Specifically, we contribute Rolling-Unrolling LSTM, a learning architecture to anticipate actions from egocentric videos.
確定! 回上一頁