The current research has created a computer-assisted three dimensional teaching system based on the skinning algorithms and EWMA algorithms to recognize dance movements and quantify GL-Compound similarity, and thus has developed a computer-based innovative dance education framework. The experimental group that employed the combined 3D system was compared with the control group that received traditional teaching using dance students in a university as subjects. The results show that the experimental group scored significantly higher in technical quality (p=0.001), musical interpretation (p=0.000) and choreography (p=0.002). In particular, the experimental group obtained a mean technical quality score of 86.37 (control group: 79.04), a mean musical interpretation score of 89.46 (control group: 81.22), and a mean choreography score of 88.52 (control group: 80.73). The model, which uses multimodal data capture and smart analysis, offers a great tool in improving specific dance technical skills and can be used as a viable tool in the process of modernization of dance education.
