Key Outcomes
- Video processing pipeline architecture — ingestion, normalisation, and AI inference across Hudl, Veo, mobile, and broadcast sources
- Computer vision model design for player detection, tracking, and identification from football footage
- Event recognition system — automated detection of passes, shots, tackles, set pieces, and tactical patterns
- Tactical analysis tools — formation detection, pressing triggers, space occupation, and passing network visualisation
- Coach and analyst-facing product design — turning raw CV output into actionable, contextualised insight
- Validated prototype demonstrating the video intelligence pipeline end-to-end on real match footage
Tools & Technology
Our Approach
Football video is the richest untapped data source in the sport. Every match, every training session, every academy game — the footage already exists across Hudl, Veo, mobile phones, and broadcast. What's been missing is the ability to automatically extract structured, actionable data from it at scale. Computer vision and AI have changed that equation entirely.
We design football-specific video intelligence products from the ground up. That means video processing pipelines that handle the messy reality of multi-source footage — different angles, resolutions, frame rates, and quality levels. It means CV models trained on football-specific patterns: player detection and tracking across broadcast and wide-angle footage, event recognition for the actions that matter tactically, and spatial analysis that turns pixel coordinates into football intelligence.
The engagement moves through three phases: video source assessment and pipeline architecture (weeks one and two), CV model design and football-specific event recognition (weeks three to five), and product design with validated prototype on real match footage (weeks six to eight).
What's Included
- Video source audit — Hudl, Veo, mobile, and broadcast footage assessment
- Video processing pipeline architecture — ingestion, normalisation, and inference
- Player detection, tracking, and identification model design
- Football event recognition — passes, shots, tackles, set pieces, tactical triggers
- Tactical analysis design — formations, pressing, space, passing networks
- Coach and analyst-facing product design and UX
- Validated prototype on real match footage
- Implementation roadmap with model training and infrastructure plan