FootballVision ⚽️

Table of Contents

1. Description
2. Technology used
3. Personal contributions
4. Demo application

Description

FootballVision is an application designed to analyze sports competitions using the YOLOv8 model from Ultralytics. The project emphasizes four main components:

1. Player detection
2. Tracking players throughout the match
3. Grouping teams by the color of their equipment
4. Calculating each team's possession based on the time in control of the ball

Technology used

• Frontend: Streamlit
• Backend: Python
  • YOLO
  • Supervision

Personal contributions

1. Created a custom dataset from scratch using video clips related to football.
2. Solved the Object Detection problem using a trained model with YOLOv8.
3. Addressed the Tracking problem with Supervision's ByteTracker.
4. Tackled the Team Clustering problem using the K-means clustering algorithm.
5. Developed a solution for calculating team possession based on ball control.

Demo application

FootballVision-Demo.mp4