Kinetic predictors of spectators' segmentation of a live dance performance

International audienceWe present a pilot study that explores the relation between acceleration paaerns of dance movements and the temporal segmentation of the dance reported by spectators during a live performance. Our data set consists of recorded accelerations from two 7 minutes long duo dances that were annotated by 12 spectators in real-time. e annotations were indications of perceived starts and endings in the dance. We were able to create an acceleration based predictor that has a signiicant correlation with the pooled subjective annotations. Our approach can be useful in analysis of improvised dance where the segmentation cannot rely on repetitive paaerns of steps. We also present suggestions for future development of acceleration based dance analysis