⚽ Robust Soccer Video Analysis System

Object Tracking & Trajectory Interpolation for Sports Analytics

📌 Introduction

This project implements a robust multi-object tracking system for soccer videos. It addresses common challenges in sports analytics, such as motion blur and occlusion, by integrating deep learning with classical engineering algorithms.

🎥 Demo

🚀 Key Features

1. Trajectory Reconstruction (Interpolation)

• Problem: Fast-moving balls often disappear due to motion blur (False Negative).
• Solution: Utilized Pandas Linear Interpolation to mathematically recover missing coordinates in the ball's trajectory.

2. Stable Classification (Majority Voting)

• Problem: Player IDs flicker between 'Player' and 'Referee' during occlusion.
• Solution: Implemented a Temporal Majority Voting algorithm using a Queue (Window size=30) to stabilize class prediction.

3. Advanced Tracking

• Integrated ByteTrack to handle low-confidence detections and maintain ID consistency.

🛠️ Installation & Usage

1. Clone the repository

   git clone [https://github.com/your-username/Soccer-Video-Analysis.git](https://github.com/your-username/Soccer-Video-Analysis.git)

2. Install dependencies

    pip install -r requirements.txt

3. Run the code

   python src/main.py --source video.mp4