Unlocking the potential of video-based markerless motion analysis to study world-class sporting performance

The ability to access and assess technique of world-class athletes in-competition has long eluded sport biomechanists due to the constraints of 3-dimensional marker-based motion capture systems. Encouragingly, the advent of video-based markerless motion analysis offers the opportunity for unobtrusive capture of sporting performances in-competition. This narrative review explores the progression of markerless systems from laboratory-like to in-competition analysis. It first explores the accuracy of markerless motion analysis for joint kinematics in dynamic and sporting movements. Next, its utility in competition is explored with ways to validate these in-competition set-ups. Existing in-competition set-ups offer a viable foundation for skeletal tracking but still requires rigorous validation. Tennis is proposed as an example where the current in-competition infrastructure may support high-fidelity motion analysis. On top of the advanced hardware (i.e. camera configurations) that has been heavily invested in, software optimization methods proposed in computer vision research (i.e. enhancing data processing efficiency, expanding and refining training datasets with anatomically-accurate labelling and sport-specific data, and increasing the quantity and types of keypoints detected), can potentially improve its accuracy for in-competition analysis. Ultimately, a markerless system rigorously validated under real-world constraints can bridge the gap between accessing and assessing world-class sporting performances for sport biomechanics.