Vis enkel innførsel

dc.contributor.advisorHendseth, Sverre
dc.contributor.authorEgeland, Tønnes
dc.date.accessioned2024-05-05T17:19:38Z
dc.date.available2024-05-05T17:19:38Z
dc.date.issued2024
dc.identifierno.ntnu:inspera:166402102:93065660
dc.identifier.urihttps://hdl.handle.net/11250/3129102
dc.descriptionFull text not available
dc.description.abstract
dc.description.abstractIn this project, a tool for tracking crossfit workouts has been developed. The tool aims to produce performance-based statistics for coaching analysis, it also aims to act as a digital judge with documents verified results which then may be used to qualify for competitions, and may also be used to generate data which may produce statistical data which may be displayed on-screen during broadcasted competitions. The project was executed using the Python modules OpenCV and Mediapipe. With these modules, the project achieved a sucient level of reliability to prove the concept of the pro- gram. The project’s end-state lacked tuning, indicated by varying results during testing, but showed that the program works and that the di↵erent modules interact and largely do as intended. The program included a module to detect CrossFit equipment, where the detection of barbells was implemented. The barbell detection experimented with using deep learning but ended up using color markers and detection, which produced great results. The project shows great potential for further development and could if continued one day be a common tool in the CrossFit community if completed and made accessible.
dc.languageeng
dc.publisherNTNU
dc.titleUsing computer vision to track progress through a CrossFit workout
dc.typeMaster thesis


Tilhørende fil(er)

FilerStørrelseFormatVis

Denne innførselen finnes i følgende samling(er)

Vis enkel innførsel