Skip to content

R-C101/CVAR

Repository files navigation

CVAR (Computer-Vision Assisted Referee)

The CVAR is a comprehensive application designed to revolutionize the way football (soccer) player performance is analyzed and documented. This project aims to provide a platform that meticulously monitors and tracks various aspects of the game, including player movement, ball possession, referee decisions, goalkeeper actions, and more.

Features

  • Player, referee, goalkeeper and ball tracking
  • Ball possession Tracker
  • Player Speed and Distance Travelled Estimator
  • Camera Movement Tracker
  • Ball Interpolation.

Installation

To get Started:

  git clone https://github.com/R-C101/CVAR.git

Get the model from Here:Model 100 epochs

The provided stub files are for the default input video. The code mentions

read_from_stub = False

If you want to save computation time remove the stubs in the stubs folder and put

read_from_stub = True

in the main.py file.

After the first run if you want to run the files again, the object tracking for the file will be saved in stubs and you will save a lot fo computation time for the second run.

The model I have used was trained for 100 epochs.

You can train your own model from the training directory. Just run the entire ipynb.

Put the model inside the models directory after the training.

TODO

  1. Get the coordinates automatically instead of having to assign them manually

  2. Ensure that the player_id doesnt go above 30

Screenshots

Output Screenshot

External Links

About

a CV model that analyses football matches including every player, the ball and the referee.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors