Detecting Short Distance Throwing Motions Using RGB-D Camera

抄録

Throwing is one of the most impotent activities in day-to-day life and for sports. This research focuses on developing a motion tracking system for detecting and analyzing short distance throwing motions and analyze the accuracy. Intel Realsense D415 camera was used as depth camera hardware. Nuitrack Software Development Kit combined with Open CV written in C++ was used to track skeleton position accurately. A simple throwing task is given to eleven volunteers to evaluate the accuracy of the system. The throwing task was to throw a tennis ball into a wastepaper bin by using the dominant hand. Distance of 1.5, 2, 3, 4 meters was given to each person to throw the ball accurately fifteen times per one distance. A tennis ball integrated with a touch sensor and a Bluetooth module was used to measure the ball's releasing point. For each throw, the shoulder angle and angular velocity of the ball releasing point were taken. The data collected and the data calculated by a mathematical model were compared to check whether the system is giving accurate data. Statistical analysis was done to measure the accuracy of the system.