Yes but the players are being tracked by AI and thats why we have a computer image and not the broadcast image.
I dont think we can get any more accurate than we have
CRO vs BEL Matchday 11 - Taylor (ENG)
- Thread starter george.g
- Start date
I dont think we can get any more accurate than we have
socal lurker
RefChat Addict
Does anyone know if that AI is designed to interpolate positions between frames of the camera where the kick point is between?
tbf, he was only making sure Lukaku didn't miss another sitter.
TV clock or his watch?
The AI system determined he was offside, by an extremely small margin, but he didn't touch the ball. That makes it a subjective rather than factual decision, and the referee has to go to the monitor to determine if Lovren was interfering with an opponent.
In fairness to FIFA, and I don't say that very often, they are being consistent as every offside where the player doesn't touch the ball is seeing the referee sent to the screen.
socal lurker
RefChat Addict
Uhh, obviously that would be the TV clock . . . but the TV clock actually tends to start when it should per the LOTG, whereas the R tends to start early.
It's actually a minor pet peeve of mine that refs don't account for the fact they start their watch and then blow. It makes a lot of sense to do that, but if it's three seconds between when you start the clock and the ball is kicked off for the second half, then the half ends at 90:03 on your watch. But the 3-4 seconds is trivial. It's not adding any time for the sub during added time that is a farce--especially when it is the team that wants the game to be over and effectively killed off 30 seconds of the game. I doubt the 30 seconds was going to matter, but especially in a WC where the mantra has been we aren't letting time wasting happen and the teams are getting their full time, I think that is a big failure.
santa sangria
RefChat Addict
Horrible offside. A human has had to draw two armpits there and the scope for error is massive.
It took a looooooong time for this decision - I think what has happened is that VAR has seen it is an unbelievably soft penalty and when they got a sniff of a potential offside they spent 4 minutes getting the armpits to prove it. I feel for Taylor there. If he waves away the dive then we don't have all the nonsense.
It took a looooooong time for this decision - I think what has happened is that VAR has seen it is an unbelievably soft penalty and when they got a sniff of a potential offside they spent 4 minutes getting the armpits to prove it. I feel for Taylor there. If he waves away the dive then we don't have all the nonsense.
santa sangria
RefChat Addict
Great Talk Sport commentary on this one. Reminding about minimum interference maximum benefit and how it was "not football".
one
RefChat Addict
What is it measured with? I know there are some high frame rate cameras but genuinely interested to know where this info is from.
In either case it's all about margin of error an I am yet to see it document anywhere. For example I have seen articles that margin of error for GLT is 3mm to 5mm.
socal lurker
RefChat Addict
As I understand SAOS, a human does not draw anything. The AI does using the various cameras to create a computer image of the player and use that computer image to identify the furthermost part of each player at the moment the ball signals it was played. It isn't at all like the systems that rely on the VAR to mark the players. Humans have to decide if there was active involvement.
santa sangria
RefChat Addict
The image of the two players included a line based on the attacker’s “armpit” - there is zero way a computer has done this automatically. The shoulder point has to be determined by a human moving the line. That’s one reason the review took so long.
It is completely automatic, they claim that the tech can work out where the furthest forward part of the body that can legally play the ball is. The VAR team just check whether the offside player is interfering.
That's not true. The system is fully automatic.
santa sangria
RefChat Addict
I do not believe the machine can choose the shoulder line in this case without human intervention.
santa sangria
RefChat Addict
…and I won’t believe it without seeing the interface
It definitely can, AI can do much more complex things these days than work out where the armpit line is.
You can do a little research yourself. There's plenty of pre-tournament media coverage about how it works.
But simply put it's tracking 29 points on the body of every player 50 times per second and tracks the exact position of those points in 3D space and can determine exactly where in 3D space the last legal part of the 2LD is. You'd simply program it to determine the line to be at x% along the line from the shoulder to the elbow.
\
one
RefChat Addict
For me it is not about if it can do it, it's about how accurately it does it. In this instance from the image it looks like we are talking offside by around 10mm or so. If the accuracy of the system is below that then all good but I don't believe it is.
For instance, assuming the 29 tracking points on the body are on the joints (they usually are) and assuming that is tracked with 100% accuracy (unlikely), then the distance between the shoulder joint and outer shoulder is different depending on the person and muscle mass. Does it really have biometric and physical dimensions data of every player?
Has anyone seen a reliable (not published by FIFA) source that gives the margin of error on the system used?
For instance, assuming the 29 tracking points on the body are on the joints (they usually are) and assuming that is tracked with 100% accuracy (unlikely), then the distance between the shoulder joint and outer shoulder is different depending on the person and muscle mass. Does it really have biometric and physical dimensions data of every player?
Has anyone seen a reliable (not published by FIFA) source that gives the margin of error on the system used?