Abstract
With the development of human sensing technology and AI technology, much research of emotion recognition has been conducted. We have already proposed an emotion estimation method from natural body movement based on Laban Movement Analysis. In this study, we applied the proposed method to the scene of design creation where creative work is performed using PC. First, we extracted and analyzed the evoked emotions in the design creation using SONY FES Watch U based on the evaluation grid method, and constructed an emotion model. Next, using the constructed emotion model,we performed a measurement experiment of a dataset in which the body movement is associated with the emotion. By using this dataset, we classified three emotional expression types based on only the motion characteristics. We achieved about 87% estimation accuracy for individuals. Nevertheless, average estimation accuracy for emotional expression types was 18%.
