The OpenBiomechanics Project (OBP) provides anonymized, elite-level athletic motion capture data, addressing the need for accessible, high-quality biomechanics datasets. It offers raw (C3D) and processed full-signal data, along with point-of-interest metrics, targeting researchers, engineers, and power users in sports science. The project aims to democratize access to elite athletic movement data for analysis and development.

How It Works

The project distributes large data archives, including raw C3D files and processed full-signal data, via GitHub Releases to maintain a lean git repository. Users fetch these assets using a provided script (scripts/download_data.sh). In-repo CSV files for point-of-interest (POI) metrics and metadata are accessible directly. A Python helper package (obp/) simplifies loading these CSVs with pandas, and runnable examples demonstrate data access and analysis workflows.

Quick Start & Requirements

• Primary install/run command: Execute scripts/download_data.sh to fetch release assets. Use the obp/ package (ensure it's in sys.path) for loading data with pandas.
• Prerequisites: Python, pandas, ezc3d (for C3D file reading examples).
• Links: Project homepage (mentioned, not linked), documentation (README), computer-vision demo media (via GitHub Releases), Mokka installer (via GitHub Releases).

Highlighted Details

• Modules: Includes dedicated sections for Baseball Pitching, Baseball Hitting, High Performance, and Computer Vision (OBP-CV) data and examples.
• Data: Offers raw C3D, processed full-signal archives, and point-of-interest (POI) kinematics/kinetics.
• Metadata: Features machine-readable data dictionaries, a DATASHEET.md following the Datasheets for Datasets framework, and a CITATION.cff file.
• Tooling: Provides a pandas loader package (obp/) and runnable example scripts for data exploration and analysis.

Maintenance & Community

The repository underwent a significant overhaul in July 2024, rewriting git history to remove large binaries and reduce repository size. Updates include the addition of the High Performance module (July 2024) and the OBP-CV module (November 2023). The project is run by Driveline Baseball.

Licensing & Compatibility

The project employs a dual-license strategy: code is licensed under MIT, while data and biomechanics documentation are under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0). The data license is strictly non-commercial and includes a specific exclusion: employees, contractors, or significant shareholders of professional sports organizations or financial analysis firms are forbidden from using the data for any purpose.

Limitations & Caveats

The recent git history rewrite necessitates re-cloning for existing users. The CC BY-NC-SA 4.0 license, particularly its non-commercial clause and the explicit exclusion for professional sports organizations and financial firms, severely restricts potential commercial applications and use by entities within those sectors.