mindspore.ops.ExtractVolumePatches

class mindspore.ops.ExtractVolumePatches(kernel_size, strides, padding)[源代码]

Extract patches from input and put them in the “depth” output dimension. 3D extension of extract_image_patches.

Parameters

  • kernel_size (Union[int, tuple[int], list[int]]) – A list of ints which’s length is 3 or 5. The size of the sliding window for each dimension of input. Must be: [1, 1, k_d, k_h, k_w] or [k_d, k_h, k_w]. If k_d = k_h = k_w, you can enter an integer.

  • strides (Union[int, tuple[int], list[int]]) – A list of ints which’s length is 3 or 5. How far the centers of two consecutive patches are in input. Must be: [1, 1, s_d, s_h, s_w] or [s_d, s_h, s_w]. If s_d = s_h = s_w, you can enter an integer.

  • padding (str) – A string from: “SAME”, “VALID”. The type of padding algorithm to use.

Inputs:

  • input_x (Tensor) - A Tensor. Must be one of the following types: float16, float32. 5-D Tensor with shape \((x_n, x_c, x_d, x_h, x_w)\).

Outputs:

Tensor, has the same type as input. If padding is VALID, the shape is \((x_n, k_d * k_h * k_w * x_c, 1 + (x_d - k_d) / s_d, 1 + (x_h - k_h) / s_h, 1 + (x_w - k_w) / s_w)\); if padding is SAME, the shape is \(( x_n, k_d * k_h * k_w * x_c, (x_d + s_d - 1) / s_d, (x_h + s_h - 1) / s_h, (x_w + s_w - 1) / s_w)\).

Raises

  • TypeError – If dtype of input_x is neither float16 nor float32.

  • TypeError – If kernel_size or strides is not a list, a tuple or an int.

  • TypeError – If input_x is not a tensor.

  • TypeError – If padding is not str.

  • ValueError – If the length of kernel_size is neither 3 nor 5 and kernel_size is not an integer.

  • ValueError – If the length of strides is neither 3 nor 5 and strides is not an integer.

  • ValueError – If padding is neither “VALID” nor “SAME”.

  • ValueError – If elements of kernel_size or strides are not positive integer.

  • ValueError – If input_x is not a tensor in dimension 5.

  • ValueError – If input_x’s shape has zero.

  • ValueError – If one of kernel_size or strides’ first two numbers is not 1.

  • ValueError – If padding = “VALID” and input - kernel_size is less than 0 in d, h or w dimension.

  • ValueError – If padding = “SAME” and \(padding_needed = ((input_x + strides - 1) / strides - 1) * strides + kernel_size - input\) is less than 0 in d, h or w dimension.

  • ValueError – If x_h is not 1 or x_w is not 1 and x_w + padding_needed - k_w - s_w is less than 0.

  • ValueError – If x_d * x_h * x_w is greater than 2048.

Supported Platforms:

Ascend

Example

>>> kernel_size = (1, 1, 2, 2, 2)
>>> strides = (1, 1, 1, 1, 1)
>>> padding = "VALID"
>>> input_x = P.Reshape()(Tensor(np.arange(1, 28), mstype.float16), (1, 1, 3, 3, 3))
>>> output_y = P.ExtractVolumePatches(kernel_size, strides, padding)(input_x)
>>> print(output_y.shape)
(1, 8, 2, 2, 2)