We present a fully automatic arm and hand tracker that detects joint positions over continuous sign language video sequences of more than an hour in length. Our framework replicates the state-of-the