{"id":191,"date":"2025-12-27T15:30:00","date_gmt":"2025-12-27T15:30:00","guid":{"rendered":"https:\/\/tebweb.com\/pages\/?p=191"},"modified":"2025-12-27T15:30:00","modified_gmt":"2025-12-27T15:30:00","slug":"artificial-intelligence-guiding-sports-live-streams","status":"publish","type":"post","link":"https:\/\/tebweb.com\/pages\/artificial-intelligence-guiding-sports-live-streams\/","title":{"rendered":"Artificial Intelligence Guiding Sports Live Streams"},"content":{"rendered":"
\"\"<\/figure>\n\n\n

It started with a client–a local Catholic School athletic program.\u00a0 During the holidays, they host a regional grade school basketball tournament. It’s a big deal–over 7 days they play 73\u00a0games\u00a0with 59 teams from 19\u00a0schools.\u00a0 10-13 hours of basketball each day!\u00a0 The small school gym does not have any bleachers, so in person attendance is minimal. In the past, they have live streamed the tournament with just a lock-down shot from an iPad. In 2022, they came to me to ask if I could improve on this. Being only 2 weeks out, I came up with a quick system–a lock down camera, and an iPad camera shooting the scoreboard. I fed these views back to my office to an RTMP server, and used OBS Studio to create a “score bug.”\u00a0 Best I could do on short notice. It worked, and I was amazed to find an average of 600-900 views for each day—with a static wide shot?<\/p>\n\n\n\n

\n
\"\"
2022 – Static Wide Shot<\/figcaption><\/figure>\n\n\n\n
\"\"
2023 – AI Camera Robot<\/figcaption><\/figure>\n<\/figure>\n\n\n\n

Moving into 2023, the school wanted to work on a better system, and one that could be permanent. It also had to be easy to run with minimal involvement from the school.\u00a0 “Start feed, end feed.”\u00a0 After researching AI driven camera systems, I settled on a French company that produces the pix4team robotic camera system. I was impressed with the videos, and liked how you can adapt it with a regular camera of choice.\u00a0 The AI camera was just part of it. I needed to create a system to control it remotely. In the broadcast sports world, it is known as a “REMI.”\u00a0 (REMote Integration).\u00a0 I needed to create my own small scale REMI.<\/p>\n\n\n\n

Tech Talk Time!<\/h2>\n\n\n\n

If the heart of the system is the robotic camera, the brains of the live stream are from a Mac Mini M1, which sits on a shelf of other network gear in a utility room near the gym.\u00a0 For the initial installation, I ran CAT6 cable for the robot and scoreboard camera, a Tongveo\u00a0PTZ\u00a0camera that can be PoE powered. It was crucial that I can connect to and control the entire system via a remote desktop connection to the Mac Mini. The Mac would use OBS Studio to combine sources and create the feed–and stream it to the world. One problem I had to overcome was that currently, the robot camera is only configured and controlled through mobile devices, Android or Apple iOS. I settled on using a cheap Android tablet.\u00a0 Thanks to GitHub, I found a utility that would allow me to screen share the tablet to the Mac desktop.\u00a0 After installing the utility, I connect the tablet via USB, and turn on Android Developer and “USB Debugging.”\u00a0 Sending a command to Terminal activates the tablet screen share, and also full mouse control from the Mac.\u00a0 I created an AppleScript to launch the Terminal command with a click.\u00a0 \u00a0This has proven to work beautifully–and has been flawless!\u00a0 My hope is that the pix4team developers will be able to make their iOS app fully functional on the M1\/M2 Macs. Until then, this arrangement is working great.<\/p>\n\n\n\n

\n
\n
\"\"
“The Robot” with a friendly face.<\/figcaption><\/figure>\n<\/div>\n\n\n\n
\n
\"\"
OBS on the left, Android Tablet to the right.<\/figcaption><\/figure>\n<\/div><\/div>\n<\/div>\n\n\n\n

A Real Scoreboard Bug<\/h2>\n\n\n\n

I come from a network sports background. So I like to do my best to mimic the look of network TV as much as I can.\u00a0 After all, it is a standard, and people are used to seeing how sports are presented on TV. The scoreboard is an important element.\u00a0 Utilizing a 2nd camera to point at the gym scoreboard, I created the lower screen “bug” to show score, time, even time outs per team. OBS lets me crop and move different scoreboard elements to create the graphic.<\/p>\n\n\n

\n
\"\"<\/figure>\n<\/div>\n\n
\n
\"\"<\/figure>\n<\/div>\n\n\n

The other part of network TV sports I borrowed from\u00a0was in the advertising.\u00a0I have set up a few options for integrating sponsorships (aka\u00a0revenue generation) into the live streams.\u00a0 I integrated the “Commercial Two Box,” now commonplace on network sports. It allows an ad to run, while still viewing the game.\u00a0 Since this stream runs automated, an ad runs at a certain time schedule, regardless of where the game is.\u00a0 Instead of abruptly cutting away from the middle of game play–the game continues in a second box. Viewers are used to this now, and it helps the school athletic programs to realize some income for their programs.\u00a0 Example below.<\/p>\n\n\n\n