作者: R. Bencina , M. Kaltenbrunner
DOI:
关键词: Orientation (computer vision) 、 Evolutionary computation 、 Fiducial marker 、 Computer vision algorithms 、 Artificial intelligence 、 Computer vision 、 Software 、 Frame rate 、 Computer science
摘要: The reacTIVision system is software for tracking specially designed fiducials (markers) in a real-time video stream. ReacTIVision was to enable expressive gestural control of musical sound, and can track many markers at high frame rate. development involved not only computer vision algorithms, but also the design new marker system. We co-designed markers, applying evolutionary computation minimise size while meeting geometric constraints required efficiently compute location 2D orientation markers. techniques adopted would have been practical had genetic algorithms able solve these constraints. This paper focuses on evolution aspects are documented elsewhere.