Let's not forget he wasn't exactly playing his best. He lost a set against Edmund and was about to lose the first set vs Carreno Busta. Sure, he probably would have turned it around but we don't really know. Thiem was always going to be a big challenge as well!
I stand by the notion (especially after that US Open match with Serena) that even if the technology is inaccurate, human line calls are just far more inaccurate to leave millions of dollars on the line for.
Exactly. People get furious when a human being with a fallible set of eyes and mind get a call wrong of whether a ball going 100 miles per hour was in or out with a margin of millimeters. Then, when you implement a computer system that is right 99% of the time and makes one slight miscalculation, they get furious again. I'll take almost never wrong over people who are wrong quite often lol.
Having watched my first lot of tennis with automatic line calling at the Australian Open, I have to say that I found it a relief as the game flows so much better without the constant challenges and their delays.
It helps Americans players as a lot of umpires have shown anti American biases before. I’m not american but I want everybody treated equally so I’m pro it although I miss line judges
I like the human error aspect of sport. Just look at the state of the premier League now and it's ridiculous by-the-book handballs and millimetre offside checks after every goal - it's a shambles
A computer is more accurate than a human. Yes computers make mistakes and have errors. But I would take a computer calculating where a ball lands going 100+ mph over a human any day of the week.
@Heather i think it's a matter of time until the playing population understands/trusts technology and the game shifts. 2020USO had some courts with only these cameras so it should come soon
@@MeFreeBee and the angry man woman swinging a bat at it 😂 although to be fair, the best news for me is that there won't be any more Djokovic incidents if line judges are replaced by computers fully
Some of those calls in the Williams/Capriati match were like what I see in my local doubles comp. Every ball that lands on the line gets called out. Lots of revenge calls in return!
So there is a little back story to that match. Earlier in the year the Olympics were held and several top umpires were caught manipulating their Olympic credential so as to be able to see other Olympic events besides tennis for free. They were caught and a final judgement was made during the 2004 US Open. The judgement resulted in several top officials being sent home from the US Open. (Look up some news articles about it) this on the court for that match as the chair umpire was a woman who was up and coming at the time but not ready for the big time just yet but in the vacuum left by the suspension of top officials she drew this assignment. She is the one who started everything by overruling a ball to out that was called in by the line judge and was clearly in. After the ensuing mayhem she panicked and stopped overruling anything. The line judges also got nervous and began blowing calls.
@@chloeharvey5684 That was no joke. That was an understandable mistake. They clearly thought the guy is Djokovic, and it is very difficult to tell them apart.
@@chloeharvey5684 Dear Jesus, what jokes? Are you a writer for Bill Maher? When he gets things right, he is a social commentator. When he makes mistakes, he is a comedian now, don't you get it? Am I supposed to see error in judgement as being a joke? I thought so, but no.
a) More accurate than humans by far. b) Unbiased. c) Always at full concentration. d) Noone for players to get mad at. The $60k price tag is prohibitive to human beings, for decent tournaments, it's affordable.
@Heyzyen Funny you should say that. Have you heard about the racist soap dispensers? Basically, the developers were racially ignorant, so they didn't test the motion sensors for non-white people, so only white people could get soap. So whenever you create something, it's possible that your own biasis slip in there. If there was a program using machine learning to get the Hawk Eye effect and it's only fed shot by men, it might not be accurate for shots by women (who are generally a bit slower but with more spin). Never underestimate the ability of humans to create flawed stuff.
@@DrZaius3141 yeah - but i presume hawkeye does not track the Players - i reckon it kind of ignores them, to test Hawk eye you would after all use some sort of serving machines for precision.
@@Zwangsworkaholic I should clarify: I don't think at all that Hawk Eye in its current form has any bias, I was just trying to point out how it could be feasible. Any automated system has certain parameters it's optimized for. To get back to the bathroom with another example: it's still not uncommon that mirros in women's bathrooms are placed at heights where many women can't see themselves. The height is optimized for men and noone put a second thought in it.
@@DrZaius3141 An AI trained on a limited dataset is not "sexist" or "racist" because it doesn't work properly on a completely different dataset. It's literally impossible for a computer to be sexist or racist, even if someone programmed a computer to say "women should stay in the kitchen" the computer still wouldn't be sexist, the programmer would be.
It’s not about fundamental accuracy in a vacuum, but about being better then a human. So long as a computer is even slightly better then a human it will be chosen. In addition people are far less likely to get angry at a computer for a seemingly bad call, rather then a human.
