Iannucci, ElenaChen, ZhutianArmeni, IroPollefeys, MarcPfister, HanspeterBeyer, JohannaHoellt, ThomasAigner, WolfgangWang, Bei2023-06-102023-06-102023978-3-03868-219-6https://doi.org/10.2312/evs.20231046https://diglib.eg.org:443/handle/10.2312/evs20231046Rowing requires physical strength and endurance in athletes as well as a precise rowing technique. The ideal rowing stroke is based on biomechanical principles and typically takes years to master. Except for time-consuming video analysis after practice, coaches currently have no means to quantitatively analyze a rower's stroke sequence and body movement. We propose ARrow, an AR application for coaches and athletes that provides real-time and situated feedback on a rower's body position and stroke. We use computer vision techniques to extract the rower's 3D skeleton and to detect the rower's stroke cycle. ARrow provides visual feedback on three levels: Tracking of basic performance metrics over time, visual feedback and guidance on a rower's stroke sequence, and a rowing ghost view that helps synchronize the body movement of two rowers. We developed ARrow in close colaboration with international rowing coaches and demonstrate its usefulness in a user study with athletes and coaches.Attribution 4.0 International LicenseCCS Concepts: Computing methodologies -> Mixed / augmented reality; Human-centered computing -> VisualizationComputing methodologiesMixed / augmented realityHuman centered computingVisualization10.2312/evs.2023104673-775 pages