Weixuan Lei1 and Yu Sun2, 1USA, 2California State Polytechnic University, USA
Training for fencing during the pandemic has changed from what it was beforehand. Students have been taking lessons online, and instead of fencing with peers, students now train by watching and analyzing fencing videos. Learning fencing from watching videos of world class fencers is an effective way of learning. However, sabre fencing is so fast that many inexperienced fencers are unable to capture the important information by watching short clips. Therefore, they are unable to learn techniques such as sabre fencing just from watching videos. This paper traces the development of an application that can utilize computer vision and pose estimation to analyze fencing video clips and output accurate scored points as well as the techniques used within the given clips. We applied our application to help less experienced fencers improve their ability to recognize points and conduct qualitative evaluations of different fencing techniques.
3D simulation, Computer vision, Artificial intelligence, Fencing lessons.