Predicting skeleton joints on egocentric video using exocentric views.
Recognizing and classifying robotic manipulation actions from video.