How to pad with zeros a tensor along some axis (Python)
This function would pad at the end of certain axis.
If you wish to pad both side, just modify it.
def pad_along_axis(array: np.ndarray, target_length: int, axis: int = 0) -> np.ndarray:
pad_size = target_length - array.shape[axis]
if pad_size <= 0:
return array
npad = [(0, 0)] * array.ndim
npad[axis] = (0, pad_size)
return np.pad(array, pad_width=npad, mode='constant', constant_values=0)
example:
>>> import numpy as np
>>> a = np.identity(5)
>>> b = pad_along_axis(a, 7, axis=1)
>>> print(a, a.shape)
[[1. 0. 0. 0. 0.]
[0. 1. 0. 0. 0.]
[0. 0. 1. 0. 0.]
[0. 0. 0. 1. 0.]
[0. 0. 0. 0. 1.]] (5, 5)
>>> print(b, b.shape)
[[1. 0. 0. 0. 0. 0. 0.]
[0. 1. 0. 0. 0. 0. 0.]
[0. 0. 1. 0. 0. 0. 0.]
[0. 0. 0. 1. 0. 0. 0.]
[0. 0. 0. 0. 1. 0. 0.]] (5, 7)
You can use np.pad()
:
a = np.ones((4, 3, 2))
# npad is a tuple of (n_before, n_after) for each dimension
npad = ((0, 0), (1, 2), (2, 1))
b = np.pad(a, pad_width=npad, mode='constant', constant_values=0)
print(b.shape)
# (4, 6, 5)
print(b)
# [[[ 0. 0. 0. 0. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 0. 0. 0.]
# [ 0. 0. 0. 0. 0.]]
# [[ 0. 0. 0. 0. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 0. 0. 0.]
# [ 0. 0. 0. 0. 0.]]
# [[ 0. 0. 0. 0. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 0. 0. 0.]
# [ 0. 0. 0. 0. 0.]]
# [[ 0. 0. 0. 0. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 1. 1. 0.]
# [ 0. 0. 0. 0. 0.]
# [ 0. 0. 0. 0. 0.]]]