Driving Cost Savings Through AI Powered Image Tagging
Client Context
We collaborated with a leading sports and entertainment company that owns multiple professional sports franchises. Like many organizations in this space, their content library is massive and growing daily, especially during live games when thousands of new photos are added to their digital repository.
Results
The impact was clear and measurable:
-
$800K in projected annual cost savings by reducing reliance on manual tagging staff.
-
Faster, more consistent workflows, enabling marketing teams to locate and use image assets in seconds rather than minutes.
-
Improved scalability as the repository continues to grow.
The Challenge
Each image needed to be tagged with details such as player, action, and context in order to be searchable by the marketing team. Historically, this tagging was performed manually by staff — a process that was slow, inconsistent across individuals, and costly.
​
-
860,000+ images were sitting in the backlog without tags.
-
700+ image requests per season required ~29 minutes each to fulfill.
-
Average fulfillment cost was $9.13 per request.
-
Inconsistent tagging reduced searchability, slowing marketing campaigns.
​
This was an ideal opportunity to explore how AI could unlock efficiency and scale.
Our Solution
We developed a computer vision–based solution that automatically tags images by both player identity and on-court action (e.g., dunk, fadeaway, 3-pointer, locker room).
​
-
Action recognition
-
Player identification
-
Training data
​
Our approach was innovative because it combined multiple models in a single workflow, tailored to the nuances of live sports imagery.