AI Football Player Tracking & Performance Analysis Model

Development of a computer-vision AI model for football match video analysis, focused on tracking a single selected player and extracting performance metrics. Scope of work: - Player detection model (YOLO) trained/fine-tuned on football broadcast footage - Multi-object tracking (ByteTrack) with single-player lock-on and ID-switch handling - Two-stage processing: fast preview for player selection, then full GPU analysis - Pitch calibration to convert video coordinates into real-world positions - Metrics extraction: distance covered, speed, sprints, activity levels (standing to sprinting), positional heatmap and zone occupancy - Structured JSON output ready to feed reporting and AI-generated analysis - GPU inference pipeline (serverless) with testing and accuracy tuning This contract covers the AI model and analysis pipeline only. The web platform, user interface and report design are not included.