@@The_SUN1234 1. I would definitely pay to watch robots compete in sports events 2. I wasn’t even advocating robots competing, but adjudicating. 3. What? Are you pro or anti anger? I genuinely can’t tell
What about robotic players? I mean where is the line drawn here? You might scoff at the notion but so did people 30 years ago at the thought of robot umpires and linesmen
While you are correct for calling balls and strikes-and heck, you could use hawkeye with a snickometer to get force plays and foul tips-there are some things that you’d need an umpire for. A computer doesn’t know what a balk is.
4go101 for example if someone spits on a ball or uses tar to have better grip it’s considered a balk . You can simply decide if a pitch is a pick off or a balk . Humans have a deeper understanding in context sensitive things . For most things in baseball especially you can use modern slow motion cameras to support almost every decision making .
Some history from cricket's side - Hawk Eye's first use in cricket back in 2001 was also for entertainment's sake. It wasn't until 2008 i.e. after tennis implementation that it was used to make official umpiring calls. Compared to tennis Hawk Eye calls in cricket have to be more predictive. The reason is because ball never reaches or touches the surface the wicket - the surface to which Hawkeye needs to predict the path to. You can see the example at 1:54 - red is the actual movement and blue is the Hawkeye prediction line to the three yellow lines (wicket). In cricket this is called LBW or Leg Before Cricket calls. Since its introduction it has as being quite controversial. Earlier Hawk Eye rules stated that if the distance between wickets and the ball impact was more than 2.5 meters Hawk eye cannot be used. This was very controversial because people couldn't understand why Hawk eye failed to predict distances above 2.5 meters. Aside from that Hawk Eye used to often misinterpret the path when a cricket ball was spinning. It has led to some hilariously wrong calls. You can find videos on YT about that. While the technology has gotten better with time it is still controversial for the "umpire's call" seen in 1:54. The reason is that in cricket Hawk Eye needs to make an actual prediction. So, if a ball partially touching the yellow lines or wickets there is no 100% certainity of that reallly happening . So, if there is some degree of uncertainity the decision is more aligned to the on-field (court) umpire's call than Hawk eye's call. Some say Hawk Eye's prediction has gotten better and it is 100% reliable but rules in cricket still don't allow that. That said, Hawk Eye had a huge impact in the way cricket is played. Cricketers used to often thrust out their legs to stop the ball from hitting the wickets. While LBW was a rule to stop people from doing that it was still a misused tactice. With Hawk Eye people need to be extra careful about using their legs and this has had a huge impact in the way cricket is played.
I feel that Hawk-Eye would work best in tandem with full replay review for visual confirmation - adding replay like in the ATP Cup this year would also help check double bounces, lets, etc. which Hawk-Eye can't do
Depending on the angle and velocity of the ball, as well as the rotation, some balls will slightly "skid" or "skim" along the ground at the moment of impact, rather than bouncing directly up or down. This is more pronounced on grass surfaces seen in your example at 7:25. This skidmark covers a larger area on the ground than the ball bouncing directly on its axis
Players cannot argue with a computer. However, I would like to see the lines themselves use some kind of touch sensitive sensor, linked to the ball coating.
I fear there would be false positives if the ball bounced close to the line and the sensors picked up the bounce of the ball even if it didn't touch the line
Worked along side Hawkeye many times. There is a LOT of work that goes into it and it's a huge team. Every single camera is configured and represented in a 3d simulation. Their smallest setup requires more people than our own crew. I highly respect their work and I understand technology can be flawed, but they are working hard to do deliver accuracy. I genuinely have nothing but respect for them and glad to work alongside them. Hoping they can bounce back after the lost of so many tournaments.
In the early '80s, Vic Braden was using hi-speed video to check the accuracy of lines calls. I seem to remember that the consensus was that human lines persons were pretty inaccurate on close line calls.
Vic’s show was awesome. He also used the videos to show that almost all players stop watching the ball at about 3 ft in front. Which makes Fed even more amazing - he watches it all the way to his strings.
I think the problem with Hawk Eye being able to see that a ball was in when no human could have ever seen it like in the one example you showed is that actually changes what is considered in and out. Tennis players have always known these sorts of shots to be out and played accordingly. It’s almost as if the net was suddenly a centimeter higher.
The reason why at 7:30 the ball looked as if it bounced out, is because it DID bounce while it was out. The ruling being the way it was, was based on it being a grass court. The ball grazed some white painted grass before it bounced(out of bounds). Based on that ruling, there can be slightly wider boundary margins on grass courts, especially with low-angled high-velocity shots, be it as few and far between as shots like that may happen.
