Human-object-object-interaction affordance

This paper presents a novel human-object-object (HOO) interaction affordance learning approach that models the interaction motions between paired objects in a human-object-object way and use the motion models to improve the object recognition reliability. The innate interaction-affordance knowledge of the paired objects is modeled from a set of labeled training data that contains relative motions of the paired objects, humans actions, and object labels. The learned knowledge of the pair relationship is represented with a Bayesian Network and the trained network is used to improve recognition reliability of the objects.