GoalSight.

The application

GoalSight is a football analysis project that combines computer vision with a cross-platform app. Match video passes through detection, tracking, team assignment, and pitch calibration before the application presents possession, movement, heatmaps, and match reports. A human review step confirms teams and player identities.

What it does

• Player and ball tracking with team assignment.
• Player review and naming before the full analysis.
• Pitch projection, possession, speed, distance, and heatmaps.
• Annotated video, match reports, and PDF export.

How it works

Flutter and Riverpod provide the app interface. A FastAPI service coordinates a two-stage Python vision pipeline and passes results to Supabase storage and PostgreSQL. The current analysis service uses an in-memory job queue; model accuracy and processing speed depend on the footage, calibration, and hardware.

Built with

• Flutter
• Dart
• Riverpod
• Python
• FastAPI
• YOLO
• OpenCV
• Supabase
• PostgreSQL