Add support for TrainingSlicer

[nicholasgcoles]

Search before asking

• [x] I have searched the Supervision issues and found no similar feature requests.

Description

I am happy to give this a shot, but wanted to make sure this would be useful.

For small object detection, I think it would be useful to slice training images. It would need a slightly different interface to InferenceSlicer.

Something like:

class TrainingSlicer:

    def __init__(
        self,
        callback: Callable[[np.ndarray, Detections], list[tuple[np.ndarray, Detections]]],
        slice_wh: tuple[int, int] = (320, 320),
        ....
    ):

    def __call__(self, image: np.ndarray, detections: Detections) -> list[tuple[np.ndarray, Detections]]:
        pass

Use case

No response

Additional

No response

Are you willing to submit a PR?

• [x] Yes I'd like to help by submitting a PR!