Computer Vision for Amateur Football Analysis ⚽

This project develops a robust Computer Vision system specifically tailored for analyzing local, amateur football matches, known as tarkam (inter-village football). It leverages advanced object detection and tracking methodologies to provide meaningful insights into match dynamics and player movements.

Key Features ✨

Object Detection

The system utilizes a high-performance YOLOv11 deep learning model for accurately detecting critical entities within match footage:

• 👤 Field Players
• ⚽ The Ball
• 🎯 Referees
• 🥅 Goalkeepers

Player Tracking

Integrated persistent tracking algorithms monitor the movement of all detected players throughout the entire video duration.

Data Visualization & Analysis

The system generates visual movement heatmaps to illustrate the density and critical areas of player movement on the field, providing tactical insights.

Performance Evaluation

Model efficiency and accuracy are thoroughly evaluated using standard computer vision metrics:

• Mean Average Precision (mAP)
• F1-score
• Inference Time

Real-time Readiness

The system is designed and optimized for potential real-time prediction and analysis during live match broadcasts or recordings.

Project Goal 🥅

This project aims to bridge the gap between amateur sports and advanced machine learning, offering detailed analytical capabilities previously limited to professional-grade sports analysis.