INTRODUCTION
Learning to play the guitar usually involves tedious
lessons in the correct ¯ngering positions for the left
hand (assuming the guitarist is right-handed). It is
sometimes di±cult for beginners to recognize whether
they are accurately positioning their ¯ngers on the
strings to make the correct guitar chords.
In this paper, we propose a guitar teaching aid that
will assist guitar players by using computer vision and
pattern recognition technologies. This system detects
the players' ¯ngers in real time and determines the
guitar chord which the player's left hand is using.
To calculate the positions of ¯ngers, colored ¯nger
markers are detected using a Bayesian classi¯er that
is bootstrapped with a small set of training data and
re¯ned through an o®-line iterative training proce-
dure [1] [2]. By using an on-line adaptation of ¯nger