ops

square_center_crop

square_center_crop(img, noise=0.0)

crop a single image of shape (channels, width, height) around the center with optional random shift along the longest dimension.

Parameters

• img (Tensor) – single image tensor of shape (channels, width, height).

• noise (float) – for shifting the crop window along the longest dimension Defaults to 0.0 (no noise).

Example

>>> import torch
>>> from hearth.vision.ops import square_center_crop
>>>
>>> img = torch.randint(0, 255, size=(3, 345, 256))
>>> square_center_crop(img).shape
torch.Size([3, 256, 256])

>>> square_center_crop(img, noise=0.5).shape
torch.Size([3, 256, 256])

>>> square_center_crop(img.transpose(1, 2), noise=0.5).shape
torch.Size([3, 256, 256])

Return type

Tensor

---