pitch-ml$\textbf{: Optimizing Pitcher Health and Performance}$

Ball tracking, biomechanics, and machine learning algorithms for optimizing pitcher health and performance.

$\textbf{1. Repository Overview}$

• .bin: AWS connection configuration.
• dev: Any development tasks (e.g., biomechanics, data science), sorted by category and project.
• prod: Research- or production-ready code, usually adapted from dev.
• packages: Modules & functions that can be accessed within the repo. Note that these currently require running pip install -e . to access.
• qa: Debugging or revision-specific tasks (e.g., biomechanics, data science), sorted by categoory and project.

$\textbf{A.1 Environment Setup}$

All code is configured to be run in a conda virtual environment (details in pitch_ml.yml) from Python 3.11.10, which has all OpenSim API dependencies installed. To activate the environment, call conda activate pitch_ml.

$\textbf{A.2 AWS Connection}$

For AWS connections, make sure to run:

• (1) Ensure executable permissions:
  • chmod +x .bin/update_ip.sh
  • chmod +x .bin/tunnel.sh
• (2) Run scripts in terminal (from root directory):
  • ./.bin/update_ip.sh: Updates the EC2 IP address (if in a new connection)
  • ./.bin/tunnel.sh: Creates a secure shell (SSH) tunnel to the EC2 instance; this enables local connection to the RDS

More details can be found in the .bin folder.

$\textbf{A.3 Publications}$

• Moore, R.C., Gurchiek, R.D. & Avedesian, J.M. A context-enhanced deep learning approach to predict baseball pitch location from ball tracking release metrics. Sports Engineering 28, 16 (2025). https://doi.org/10.1007/s12283-025-00497-5
  • Relevant repository sections: Coming Soon.