Air Drums: Playing Drums Using Computer Vision

Abstract

The cost of a drum set is an investment that most aspiring drummers would eventually need to shoulder in order to continue their craft. What this research aims to do is to hasten the introduction of drummers to the drumming experience without the costs, and also to allow for drummers to be able to practice, at least casually, without a full drum set. This thus allows the experience of drumming to a wider audience. The solution we explore is the development of a prototype virtual drum set that would only require users to have a laptop with a camera, along with easily accessible markers representing the tips of drum sticks and knee movement, such as colored papers. OpenCV based on Python was used to implement this, and used the concept of color-based blob detection for detecting the markers. The contributions of this work are the following: (1) the built prototype system is the first attempt as far as we know in building a purely camera-based air drum system that includes the feet movement for bass drum control; (2) the user interface of the system gives access to a more comprehensive set of drum components with the exclusion of the hi-hat due to its complicated dynamic properties; and (3) a more flexible system was built since it allows the usage of markers with different colors, shapes, and sizes. This prototype has also shown to have potential for further development as a released application, along with the possibility of use as a USB controller and MIDI controller.