UA Bball Wearable Project

Logo

CS495 Capstone Project

View the Project on GitHub wriemer/UA-Bball-Wearable-Project

Project Goals

    The original goal of this project was to create software for a pair of smart glasses which could analyze live basketball film for team personnel, ball location, and posession identification, and use this data to select and popup relevant stats on the glasses’ display. This would allow basketball coaching personnel to get access to live in-game statistics quicker, which would help with quicker decision making on the court. However, due to problems with the supplier of the smart glasses, we decided to pivot the project goal to focus on a web app leveraging the functionality described above.
    Our new goal was twofold. The first part of the goal was to create a software tool which could use computer vision to analyze basketball film and annotate it with relevant statistics. The second, and arguably more important goal was to build the foundation for the original goal of the project while also creating a potentially helpful tool. We specifically chose the path that we did because it would allow for the vast majority of our work to be built upon in the future. The three specific sub-goals we focused on can be seen below:

Approach

Approach

Tech Stack

Tool Purpose
Streamlit Build web app wrapper
Roboflow Create image dataset
YOLOv8 Train and implement CV model
OpenCV CV execution
SQLite Store roster info and stats

Technology

Documentation

    For documentation on the work which we have completed, please see the github link below. Read the relevant README.md documents for each module, and view the commented code to see our implementation. Project GitHub

Project Video (Presentation + Demo)

Team Members