University Honors

This project develops a robust chess vision system capable of detecting and tracking agame of chess under a variety of lighting conditions and angles consistent with the abilities of ahuman. It includes a graphic that displays the current state of the chess game, the last movemade, and the best move to make. It works using a single camera and a continuous video streamand uses machine learning for piece recognition. It is able to adapt to different viewing angles,changes in lighting, obstructions on the board, or unexpected adjustments or movements ofpieces. However, unlike a human, it runs every move it detects through a state-of-the-art chessengine to determine the best possible move to make in response. This system can beimplemented in a consumer smartphone app or be used during professional chess competitions asa much more economical alternative to electronic chess computer boards.