This summer, Johan Ribbeklint, our CEO, had the chance to step out of the office and into the world of professional cycling. He spent a few days on the road with the Lucky Sport Cycling Team during a race in Poland, an unusual but exciting opportunity to support Swedish cycling up close.
Of course, he didn’t just watch from the sidelines. Instead, he and the team here at Buzzcloud developed an AI-powered solution that automatically measures media time and exposure for the team across different streaming platforms. A smart project that blends sport and technology to create real value for both the team and their sponsors.
But how do you actually turn hours of dynamic race footage into clear, data-driven insights for sponsors?
To answer that question, we combined computer vision and machine learning to build a system that could detect and measure every appearance of a key sponsors logo and Team Lucky’s purple jerseys during a full-length race broadcast. Here’s how we made it happen.
Why this matters
For sports teams and their sponsors, data like this is a game-changer. It allows them to:
- Prove media value with hard numbers
- Identify high-value race moments and optimize camera coverage
- Refine jersey designs to maximize logo visibility in future broadcasts
- Strengthen sponsorship proposals with evidence of past exposure
When a brand can clearly see the exposure it’s getting, it’s much easier to justify ongoing or increased investment.
Measuring sponsor visibility with AI: Tracking logos and teams in cycling broadcasts
In professional sports, sponsorship is big business. Logos on jerseys, banners along race routes, and even subtle product placements all form part of a team’s commercial strategy. But how can a sponsor really know how much exposure they’re getting during a televised event?
That’s the question we set out to answer with a recent project for a professional cycling team. Using AI techniques, we analyzed a five-hour race broadcast to measure how often and how prominently one of the team’s key sponsors and the Team Lucky appeared on screen. The results provide a fascinating glimpse into how technology can bring clarity and measurable value to sponsorship deals.
Turning video footage into sponsor insights
Cycling broadcasts are dynamic and unpredictable. Riders weave through the peloton, cameras switch rapidly between angles, and sponsors’ logos may only appear briefly before disappearing into the pack. To track exposure at this scale, we turned to computer vision: a field of AI that enables machines to “see” and analyze images and videos.
We trained two custom object detection models using Ultralytics’ YOLOv8 architecture (one of the latest in computer vision):
✅ One to detect the sponsors logo on rider jerseys
✅ Another to identify Team Lucky’s distinctive purple jerseys
By feeding the models hundreds of annotated images captured in different lighting, angles, and levels of motion blur, we taught them to reliably spot these elements in a wide range of conditions.
What did we find?
We trained the AI models on curated datasets: about 150 labeled images of the sponsor logo and 500 images of Team Lucky’s distinctive purple jerseys, captured from various angles and lighting conditions. Once trained, the models were applied to the full-length race broadcast, sampling one frame per second to detect appearances of the sponsor logo and Team Lucky jerseys while logging details like timestamps and screen coverage.
Here’s what the analysis revealed:
Sponsor logo exposure
- Total exposure time: 1,235 seconds (~20.6 minutes), representing 6.6% of the total broadcast
- Logo prominence: On average, the logo covered 0.27% of the screen area, peaking at 3.91% during close-ups
- Insights into camera angles: Breakaways and sprint finishes delivered the highest visibility
Team Lucky presence
- Total exposure time: 2,213 seconds (~36.9 minutes), or 12% of the race
- Average screen share when visible: 1.63%, with an overall screen share of 0.19%
- Insights into team coverage: Team Lucky was frequently visible but often on the periphery of shots, suggesting opportunities to optimize rider positioning or collaborate with broadcasters for better coverage
This level of detail transforms how sponsors and teams can assess their investment. Instead of relying on rough estimates or viewer impressions, they now have objective, frame-by-frame data on their brand and team presence.
The future of sponsorship analytics
While this project focused on one sponsor and one team in a single race, the potential applications are far broader. Imagine tracking multiple sponsors simultaneously, analyzing exposure across an entire season, or even comparing different events to see where logos and teams get the most airtime.
As AI models continue to improve, we can move toward real-time analysis and richer metrics, like weighting exposure by screen size, clarity, and even audience engagement.
Conclusion
By applying computer vision to sports broadcasting, we’ve shown how technology can bring transparency and precision to sponsorship evaluation. For teams and brands alike, this isn’t just about counting logos; it’s about unlocking insights that can drive smarter commercial strategies in an increasingly competitive landscape.
Want to read more insights from Buzzcloud? You can find more here.
