Joining the dots...
Okay, so you know what's Offside, and you know how VAR has changed that...well, almost! The problem is that it's still not fully automated i.e. the decision-making
is left up to human beings (which has its own set of complications like preference, consistency in performance etc.) Even if they were to get the
decison right eventually, it results in the stoppage of play. It's not real time and has an impact on the viewing experience.
This is where my project comes in. What if we could use AI to make the ofside decisions for us?
There is a ml5 module called YOLO which allows for object detection in real time. We could harness this capability and extrapolate
it to football matches wherein our AI model is trained to detect players and their on-field positioning at all times. And the offside call would be activated instantaneously upon infringement.
Thereby reducing the time required to actually check and verify if the correct call was made.
How do we go about it?
We train our AI model by creating our dataset comprised of offside images and videos. Here's how we can do it:
Build A Data Set For Your Machine Learning Project
Below are a few examples of what the dataset would look like.
.gif?v=1575989257049)
