M+1], ..., input_shape[N-1]]

Some examples:

(1) For the following input of shape [4, 1, 1, 1], block_shape = [2, 2], andcrops = [[0, 0], [0, 0]]:

[[[[1]]], [[[2]]], [[[3]]], [[[4]]]]

The output tensor has shape [1, 2, 2, 1] and value:

x = [[[[1], [2]], [[3], [4]]]]

(2) For the following input of shape [4, 1, 1, 3], block_shape = [2, 2], andcrops = [[0, 0], [0, 0]]:

[[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]

The output tensor has shape [1, 2, 2, 3] and value:

x = [[[[1, 2, 3], [4, 5, 6]],
      [[7, 8, 9], [10, 11, 12]]]]

(3) For the following input of shape [4, 2, 2, 1], block_shape = [2, 2], andcrops = [[0, 0], [0, 0]]:

x = [[[[1], [3]], [[9], [11]]],
     [[[2], [4]], [[10], [12]]],
     [[[5], [7]], [[13], [15]]],
     [[[6], [8]], [[14], [16]]]]

The output tensor has shape [1, 4, 4, 1] and value:

x = [[[[1],   [2],  [3],  [4]],
     [[5],   [6],  [7],  [8]],
     [[9],  [10], [11],  [12]],
     [[13], [14], [15],  [16]]]]

(4) For the following input of shape [8, 1, 3, 1], block_shape = [2, 2], andcrops = [[0, 0], [2, 0]]:

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]]]]

The output tensor has shape [2, 2, 4, 1] and value:

x = [[[[1],   [2],  [3],  [4]],
      [[5],   [6],  [7],  [8]]],
     [[[9],  [10], [11],  [12]],
      [[13], [14], [15],  [16]]]]