Function random_crop

slowfast/datasets/transform.py:117–150

Perform random spatial crop on the given images and corresponding boxes. Args: images (tensor): images to perform random crop. The dimension is `num frames` x `channel` x `height` x `width`. size (int): the size of height and width to crop on the image. b

(images, size, boxes=None)

def random_crop(images, size, boxes=None):
    """
    Perform random spatial crop on the given images and corresponding boxes.
    Args:
        images (tensor): images to perform random crop. The dimension is
            `num frames` x `channel` x `height` x `width`.
        size (int): the size of height and width to crop on the image.
        boxes (ndarray or None): optional. Corresponding boxes to images.
            Dimension is `num boxes` x 4.
    Returns:
        cropped (tensor): cropped images with dimension of
            `num frames` x `channel` x `size` x `size`.
        cropped_boxes (ndarray or None): the cropped boxes with dimension of
            `num boxes` x 4.
    """
    if images.shape[2] == size and images.shape[3] == size:
        return images
    height = images.shape[2]
    width = images.shape[3]
    y_offset = 0
    if height > size:
        y_offset = int(np.random.randint(0, height - size))
    x_offset = 0
    if width > size:
        x_offset = int(np.random.randint(0, width - size))
    cropped = images[
        :, :, y_offset : y_offset + size, x_offset : x_offset + size
    ]

    cropped_boxes = (
        crop_boxes(boxes, x_offset, y_offset) if boxes is not None else None
    )

    return cropped, cropped_boxes


def horizontal_flip(prob, images, boxes=None):