Olympic weightlifters' performance assessment module using computer vision

Abstract

Olympic weightlifting is a highly technical and physically demanding sport that requires athletes to lift heavy weights with precise technique. Accurately assessing an athlete's performance is crucial for identifying areas of improvement in terms of strength, power output, and movement efficiency. Traditional methods ofperformance assessment, such as manual observation and analysis, are time-consuming and subject to human error. Advances in wearable sensor technology, motion capture systems and computer vision algorithms have the potential to revolutionize sports analytics. We report a computer vision pipeline for automated analysis of weightlifting videos and demonstrate the markerless estimation of major joints, joint angles and accurate tracking of barbell trajectory allowing us to superimpose the estimated CoG trajectory, the base of support, and the line of gravity information into the video. The augmented videos can provide instant, objective and detailed feedback and allow longitudinal insights for comprehensive weightlifting performance analysis.