Abstract

Methods for creating a system to automate the collection of swimminganalytics on a pool-wide scale are considered in this paper. There has not beenmuch work on swimmer tracking or the creation of a swimmer database for machinelearning purposes. Consequently, methods for collecting swimmer data fromvideos of swim competitions are explored and analyzed. The result is a guide tothe creation of a comprehensive collection of swimming data suitable fortraining swimmer detection and tracking systems. With this database in place,systems can then be created to automate the collection of swimming analytics.

 

Quick Read (beta)