Soccer Analysis System

A computer vision-based system that automatically detects and tracks soccer players in a match video, reads their jersey numbers using OCR, and stores analytics data for further tactical analysis.

---

Features

• 👥 Player Detection & Tracking using YOLOv5 + Deep SORT
• 🔢 Jersey Number Recognition via Tesseract OCR
• 📊 Real-Time Match Analysis with unique player IDs
• 💾 CSV Logging of jersey numbers for each frame
• 🔍 Debug Visualizations for crop inspections and OCR tuning

---

How It Works

1. Detect Players: YOLOv5 detects players as person class.
2. Track Them: Deep SORT assigns a consistent ID to each player across frames.
3. Read Jersey Numbers: Tesseract OCR extracts digits from bounding boxes.
4. Save Analysis: Logs frame ID, player ID, and jersey number to a CSV.

---

🔧 Setup Instructions

1. Clone & Install Requirements

1. git clone https://github.com/Sarthak1311/soccerAnalysisSystem.git
2. cd soccerAnalysisSystem
3. pip install -r requirements.txt

Potential Use Cases

1. Tactical player heatmaps based on movement
2. Player performance analytics via jersey number
3. Automated highlight generation
4. Real-time scoreboard overlay systems

Future Improvements

1. Pose estimation for more accurate jersey region detection
2. Better OCR preprocessing tuned for soccer videos
3. Automatic team detection (home/away color)
4. Web dashboard for visualizing analytics

Author

Sarthak Tyagi 📍 Gurgaon, India ML Engineer | CV & NLP | GitHub