This paper investigates the impact of contemporary entertainment preferences among youth, who increasingly favor novel forms over traditional theater arts, driven by the rapid pace of modern life. Specifically, it explores how action recognition technology can transform theatrical performance forms through the lens of intelligent media. Utilizing the inertial sensor MPU9250, the study mathematically captures the action characteristics of theatrical performances using quaternion algebra. This process involves preprocessing steps that set the stage for coordinate system transformations. Furthermore, the research constructs a database of theatrical morphology by classifying and storing morphological features derived from a human 3D skeleton dataset. The effectiveness of motion recognition technology in enhancing theatrical performances is then empirically tested and analyzed. The results reveal that this technology successfully identifies morphological movements in traditional theater with accuracies of 74.42%, 73.77%, and 64.45%, respectively. An additional pedagogical assessment highlights the significant role of motion recognition technology in educational contexts. In particular, the study notes a substantial difference in the ‘imagination’ dimension scores between control and experimental groups. The control group scored an average of 6.67, whereas the experimental group achieved 8.07, indicating enhanced engagement and cognitive stimulation through the application of motion recognition. These findings underscore the potential of such technology to offer theoretical and practical insights for the revitalization and progressive development of traditional theatrical performance forms.