Validation of a low-cost swimming competition analysis system using the panning technique

Abstract

Background: The analysis of swimming competition provides relevant data for the performance of swimmers, however, the techniques used require expensive instruments. The appearance of economic alternatives such as swimmer tracking techniques, known as panning, are an alternative for the democratization of competition analysis.

Goals: To validate a panning competition analysis system in comparison with the traditional system of fixed cameras.

Method: 10 swimmers were analyzed during a 25-meter swim using a fixed camera system and a GoPro camera that made the measurement by following the swimmer. Previously, the lane ropes were marked every 5 meters so that they were visible to the cameras. The recordings were made simultaneously. Observer repeatability was tested, as well as cross-system validation.

Results: The intra-subject repeatability indicated that the analyzed observer presented a high degree of repeatability (error < 2%). The reliability study between systems (Fixed vs. Panning cameras) yielded extremely high levels of Pearson's correlation and Lin's concordance, as well as an excellent intraclass correlation coefficient. Likewise, no significant differences were observed between systems. Finally, the differences between the two systems using Bland-Altman plots indicated that all observations were distributed between the 95% percentile confidence intervals, indicating a high agreement between systems.

Conclusions: The panning system of competition analysis is valid with respect to the fixed camera system at the training level.