By leveraging human intention as a high-level guidance, the proposed framework can effectively anticipate long-term sequences of future human actions in egocentric videos.
