Technology is having an increasingly pervasive role in many sports disciplines, allowing athletes and trainers to exploit the possibilities opened by advanced sensing and intelligent data analytics. For example, heart rate monitoring vests, in- field positioning monitoring, goal line, and video-referee are currently common practice in football. In the case of tennis, rugby, or volleyball, the "hawk eye" is also a widely used technology also.
Within this landscape, swimmer analysis is somewhat of an underdeveloped area, where the typical approaches consist of video-based systems (which are expensive to setup and fixed to a particular location) and wearable devices still present limitations in terms of functionality and level of detail of the extracted information.
To address some of these challenges, researchers from IT in Lisbon recently developed SwimIT, a novel approach composed of hardware and software based on attitude and heading reference system (AHRS) and a machine learning workflow for data analytics. This is intended to help swimmers and coaches in performance evaluation and improvement, through the extraction of an extended set of dynamic parameters.
In his thesis, Eduardo Félix, explored new methodologies that have shown promising results, with a 100% accuracy in swam laps segmentation, a precision of 100% in the recognition of backstroke style and a precision of 89.60% in the three remaining swimming techniques (butterfly, breaststroke and front crawl).
Real swimming data was collected from swimmers with different skill levels, in order to develop a representative database and allow assessing the system performance in real-world swimming conditions. From the AHRS data, three novel indicators were also introduced, namely, trunk elevation, body balance and body rotation.