I'm so glad that the line calling technology is being widely adopted. I've always felt it was absurd to have all of those people staring down the lines when we know that that the human eye is incapable of judging the ball accurately at those speeds. Prior to this technology it would've made as much or more sense to simply have the chair umpire make the calls. Hopefully linespeople will soon be a thing of the past.
What I think is interesting about the introduction of technology into referee calls of various kinds is that it will require getting specific with rules, finally. So many professional sports, with millions and millions of dollars at stake, have vague rules that are up to the subjective judgment of refs. That's the way it has had to be to cope with the limitations of humans, but no more. We can get be specific down to the millisecond over what counts as a traveling call, then let the computers and cameras call them. We can be specific down to the precise positions of hands and distance to the ball about what counts as pass interference, and let computers and cameras call it.
There is 1 major inaccuracy in this video. While Hawkeye cameras are used to view the baseline in a Hawkeye live setup, they do NOT “track footfaults as well” as stated in the video. The review official watches the baseline on a screen and if he/she sees a footfault, they hit a button and a voice says “footfault “over the PA system. So there still could be footfault controversy with Hawkeye live.
because computer calculations understand that where a ball bounced from is not necessarily where the ball made contact with the ground. in other words, they can account for skid.
Wow, your videos are VERY well made -- thoroughly researched, good pace, enjoyable editing, and all the right topics that satisfy a curious itch. Keep it up, buddy! Love your work! Cheers from Korea
Hawk Eye operates with 10-12 "slow" cameras with 150 frames per second per camera (the video mentions 340 fps -perhaps Hawk Eye has improved). In contrast, Foxtenn ("Real Bounce" system) works with 40 high speed cameras with 2500 (!!) frames per second per camera (already used in quite many ATP tournaments). Hawk Eye interpolates/calculates/estimates the ball trajectory within the frames and shows a computer generated image on the basis of the calculations. Therefore it is sensitive to shocks like wind and trembles, etc. Its expected systematic error is 2,6 mm. Foxtenn shows the real video image of the ball's path and due to its extra fine time resolution, its systematic error is basically zero. Think for a moment to understand why the difference in resolution is so important: a ball flying at 126 km/h travels 35 cm in 0.01 second (!) or 23 cm between Hawk Eye frames. For 90 km/h it is 25 cm or 17 cm. So in the case of Hawk Eye, some wind gust after the last frame taken before landing can easily divert the ball with a few mm from the expected path.
I feel that when we get to the point where we start asking questions like "what actually counts as the ball touching the ground", we may as well just flip a coin. It's like trying to measure sprints to the millisecond, where it's just about how we define crossing the line.
In cricket, even now in 2020 after 19 years of advancement, the use of Hawk Eye has limitations, and isn't as accurate due to the more inconsistent nature of a cricket pitch, the unique spin and swing nature of each ball as it is worn out through a match, and the effect of the seam of the ball in a bounce. If I recall, something like more than half the ball has to be impacting the stumps in a hawk eye replay for it to be considered confidently hitting. This is of course a bit different from tennis where the ball does eventually bounce after the estimation, whereas in cricket it remains an estimation and is used to predict the path of a ball if there had been no other action or impedance against it. Great vid! loved it.
Tennis is better with Hawkeye. Gone are the challenges and arguments with line judges. Players get to see the call immediately and then move on. It doesn’t need to be perfect, just consistent.
I did the math and on a 150mph serve Hawk eye’s camera take a picture of a frame every 8 inches the ball moves. Which is pretty large especially relative to the ball’s diameter. I think having the camera’ be better will dramatically reduce the likelihood of a wrong call
Isn't it also true, that it is still vastly superior to line judges. Plus, changing the rules to giving the benefit of the doubt the ball being out will make things a lot easier. The system is still largely superior to human judgement. I feel the data set generated can predict the parabola of a tennis really well.
You know why they don’t want to even have it on clay courts for even “fun reviews”, similar to how it was years ago, is because with the actual clay marks, you will easily see just how much room for error 2.6mm really is. That 2.6mm is 3D spacial error, which is spherical… meaning that the ball itself could flex or wobble its shape enough to throw the calculations off. A calculated “in” because of its estimation may actually be an out once it actually hits the court and vice versa. A good example of this is that hard court call that Hawkeye had wrong and you could clearly see the fuzz mark with no other ones around… and that was way more than 2.6mm off in error. If they truly believe it’s that good, let it prove itself on clay courts too. Heck, if anything else, they could use clay reality to calibrate it even more. I would put down money that it would make errors on clay that we could actually see, but obviously they don’t want to do that and lose credibility. 🙄
I never see TH-cam videos on tennis other than highlights compilations of the same old points. This is a welcome change for all tennis fans and I hope you keep making awesome videos like this!
