UC Davis

This paper studies artistic expression in human movement by exploring the performance art form salsa. The motions of a salsa performance are constructed as concatenations of motion primitives, each of which specifies the movement of the dance pair over the course of eight musical beats. To analyze the syntax of artistic expression, the choreography of dance performances is represented by a transition model that is based on humanoid robot representations of the dancers. In order to assess the quality of a performance, two distinct metrics are explored. By integrating the performance metrics into the proposed transition system, it is possible to create an algorithm that is capable of autonomously recognizing the dance moves and evaluating the quality of the performance with a score. To validate the model, a dance pair performed four distinct salsa dance sequences observed by an artificially intelligent (AI) judge. The video recordings of the performances are also shown to a dance audience for evaluation. By looking at the correlation between the dance audience and the AI judge's scores, we conclude that the proposed model performs well in evaluating the artistic merit of the dance.