HAAR Classification for Evaluation of Dance Performance - Bharatanatyam

Abstract

Automated evaluation of dance performances against a gold standard is implemented in this project and a scorecard is provided to the performer in the form of a graph with improvement suggestions. Joint detection is used by the system to acquire the motion of the performer. The dancer's frontal view is captured by a camera. To separate the video into individual frames, it is input into the system. As the HAAR classification system detects eight essential joints of the body in each frame, the position changes of these joints are used to calculate the joint velocity. A dancer's performance is evaluated in contrast to a gold standard using a scoring algorithm. Upper- and lower-body movements are scored independently. As a result of the upper and lower body scores, a final score is given that incorporates both scores. In addition to providing a score, the algorithm also attempts to pinpoint areas where the dancer can improve. This system attempts to automate the complete evaluation process of a dancer during examinations in order to reduce the amount of time and effort required by humans.