Crop

class deeptrack.augmentations.Crop(*args, crop: int | list[int] | tuple[int] | Callable[[Image], tuple[int]] = (64, 64), crop_mode: PropertyLike[str] = 'retain', corner: PropertyLike[str] = 'random', **kwargs)

Bases: Augmentation

Crops a regions of an image.

Parameters

feature: Feature or list of Features

Feature(s) to augment.

crop: int or tuple of ints or list of ints or Callable[Image]->tuple of ints

Number of pixels to remove or retain (depending in crop_mode) If a tuple or list, it is assumed to be per axis. Can also be a function that returns any of the other types.

crop_mode: str {“retain”, “remove”}

How the crop argument is interpreted. If “remove”, then crop denotes the amount to crop from the edges. If “retain”, crop denotes the size of the output.

corner: tuple of ints or Callable[Image]->tuple of ints or “random”

Top left corner of the cropped region. Can be a tuple of ints, a function that returns a tuple of ints or the string random. If corner is placed so that the cropping cannot be performed, the modulo of the corner with the allowed region is used.

Methods

get(image: Image | np.ndarray, corner: PropertyLike[str], crop: PropertyLike[int], crop_mode: PropertyLike[str], **kwargs) -> Image

Abstract method which performs the Crop augmentation.

Methods Summary

get(image, corner, crop, crop_mode, **kwargs)

Abstract method which performs the Crop augmentation.

Methods Documentation

get(image: Image | np.ndarray, corner: PropertyLike[str], crop: int | list[int] | tuple[int], crop_mode: PropertyLike[str], **kwargs) → Image

Abstract method which performs the Crop augmentation.