Athletic posture optimisation

Abstract

Injury prevention is a common area which athletes aim to work on the most throughout their careers. Sustaining an injury could mean that an athlete would have to sit out of competitions and games for a lengthy period of time, often affecting their mindset and performance in a negative way. This project is motivated by the need to aid athletes and other individuals to prevent injuries that result from incorrect posture during exercise. Attempting to avoid unnecessary strains on the individual’s joints and muscles would allow the individual to work out in a healthier and harmless way on the road to achieving their goals. To achieve this, a real‐time application is developed where the user would be able to receive constant, visual feedback regarding their posture during exercise. The squatting exercise was used as a proof of concept, paving the way for a wider variety of exercises to be included eventually. In this study, a dataset of images is collected and different classification algorithms are investigated to be able to pick out the most ac‐ curate and timely ones. When combined together, these algorithms produce a solution which is able to provide feedback to the individual in a real‐time, precise manner. The implementation continued to build on several works that have been done around the same area. When comparing the final implementation to other works, the timeliness and accuracy at which the application can generate results to the user sets it a step higher than most other implementations.