A Framework for Analysis of Softball Pitching, as Applied to Legal and Illegal Pitches

Abstract

In response to the NCAA’s 2023 rule change allowing softball pitchers to legally disengage from the playing surface while delivering a pitch, this study develops a framework to analyze and compare the legal drag, legal leap, and illegal replant pitching techniques. By developing a pose estimation algorithm and Recurrent Neural Network (RNN) for use on videos of real collegiate pitchers, we aim to distinguish physiological differences between these types of pitches and use our RNN to automatically detect illegal pitches. Our pose estimation results demonstrate the algorithm's effectiveness in extracting patterns from pitching videos. Key features such as the distance between the pitcher’s right knee and right toe, as well as the right toe x-position vs. time, emerge as crucial indicators for distinguishing legal and illegal pitches. The RNN achieved an accuracy of 71.4%, with a loss rate of 0.875. This framework offers a data-driven approach to softball pitching mechanics, providing valuable insights for researchers and coaches alike.