The future for tennis is probably a combination of Hawkeye and the Foxtenn one you mentioned in the last minute. Foxtenn coupled with Hawkeye will give real data to improve Hawkeye's algorithm. On the other hand if players feel like Hawkeye f***ed up they can use the Foxtenn footage as a challenge.
No need for combination -at least for the basic service of linejudgeing (Hawk Eye has quite many additional smart services, too). Foxtenn is simply superior in accuracy. Hawk Eye operates with 10-12 "slow" cameras with 150 frames per second per camera (the video mentions 340 fps -perhaps Hawk Eye has improved).. In contrast, Foxtenn ("Real Bounce" system) works with 40 high speed cameras with 2500 (!!) frames per second per camera (already used in quite many ATP tournaments). Hawk Eye interpolates/calculates/estimates the ball trajectory within the frames and shows a computer generated image on the basis of the calculations. Therefore it is sensitive to shocks like wind and trembles, etc. Its expected systematic error is 2,6 mm. Foxtenn shows the real video image of the ball's path and due to its extra fine time resolution, its systematic error is basically zero. Think for a moment to understand why the difference in resolution is so important: a ball flying at 126 km/h travels 35 cm in 0.01 second (!) or 23 cm between Hawk Eye frames. For 90 km/h it is 25 cm or 17 cm. So in the case of Hawk Eye, some wind gust after the last frame taken before landing can easily divert the ball with a few mm from the expected path.
Your content and voice are like the Core-A Gaming of tennis. I don't even watch tennis, yet I subscribed to your channel, I don't play fighting games, but I subscribed to Core-A.
Even if it calls your ball out when it was in, it will also call it in when it was out so chances are throughout a player's career they will break even.
@Vlasko60 I think Cricket matured a lot more than Tennis with usage of Hawkeye. Since it is not 100% accurate, in cricket, umpires decisions are considered for marginal decisions. Instead of showing them as hitting or missing, which is akin to in or out for tennis, they are referred to as "Umpire's Call".
This is only true if those errors apear very often. Its called the law of big numbers. Basically, if you roll a dice 6 times and track the results, it would be unlikely to be evenly distributed. But if you roll dice 60 times, 600 times or more its more likely to distribute evenly between all 6 numbers.
I’m reminded of the Agassi-McEnroe match in the Wimbledon semis in 1992, in which after the whistle for out-of-bounds kept going off at the wrong time on McEnroe’s serve, Agassi came up to the ref and said, “Turn it off. Just turn it off. This things been screwing up for the whole tournament.”
Capriati in that 2004 match against Williams was incredibly dishonest; how can you not see those balls were in and steal your opponent's points. But the look on her face during the on-court interview when faced with questions about this was very telling. She knew Serena was being cheated, but was cool with it. I suppose that's how she'd have liked it for herself.
It's not the players job to make calls, it's the ref. There's no telling that Williams wouldn't have returned the favor if Capriati had been honest, so there's no incentive for her to do a job that's not hers by making calls she thinks are in or out (especially when she may not have good vision on it either.)
@@squidge903 thats on paper.. But in Human heart there is something called sportsman ship .. when you know you don;t desrve a point you don't take it.. if u r taking just because it's given .. it may not be your fault but you are not worth it..
@@Straightforward786 It's not that simple. In reality, it's often hard to say for sure if a ball was in or not, especially as a player on the baseline with the ball coming at you fast and landing near the baseline. Sure, it's often clearly out. But by definition, it's the closer calls that line judges get wrong. So, in practice, most of the time. As a player there will be times when your opponents ball is called out but you know it might have been in. But equally, you're far from certain. The only reasonable thing to do is let the call stand unless it's very clearly wrong. And, like I said, the very clearly wrong doesn't happen much - and I don't mean very clearly wrong on the replay, I mean very clearly wrong as you see it in the real time as the player. That rarely happens because nearly all incorrect calls are when the ball lands very close to the line.
To me the fact that it predicts where the shot should go instead of where it might actually go is a problem. It might be accurate most of the time, but if there is a slight breeze down the middle of the court, it would be enough to move the ball a millimeter, which is the difference from a ball being in or out. It makes more sense to me to just have those cameras focus on every inch of all the lines (and from varying angles) to determine if it’s out or not.
