dgs.utils.image.CustomResize¶

class dgs.utils.image.CustomResize(*args: Any, **kwargs: Any)[source]¶

Resize image, bbox and key points with this custom transform.

The image and bbox are resized using regular torch resize transforms.

Methods

forward(*args, **kwargs) → dict[str, any][source]¶

Resize image, bbox and key points in one go.

Keyword Arguments:

image – One single image as tv_tensors.Image of shape [B x C x H x W]
box – tv_tensors.BoundingBoxes in XYWH box_format of shape [N x 4], with N detections.
keypoints – The joint coordinates in global frame as [N x J x 2|3]
output_size – (h, w) as target height and width of the image

Returns:

Will overwrite the image, bbox, and key points with the newly computed values. Key Points will be in local image coordinates.

The new shape of the images is [B x C x h x w].

Attributes