Project flashbacks:

A cursory look at the road behind the team

Who’s the client?

A US-based sports technology company

What’s the solution about?

A mobile real-time tracking and analytics system for basketball teams

What tech innovations were used?

Potent combination of cutting-edge computer vision and statistical analysis technologies

Engaging basketball amateurs during live matches

Imagine that you can transcend the boundaries of a single basketball game by making the fan experience smarter with a data-driven approach. What if they could scan the basketball player with a smartphone and discover more about their favorites and rivals?

Building such an engaging mobile application with real-time tracking and analytics capabilities was the key goal when the client turned to Oxagile.

Innovative object detection approach to involve fans in the game action

After doing deep research, Oxagile’s team generated an ML-based workflow that made the ambitious idea on a napkin work. Here’s what it’s based on:

  • Recognizing only moving players on the court
  • Sorting through smartphone positions in the backend to predict possible detection scenarios
  • Sending player coordinates to the backend system
  • Rendering the detected player ID and performance stats

Equipped with a mobile camera, fans feel like team trainers

Users are now able to point their mobile camera at the court and tap any player to view a wealth of related information.

Player recognition

  • Real-time mobile video processing
  • Real-time data collection from ball- and player-tracking sensors

Player performance tracking

  • Instant access to the captured players’ performance stats
  • Calculating precise coordinates of players from any position around the court


For recognizing players during a game


Of data about players’ coordinates generated


Player detection capacity in a mobile app

Bringing extra value to scouts and trainers

Besides increasing the fans’ involvement in the action, the AI basketball tracking and video analysis app is poised to be a useful assistant for basketball scouts and trainers who want to get instant access to a certain player’s performance stats during the game.

Delivery Model
SLA-driven task backlog delivery (TM-based)
Effort and Duration
2 months, 4 man-months
Technologies
Java, Mobile, OpenCV, Unity, Spring, Jersey (JAX-RS), Apache Commons, Apache Spark, Maven, Tomcat