Skip to content

moadel2002/KooraKit

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Koora-Kit: Your Ultimate Football Analysis Tool

Welcome to Koora-Kit, the one-stop platform for comprehensive pre-match and post-match analysis of football matches and players. Whether you are a passionate fan, a tactical enthusiast, or a professional analyst, Koora-Kit is designed to provide you with actionable insights and detailed evaluations of every aspect of the beautiful game.

What is Koora-Kit?

Koora-Kit empowers users to explore football data intuitively and insightfully. By combining advanced analytics, rich visualizations, and user-friendly interfaces, it enables a deep dive into:

  • Pre-Match Analysis: Evaluate team strategies, player form, and key statistics to predict match outcomes or prepare for game day.
  • Post-Match Breakdown: Gain insights into team performance, player contributions, and pivotal moments with data-driven evaluations.

Key Features:

  • Pre-Match Analysis:

    • Team Comparisons: Analyze strengths, weaknesses, and trends for both competing teams.
    • Player Comparisons: Analyze strengths, weaknesses, and trends for different players.
    • Player Performance Metrics: Dive deep into player statistics to understand individual contributions.
    • Customizable Reports: Generate reports using LLMs to differentiate between two players in terms of their KPIs. Screenshot from 2024-11-24 16-10-21
  • Post-Match Analysis:

    • Player Detection and Tracking: Automatically identify and track player movements throughout the match using YOLOv8 and Deep-EIOU tracker.
    • Ball Detection and Tracking: Monitor the ball's position and movement, capturing key moments and transitions during gameplay using YOLOv8. Screenshot from 2024-11-24 16-36-40 Screenshot from 2024-11-24 16-37-32 Screenshot from 2024-11-24 16-37-42
    • Team Assignment: Assign detected players to their respective teams using KMeans on players' VGG19 embeddings. Screenshot from 2024-11-24 16-39-12 Screenshot from 2024-11-24 16-40-28
    • Field Localization: Detecting the field's key points using YOLOv8pose to map all activities onto the football pitch for spatial and tactical insights, such as heatmaps and formation analysis. Screenshot from 2024-11-24 16-42-55 Screenshot from 2024-11-24 16-42-06

    Using the output of these models, the following analytics were driven:

    • Heatmap calculation: it highlights the areas where the player was most or least effective.

      Screen-Recording.mp4

    • Speed and Distance Calculation: it calculates the average speed and distance of each player throughout the match.

The whole post-processing pipeline is shown below: Screenshot from 2024-11-24 16-50-07

All of the detection and field localization models were fine-tuned on Soccernet dataset

The models' weights can be found in this drive link.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published