Joseph

Meet Joseph

Joseph's keen interest in computer programming, artificial intelligence, and basketball converged to inspire this innovative project idea.

He had a desire to create an AI-based tool to help basketball coaches identify plays and track a team's use of certain plays. This data collection will help coaches develop a deeper understanding of their opponents and get a better sense of what to expect when competing against certain teams.

8 Questions with Joseph

Describe your innovation (what does it do?):

I developed an algorithm that tracks and traces the movement of basketball players from a video and plots the movement on a half-court map to help the coach analyze the player's movement on the court.

Describe your development process (what did you learn during this process?)

Technically, I learned about using OpenCV and YOLO for object recognition and some skills regarding mapping and projecting. Regarding the world of computer science, I realized that the world of technology is constantly changing, and we always need to stay up-to-date with emerging information and to continuously adapt and learn new skills.

What challenges did you overcome during development?

During the project, I was unable to recognize the players using OpenCV correctly; therefore, I did some research and switched to Yolo, an object recognition algorithm using convolution neural networks. After switching, everything works well, and most human characters in the frame can be successfully recognized by the algorithm.

What challenges still exist with your innovation?

Firstly, colour recognition is still an existing problem. The algorithm is unable to recognize the referees on the court and always recognizes them as one of the players. This is due to the fact that team recognition is based on the colour recognition of the character’s jersey, and they have both black and white on their jersey.

Secondly, I was unable to connect all the dots on the map into a line. Currently, it looks like a scatter plot on the map and it is really confusing.

Where do you hope to go with this innovation next?

I wish to complete the algorithm for colour recognition and get in touch with some experts in the field to help me solve the problem of connecting the player’s movement into a complete line instead of scatter plots.

What / who at GNS helped you along the way?

Mr. Hamlin, as my personal project coordinator, guided me throughout the whole project. I plan to keep developing my project in Launch Pad this year.

Page updated

Report abuse