That's really interesting, to be honest I thought it would be more accurate. Before Hawkeye I used to think that one day the actual lines would be "smart" with the ability to sense contact with the ball/ players feet. Thanks for posting
I love that idea! I find it interesting how different minds work. I would not have looked at it from that aspect. To me, those examples of the camera seeing the exact bounce and skid and everything would be a good idea. It is apparent that the technology exists now, but maybe those cameras have to be so close to get that level of accuracy? I don’t know, but I do like your idea
The key here is even if the computer is inaccurate, it is consistent. If its precision is high, that becomes the new standard and any pro athlete can and should adapt. One could argue that a line judge is less consistent due to inherent biases. Thus the precise, if not perfectly accurate, behavior of the computer system is superior to the imprecise (and still not perfectly accurate) human judge.
One thing that always confuses me is that the rules of tennis state that the ball has to TOUCH the line, but hawkeye will call IN when the shadow of the ball OVERHANGS the line. Like if the very, very edge of the ball touches the line on a hawkeye replay it's "IN" despite the fact that that part of the ball can't possibly have touched the line. Doesn't make any sense to me...
The question is - does being "accurate" helps or hurts the spirit of the game? "You cannot be serious" moment is legendary. It would not exist if McEnroe had to argue with a computer... I'm leaning more toward being human and imperfect instead of having sterile, 100% accurate robot. I think the way it is used now - for a limited number of "challenges" is a good compromise. Replacing human judges completely would probably be a huge mistake.
Then any motion that blocks the laser would trip it. Bugs, leafs, trash, rocks... even a large enough amount of dust can trigger a laser sensor. Also, what if a player's foot is already blocking the lasers line? The laser would also have to be above the ground, not on the ground so it would actually get trigger before hitting the ground saying the ball is in when it could just barely be out.
I agree that lasers alone would not work but perhaps in addition to the cameras. Might be a lot of extra work and cost for little gain. I suspect you would need laser on each side in case there is a foot and a ball at the same time. AI would have to figure out which trip to use and hopefully Biggest issue trying make sure there timing is the same too because at 320 fps if they are out of sync a little bit it could cause trouble. Ball hits at .002. Was the trip at .001 the ball or something else. I think it would make system WAY more expensive and in a little if any gain. I do like your thinking behind it though.
The issue with Hawkeye is: If it breaks during a match (which is certain to happen with thousands of matches each year), then how do any line calls get made? You would have to have some line judges on reserve in case that happens.
what do you mean "break"? how can a program break at all? i suppose game stops for a while and programmers will get it working again i think. it would not be an issue at all.
one viewer - what I mean is at some matches, players will challenge a call. Then, the umpire is told Hawkeye isn’t working. Maybe it can be fixed, but I don’t think the tournament would pause the match to fix it. No telling how long it might take to get it working again.
@@saberswordsmen1 and yet, that system would be million times better than human eye. as i said, it should be maximally optimaized before using it and i think it should work fine.
@@CULTTENNIS haha no doubt. Just came across your channel and I subbed immediately. Keep doing what you're doing and this channel will grow huge. Just got done watching the Coria video. Classic footage!
In the example at 7:50, the ball is clearly out. What, are we taking into account how hairy the ball happens to be? Also, an ‘average’ accuracy of 2.6mm doesn’t give the full picture. What is the spread of accuracy? And the worst case?
First off, well produced video. Second: 8:43 is NOT a foot fault. Also, Serena might not have actually foot faulted. Too close to tell, but saying she would "get away" with it, is really snide of you.
I agree with hawk-eye technology. It is amazing that 10 cameras are focusing on the movement of a single ball, speed and bounce, and mathematically can predict whether in or out. Other factors that influence the ball movement, such as humidity, temperature, shades and wind gusts are components of mathematical model. It is more accurate of course than human eyes, considering that players now can hit the balls with much velocity. May be putting more cameras (say 20) can achieve 100% accuracy. Cult Tennis, Great presentation. Thank you.
The truly weird part is incorporating the deformation of the ball into the call and ignoring that a spherical object can only make initial contact with the ground at a single point, meaning if only the deformation of the ball touches the line, the ball was in contact with the ground outside the line before that and the call should be Out.
Totally agree. In cricket it is 'not out' (umpire's call) if less than 50% of the ball is touching the stumps/bails. Cricket balls have no deformation. In tennis, 10% would do away with stupid edge cases.
