Keywords: eBoard Games | Computer Vision | eSports
The Nvision project was completed at the University of Sydney in 2013 along with my phenomenal team of Sue Ann Wong and Lloyd Windrim. It was part of a computer vision challenge in which we were tasked with applying CV techniques to a real-world situation with commercial potential. Our idea was a system that could spice up complex board games by adding a virtual referee into the mix. Our virtual referee ‘enVision’ would ensure players were following the rules of play and that no errors were made. We saw this being especially useful for competitive board game tournaments or even making the board game medium more user-friendly to new players.
The aim of this project was to create a real-time, unsupervised referee system for a Monopoly board game using computer vision techniques. There have been many attempts made to automate different elements of games, for example, creating an artificially intelligent opponent which learns from experience (Spronck et al., 2003) or creating a system which augments aspects of a sporting match to enhance the viewing experience (Gedikli et al., 2007). The system we developed behaved as a referee/supervisor for a Monopoly board game by producing three different warnings when errors occur during gameplay.
The idea was that through the automation of error checking, players will be more relaxed as they take comfort in the idea that a computer is making sure that no mistakes are made during the game. The system used an overhead camera which takes a photo of the scene containing the board after each turn is completed. Processing occurs on this photo to determine which errors are occurring.
The system was comprised of four primary modules- each one reliant on computer vision and image processing. Each of these modules has been designed and evaluated in a self-contained manner, although they all interact in the final system. The modules we developed include the (1) Game Board Setup (2) Dice Roll Detection (3) Token Movement Verification and (4) Token Tracking. Combined, these subsystems could be used to monitor and track a vast range of board games.