Tensorflowjs Object Detection with Keypoints

Hi

Is it at all possible to add keypoints to a custom model for object detection in the browser?
I’m wanting to try and figure out the orientation of the object thats detected, and I think keypoints would help with this.