I would be livid if it was disallowed.Technically it didn’t but our first goal was bollocks.
I’d be livid if I was a Sheffield U fan.
Technically it was a good goal.Get over it.
I would be livid if it was disallowed.Technically it didn’t but our first goal was bollocks.
I’d be livid if I was a Sheffield U fan.
It was unfortunate for SU... but, according to the current rule of the game (and the way I'd like for the rules to be) our first goal was allowed.Technically it didn’t but our first goal was bollocks.
I’d be livid if I was a Sheffield U fan.
It would be extremely difficult to bring AI in to deal with this, primarily because there's that many different scenarios it would be very difficult to get a training set that would allow the model to be able to deal with all the different possible calls and from all the different angles. The main problem they have is they claimed that VAR would solve all inconsistencies and when you are getting down to pixel level decisions it becomes impossible to make that call definitively. For example, if a pixel represents say 10 cm of pitch then it is impossible to make a definitive call smaller than that.VAR is a fledgling technology going through toothing pains - it's flawed at the moment - but IMO unbiased - to make the technology take the next step, AI is required.
That is, to remove all possibility of bias, AI needs to be employed to referee all VAR decisions - further, anyone should be able to download the current AI state and run games past it for sake of transparency.
I think that we're close to computer-assisted refereeing - maybe 5 - 20 years - eliminating missed calls from deciding matches, and moreover, eliminating refereeing as a possible source of corruption (bribing players to fix games will remain an issue, but I think that this is easier to police).
I very much disagree - there's a huge data set of offside/not-offside calls. It's, IMO, just a skin-cancer type of problem. AI is perfect for this and there's plenty of data to make an offside AI "perfect."It would be extremely difficult to bring AI in to deal with this, primarily because there's that many different scenarios it would be very difficult to get a training set that would allow the model to be able to deal with all the different possible calls and from all the different angles. The main problem they have is they claimed that VAR would solve all inconsistencies and when you are getting down to pixel level decisions it becomes impossible to make that call definitively. For example, if a pixel represents say 10 cm of pitch then it is impossible to make a definitive call smaller than that.
I would be livid if it was disallowed.
Technically it was a good goal.Get over it.
There's huge datasets of offside/onside, but the variety of angles, instances etc means you'd need an enormous dataset that would need to cover every possible version. I would suggest it would make more sense to produce a system that can detect exactly when contact is made to the ball and then detect exactly where the players are at that point. As for the pixel thing, I work in medical imaging (using AI) which has to deal with that exact problem. A model will give you an answer, but the margin of error at distances this small will mean that the answer doesn't mean much, it's guessing.I very much disagree - there's a huge data set of offside/not-offside calls. It's, IMO, just a skin-cancer type of problem. AI is perfect for this and there's plenty of data to make an offside AI "perfect."
As far as pixels are concerned - this is precisely where AI excels.
Due to the huge dataset - all of this will work out. There is no model per-se it's largely just training - AI will self-discover all of the of offside.There's huge datasets of offside/onside, but the variety of angles, instances etc means you'd need an enormous dataset that would need to cover every possible version. I would suggest it would make more sense to produce a system that can detect exactly when contact is made to the ball and then detect exactly where the players are at that point. As for the pixel thing, I work in medical imaging (using AI) which has to deal with that exact problem. A model will give you an answer, but the margin of error at distances this small will mean that the answer doesn't mean much, it's guessing.
Technically it didn’t but our first goal was bollocks.
I’d be livid if I was a Sheffield U fan.
why?? The rule states if the ref touches it, should be a drop ball. He didn’t touch it!!
Are you suggesting the referee should make up the rules as he goes along...??!!
No I am not suggesting that, I’m suggesting the rule is rubbish as yesterday proved.
I want them to sack VAR off and go back to last season’s rules.