Is there a Mask-Keypoint R-CNN available?

In torchvision we got keypoint R-CNN to perform object detection as well as keypoint detection; and we got mask R-CNN to perform object detection as well as instance segmentation. Say, if for an detected object, I want to perform both keypoint detection and instance segmentation, is there a “mask-keypoint R-CNN” for this?

Wrote one myself here. Yet may not be very stable.