Coach Cahtah

Coach Cahtah is an AI-powered basketball coach used to help athletes identify any mistakes in their shooting form.

Comment

Inspiration

Our inspiration came from a shared interest in sports and the unfortunate realization that-- getting better was getting harder! Private Coaching was getting more expensive, and self-coaching was difficult and time-consuming. To learn how to shoot with proper basketball shooting form, an athlete would have to be willing to pay inflated prices for lessons or be left to record themselves, play, analyze the film, and scour the internet to get better. We realized that we needed a coach; we needed an AI-powered coach to make our jobs as athletes much more efficient. Inspired by the hit film Coach Carter (2008) and the Bostonian accent, we developed "Coach Cahtah"!

What it does

Coach Cahtah takes a video input of a player shooting a basketball anywhere on the court and uses Machine Learning to track and trace the ball's path to the hoop so that the user can better understand how their shooting form affects their field goal percentage.

How we built it

We used Python for the backend of the project and the public Machine Learning library, OpenCV, to track the path of the basketball in the video. Instead of using hours of nonexistent basketball training data, we used a color-mapping method on the basketball to isolate its motion and cut our time to production exponentially.

Challenges we ran into

First off, we needed to realize how difficult image recognition was. We did not have enough workforce or existing data to fulfill our intended goals, so instead, we simplified our ambitions to reach an MVP. We also attempted to use the Terra API with the Apple Watch, but due to its closed-system nature, it wasn't easy to pull gyroscopic data from the device.

Accomplishments that we're proud of

Fortunately enough, we were able to adapt very well considering the situation. Instead of using TensorFlow, we ended up using OpenCV, which was more beginner-friendly, and we were able to circumvent the hours of training data we needed to train a model by taking a shortcut through the whole process. We used color mapping to isolate the basketball to trace and view its path to the hoop instead.

What we learned

Machine learning has INCREDIBLE opportunities to make our daily lives INCREDIBLY efficient. The problem is that it requires a strong understanding and high-quality training data. With time and effort, using this tool can create unique products.

What's next for Coach Cahtah

Coach Cahtah's greatest asset is its expandability. We can integrate all of these programs into a single app to allow athletes to analyze their shooting on the court efficiently. We can also move into other sports like Tennis.

Built With

Share this project:

Updates