[1G3] VR Applications 1

[1G3-01] Presentation Training System Based On Imitating Past Famous Speech

○Yuhua Shi¹, Kohei Matsumura¹, Roberto Lopez-Gulliver¹, Haruo Noma¹ (1. Ritsumeikan University)

Conference presentation is a non-trivial task since it is affected by both presentation content and nonverbal behaviors. To improve presentation ability, we propose a new training method imitating famous past speeches. We employed the OpenPose Library to extract orators' motion data from past famous speech 2D video. During training, the system captures trainee's motion in real-time using the Kinect, then calculate the cosine similarity of adjacent limbs to get the similarity of the trainees' and orator's motion data. In this paper, we evaluate the effectiveness of our motion data matching algorithm. Experiment results show that the proposed algorithm is able to match the orator's motion in the past famous speech and trainee's motion. In the future, we plan to allow trainees to imitate the real past famous speech and evaluate the improvement of the trainees' presentation skills using our method.