JATICDetectionAugmentation

class nrtk.interop.maite.interop.object_detection.augmentation.JATICDetectionAugmentation(augment: PerturbImage, augment_id: str)

Implementation of JATIC Augmentation for NRTK perturbers.

Implementation of JATIC Augmentation for NRTK perturbers operating on a MAITE-protocol compliant Object Detection dataset.

Given a set of ground truth labels alongside an image and image metadata, JATICDetectionAugmentation will properly scale the labels in accordance with the change in the image scale due to applied pertubation. At this time JATICDetectionAugmentation does not support any other augmentation to the labels such as cropping, translation, or rotation.

Parameters

augmentPerturbImage

Augmentations to apply to an image.

metadata: AugmentationMetadata

Metadata for this augmentation.

Methods

__call__(batch: tuple[Sequence[ArrayLike], Sequence[ObjectDetectionTarget], Sequence[DatumMetadata]]) → tuple[Sequence[ArrayLike], Sequence[ObjectDetectionTarget], Sequence[NRTKDatumMetadata]]

Apply augmentations to the given data batch.

__init__(augment: PerturbImage, augment_id: str) → None

Initialize augmentation wrapper

__subclasshook__()

Abstract classes can override this to customize issubclass().

This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).