A Non-Invasive Vision-Based Approach to Velocity Measurement of Skeleton Training

Research output: Contribution to conferencePaper


Skeleton is a winter sport where performance is greatly affected by the velocity an athlete can achieve during their start up to the point where they load themselves onto their sled. As such, it is of interest to athletes and coaching staff to be able to monitor the performance of their athletes and how they respond to different training schedules and techniques. This paper proposes a non-invasive vision based method for measuring the velocity of a skeleton athlete and their sled during the push start. Mean differences in estimated velocity between ground truth data and our proposed system were -0.005 (+/- 0.186) m/s for the athlete mass centre and -0.017 (+/- 0.133) m/s for the sled. The results compare favourably to techniques previously presented in the biomechanics and sport science literature.
Original languageEnglish
Publication statusPublished - 16 Jun 2020
EventIEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2020 - The Washington State Convention Center, Seattle, USA United States
Duration: 16 Jun 202018 Jun 2020


ConferenceIEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2020
Abbreviated titleCVPR 2020
CountryUSA United States
Internet address


Cite this