Pose with object detection

We want to detect object as well do pose estimation as part of one model. Could someone please suggest what would be the new model architecture shoould be?