Practice Makes Perfect: An Analytical Approach to Eliminating Judging Bias in Springboard Diving

Abstract

The judging system in springboard diving is fundamentally flawed due to the influence of subjectivity and bias. My senior thesis project attempts to mitigate these flaws by creating a computer program that uses analytical methods to score dives. Using a combination of computer vision and machine learning algorithms, I have created a system that can pass a binary judgment of a dive (success/failure) and gather metrics that can ultimately help determine a raw score. These metrics provide informationabout various components of the dive: height, distance from the diving board, angle of takeoff, angle of entry, and size of splash. Though it is not perfect, the pipeline I have created is a relatively successful preliminary method of computerized judging. The results indicate that with more data and fine-tuning of the methodology, the ultimate goal of removing human error from judging is feasible.