Tracking transients in steelpan strikes using surveillance technology.

This paper presents advancements in tracking features in high-speed videos of Caribbean steelpans illuminated by electronic speckle pattern interferometry, made possible by incorporating robust computer vision libraries for object detection and image segmentation, and cleaning of the training dataset. Besides increasing the accuracy of fringe counts by 10% or more compared to previous work, this paper introduces a segmentation-regression map for the entire drum surface yielding interference fringe counts comparable to those obtained via object detection. Once trained, this model can count fringes for musical instruments not part of the training set, including those with non-elliptical antinode shapes.