抄録

◎Zhao Wang・Jun Ohya（早大）

Fingering is one of the most important aspects in guitar learning. As the guitar is a string instrument, it is incomprehensive to check the fingering information of the player only through audio signal; visual information is also needed to the analysis. This paper proposes a method that can track guitar neck and guitarist fingertips respectively. Compared with related works, first and foremost, this paper’s method can cope with occlusion problems, which are the frequently happened issues during real guitar playing; second, it can relax some problems and inconvenience such us color mark at fingers and AR Tags. Moreover, the recognition accuracy can be highly improved