Authors:
Lisa Baily
1
;
Nghia Truong
2
;
Jonathan Lai
3
and
Phong Nguyen
2
Affiliations:
1
The American School in Japan, 1-1-1 Nomizu, Chofu-shi, Tokyo, Japan
;
2
Tokyo Techies, Shinbashi 2-16-1, Minato, Tokyo, Japan
;
3
Tokyo Coding Club, Nishi Azabu 3-24-16, Minato, Tokyo, Japan
Keyword(s):
Pose Estimation, Pose Tracking, Machine Learning, Computer Vision, Euclidean Distance, Tennis, Analysis.
Abstract:
In this paper, we created a method to find how professional and amateur tennis serves differ from each other. We collected videos from online and from our own recordings and turned those videos into frames. From those frames, we manually selected ones appropriate for our study and ran those through a pose estimation system, which turned those frames into simple stick figures of the players including all the x and y coordinates of the player. By normalizing all data, we were able to calculate the Euclidean distance between two compared players’ joints and analyze their consistency in their serves. Our results from our t-tests showed that there was a significant difference between the amateur’s consistency and the pro’s consistency and body parts like both shoulders showed a significant difference.