Skip to content

Repository files navigation

🏆 Scoreboard Analyzer AI

🎯 Problem Addressed

Addressing the lack of visibility and limited coverage for lesser-known sports leagues and women's sports. Gen Z demands shorter, engaging, and easily shareable content.

✨ Our Solution

  • Automatically processes video footage using OpenCV and PyTesseract OCR to accurately detect and track scoreboard changes.
  • Utilizes NLP (NeMo, NVIDIA) to analyze commentary and rank clips by excitement level.
  • Enhances video production efficiency using parallel processing with FFmpeg.

🚀 Features

  • Real-Time Score Tracking: Precisely extracts scoreboard data from diverse footage.
  • Commentary Analysis: Identifies exciting moments using NLP and ranks content for maximum viewer engagement.
  • Cost & Time Efficient: Saves up to 14 hours per project by automating the editing process.

⚙️ Technology Stack

  • OpenCV
  • PyTesseract OCR
  • NLP (NeMo, NVIDIA)
  • FFmpeg
  • FAISS Vector Database

🔮 Future Enhancements

  • Incorporate pose detection and object tracking for advanced analytics.
  • Expand event recognition to include assists, rebounds, steals, and defensive actions.
  • Broaden training dataset for improved accuracy and reduced misclassification.

🙏 Acknowledgments

Special thanks to our hackathon team and mentors who supported this project!

About

Uses machine learning to automatically clip the moments in a game and shortens it to a specified time

Resources

Stars

Watchers

Forks

Releases

Packages

Used by

Contributors

Languages