Google Cloud x MLB Hackathon Grand Prize Winner Pitcher Scorecard



AI Summary

Video Summary: Customizable Pitch Analysis Tool

  • Introduction: The speaker, a former shot putter, discusses the limitations of motion capture technology for analyzing sports techniques.

  • Concept: Developed a customizable prompt generator for analyzing pitches, using Gemini API and Google Cloud, completed in one week.

  • Target Audience: Designed for both professional and amateur pitchers.

  • Professional Analysis:

    • Uses video of MLB pitcher Clayton Kershaw.
    • Pre-processes video with OpenCV and stores in Google Cloud.
    • Pulls MLB data for pitch type and game state.
    • Generates tailored instructions and prompt details.
    • Example: Kershaw’s curveball performance analysis shows minimal deviation from ideal.
  • Amateur Analysis:

    • Includes parameters for less experienced players.
    • Co-host Richard’s pitch analyzed, showing need for improvement (tightening arm and increased leg drive).
    • Gemini adjusts its grading rubric based on skill level.
  • Technical Insight:

    • Gemini processes multiple frames simultaneously for a comprehensive analysis.
    • The application worked efficiently without the need for complex models or datasets.
  • Conclusion: This tool democratizes personalized coaching for athletes at any level, making advanced analysis accessible without high-resolution cameras.