This collaborative research project will focus on image recognition as a primary application of a Beowulf parallel processing system. The goal is to develop accurate and efficient image recognition algorithms so that Beowulf can use a video camera to monitor the state of an actual Mancala game board. This involves recognizing the position and orientation of the board, locating the pits and mancalas, and locating the individual stones in each pit or mancala. This project will provide the input to a separate project that will be developing the game playing algorithms. The long-term goal of these projects is to develop algorithms so that the Beowulf system can play Mancala against a human opponent, using a real board.