Bocce Ball Computer Vision Technology

Howdy 👋

Last week, building upon Pearson's Law, I introduced you to Oddball's Tech Tuesday series of videos.

This week is story time. we're going to dive into my inspiration for an ambitious bocce scoring and stats system powered by cameras and software.

Welcome to the 6 new subscribers from last week. Thanks for spreading the word.

Bocce ball meets Computer Vision

Years ago circa 2010, I used to play bocce with my buddy Adam at music fests (Cozy Log Cabin, Pittsboro, NC), on the beach (Tybee Island, GA), and at camp sites (wherever). Back then, I never traveled without bocce balls in the trunk of my Volvo or Saab 🇸🇪 -- bocce is free fun in the sun ☀️.

I had never played bocce in a rectangular court. That is until my Chicago, IL neighbors, Kyle & Katie, introduced me to the American Bocce Co leagues.

One day in October 2020, Kyle and I were having a socially distant conversation on our roof decks and I told him about the technology idea I've always had for bocce. Back on the beaches of Tybee Island, the idea involved a drone that would zoom over to the balls down the beach and indicate with a light what color balls are "in" (i.e., closer to the pallino). Too impractical.

Given that real bocce the way the Italians play is supposed to be played on a court, the drone isn't really needed. Instead, mounting cameras to a ceiling, structure over a court, or pole on the side of the court makes more sense.

The concept is simple: Put cameras above a court, then use software to locate & distinguish balls and determine the distances to the pallino. Voilà — now you know who is IN/OUT throughout the frame and at the end of the frame.

Simple concept to explain, but difficult to execute, especially with a low budget and limited time.

Neighbor Kyle encouraged me to share the idea with ABC. So one day I pitched the idea to Matt David of ABC. This led to a conversation with Alex Gara and before you knew it, we were dreaming bigger than scoring as we knew that the value is in the statistics that such a system could produce.

I've been down a rabbit hole ever since.

Feeling inspired, in late 2020 I was writing code to model the game of bocce, and soon I was tasting the rainbow with a scale model of Skittles, a webcam, and working code on my desk.

This made me a bit too overconfident that the problem could be solved in a short amount of time.

A desktop Skittles experiment compared to the real world presents many challenges:

• Consistent lighting on my desk vs. uncontrollably lit rooms with windows and shadows that change as the day progresses

• Relatively flat Skittles vs. round balls where a crescent shaped shadow is usually present

• A good, inexpensive USB HD Webcam 8 inches from the target vs. multiple good, expensive HD Ethernet cameras that are 12 feet from the target

Well, two years later, a few camera upgrades, and some professional computer vision software, I'm still working on the project in my spare time.

Alex calls it my "winter project" and as winter approaches, I'm trying to find inspiration and motivation to continue hacking on it. And through the process of writing and sharing mini updates on Insta and Twitter, I'm hoping I can sustain the progress beyond winter too.

I draw inspiration from AccuTennis and Hawkeye

Hawkeye is the IBM and Sony funded tennis system for tracking balls at the Australian Open, Wimbledon, French Open, and US Open. Pros have the opportunity to challenge a line judge or umpire and a computerized animation of the court and ball shows up on the jumbotron with the crowd waiting in suspense for IN or OUT to be drawn on the screen. All of these calculations are done by super high-end cameras and a supercomputer. The cost to develop this system was in the many millions of dollars and to this day the system has been expanded to sports beyond tennis.

In contrast, AccuTennis, developed by 3 ambitious engineers in Philadelphia initially as a side gig, brings the same concept of technology to local tennis facilities with more of an emphasis on interactive training and practice. This system uses very inexpensive cameras and it doesn't use a supercomputer. It makes your local tennis player feel like they are rolling with the big dogs as they have their own mini Jumbotrons (TVs that can sustain a 120mph tennis ball impact) at each end of the court.

My thought is — if my friends at AccuTennis can do it...then I should be able to do the same for bocce.

Bocce needs this type of technology if it is ever to be broadcast on TV and if it is to re-emerge in the Olympics.

There's no doubt in my mind that this ambitious bocce project is possible to solve.

Whether it is practical is a story for another day. Ask my wife how practical it is that I have two expensive cameras mounted to my living room ceiling and an astroturf court on the floor... (head to Instagram for a recent reel of our living room)

See ya next week,

~ Digital Dave