Decoding Directorial Style | Using Pose & Action Estimation to Analyze Theater Performances

Abstract:

At this seminar, hear how this research project employs machine learning and computer vision to analyze directorial styles and other aspects of theater performances, focusing on pose and action recognition. By applying these technologies to video recordings of theatrical productions, multiple performances by the same director are compared to identify distinctive patterns in choreography and staging. The approach combines both distant and close viewing techniques, enabling a more nuanced understanding of theatrical gestures and movements. Through this analysis, the abstract concept of directorial style is quantified.

Bridging the fields of performing arts, computer science, and digital humanities, this interdisciplinary project offers new insights into theatrical analysis and enhances the understanding of directorial signatures in live performance.