Detection and tracking
Connected frame capture to a football detection model and persistent object tracking. Added team-colour classification, ball stabilisation and smoother overlay movement.
AI Vision / Sports analytics
Another layer of the match.
A desktop computer-vision prototype that adds player and ball tracking, team colours and analytical overlays to football footage.
Inside the project

Project stage
Research prototype
Contribution
Computer-vision prototyping, analytical overlay design and desktop controls.
Disciplines
AI & automation · Websites & digital products
01 / The challenge
A broadcast shows the action, but understanding player movement and match patterns needs another layer of information. The project explored how to add that layer to footage on a local computer.
02 / The work
Connected frame capture to a football detection model and persistent object tracking. Added team-colour classification, ball stabilisation and smoother overlay movement.
Developed optional layers for possession, movement trails, heatmaps, formation estimates, pass maps and a radar view. Clean and Full modes let the viewer choose the amount of information.
Exposed monitor selection, confidence thresholds and individual analytical layers in a desktop panel. A separate broadcast window was implemented to combine footage and overlays for capture.
Inside the product
Original development captures.
Open an image at full size.
03 / What the work produced
A local prototype implementation, a modular analysis pipeline and training records across five football-model iterations, covering players, goalkeepers, referees and the ball.
Archived development captures from SportsOverlay / Odds Vision AI. This research prototype produces analytical estimates that depend on the model, camera view and calibration.

Start a conversation
A brand, a website or a campaign. Tell us what you have in mind.