mindsponge.common.rots_from_tensor

View Source On Gitee
mindsponge.common.rots_from_tensor(rots, use_numpy=False)[source]
Amortize and split the 3*3 rotation matrix corresponding to the last two axes of input Tensor

to obtain each component of the rotation matrix, inverse of ‘rots_to_tensor’.

Parameters

  • rots (Tensor) – Represent the rotation matrix, shape is \((..., 3, 3)\) .

  • use_numpy (bool) – Whether to use numpy to calculate. Default: False.

Returns

Tuple, rots represented by vectors, rots is \((xx, xy, xz, yx, yy, yz, zx, zy, zz)\).

Supported Platforms:

Ascend GPU

Examples

>>> import numpy as np
>>> import mindspore as ms
>>> from mindspore import Tensor
>>> from mindsponge.common.geometry import rots_from_tensor
>>> input_0 = Tensor(np.ones((256, 3, 3)), ms.float32)
>>> output = rots_from_tensor(input_0)
>>> print(len(output), output[0].shape)
9, (256,)