Weakly supervised methods for learning actions and objects: Reducing human intervention in learning visual concepts for Artificial Intelligence