torchcvnn.transforms.CenterCrop

class torchcvnn.transforms.CenterCrop(height: int, width: int)

Center crops an input array/tensor to the specified size.

This transform extracts a centered rectangular region from the input array/tensor with the specified dimensions. The crop is centered on both height and width axes.

Parameters:

• height (int) – Target height of the cropped output

• width (int) – Target width of the cropped output

Returns:

Center cropped array/tensor with shape (C, height, width),

where C is the number of channels

Return type:

np.ndarray | torch.Tensor

Examples

>>> transform = CenterCrop(height=224, width=224)
>>> output = transform(input_tensor)  # Center crops to 224x224

Notes

• If input is smaller than crop size, it will return the original input

• Crop is applied identically to all channels

• Uses functional.center_crop() implementation for both numpy and torch

__init__(height: int, width: int) → None

Methods

__init__(height, width)