I think it is generally accepted that the interpretation of the point of contact is not the instantaneous first point of contact, but the entire area the ball contacts on that first bounce. So though the very first point of touch could be out, if the ball flattens to impact onto the line as well then it is in.
Weird tennis would set itself up like that with the impossibility of trying to "see" the complete deformation of the ball as it strikes the ground. Seems to me if any part of the ball touches out of bounds, before any part of the ball touches inbounds, the ball is out. Like in every other sport with boundaries. But then I don't play tennis. ¯\_("/)_/¯@@mattc3581
Part of the reason that Hawkeye isn't used on Clay is that the spot of the ball is visible. You'll notice that the balls always produce a small amount of clay spray every time that they hit. Watch the close ups. That clay dust can mimic the ball with Hawkeye. The ball could be a fraction of an inch out but still register as in. If they can ever come up with Hawkeye that recognizes the ball and not the clay dust then they'll use it, but I just don't see that happening.. More importantly, and I've played on clay so have seen it. That's why they're not used on clay.
It doesn't really matter if it's off by 5mm or 5cm, what matters is it's unbiased towards players and off by a consistent amount. If you're not hitting the ball because you think it's going to be 1mm out of bounds, you're making a bad judgement call. You should be hitting every ball that has a good chance of being in. 1mm out of bounds is still a very well placed shot.
Here is the deal, if a ball is hit with 100 miles per hour, a camera that takes a picture 340 times per second, is only accurate for 13cm. 100 mph => 160 kmh => 44m/s => 44m/s *(1/340)s => 0.13m. Considering the two images beofre impact and after impact is pretty much what the system does, only more sophisticated. For fun if you want to be as precise as 3.6mm you need 12356 images per second, which is bonkers. (Funny number 12345).
@@Schnorzel1337 no, that's not how it works. If you see a truck move at a steady speed from a to b, you don't need to have recorded its position at every single point to know where it would be at a certain time
@@patrickbeart7091 Good that was never the question but good. @Nazraq S suggested a "real camera setup" that shows where it actually hits, which is not possible at all. If you want to get decent results just for the camera you would need 12356 images per second so the first image where the ball touches the ground captures the ball at most 3.6mm from the initial impact. That part obviously ignores friction and everything else.
@@patrickbeart7091 For context if you had a truck moving at 100 mph then the truck would be blurred across in the direction of travel, making the truck not only appear longer than it actually was, it also blurs the edges and distorts it. Looking at a tennis ball moving at 100mph at 300fps looks like a pill, when it hits the ground it deforms and skids and all of those actions are, consistently, declared as being part of the contact and determines whether it's in or not. To achieve such a system you'd need cameras that either play at insane framerates, or you'd have to combine THIS system to predict where it would hit with a camera capable of a 10000+ of a second exposure time to achieve a "picture" of the ball actually hitting, which adds another problem, the 10000+ frame time would likely need to be a series of shots to get all of the frames where it's contacting. So, no, a system that actually monitors the contact of the ball is just a technological hurdle that nobody is really willing to foot the bill for.
I have never ever considered hawk eye to be wrong. It would be highly audacious and stupid of me to sit infront of my screen, consider a ball out, have hawk-eye disagree, and then say hawk-eye was wrong... the truth is unless there's a great replay, its impossible to tell many times. But people really do underestimate a ball brushing the surface and landing beyond the line, which was greatly exhibited in the Tomic example. All in all hawk-eye has a what, proposed 99.7% accuracy? Far greater than the human eye, OBVIOUSLY. And its few malfunctions of not working are one out of million. In fact, ive only seen the machine not work entirely once (Azarenka example). The other times it just couldnt appear as visual.
As the game becomes more precise maybe people will learn the tendencies of what they can get away with on each court and change their strategy with that knowledge
The only question to ask is "What is the most accurate calling system, the human eye or the electronic Hawkeye?" Since the whole point of calling balls in tennis is to determine which hit is in or which is out, then whichever system does this best is therefore best for the game. Make the accuracy determination and you have your answer.
I'm appalled at the overall quality of this discussion thread, so far as I read. Integrating multiple observations of a moving body in a clever way goes all the way back to Gauss, who invented least squares for the purpose. The real issues here are optical distortion in the camera lens, imperfections in the court surface (e.g. drainage curve), synchronization of the camera feeds, distortion of the shape of the ball after impact with the racquet (which likely oscillates in the air for a few meters as it sorts itself out), non-uniformity of the ball (e.g. moisture off the court surface, scuff marks), motion blur in frame capture, variable illumination, optical turbulence (which also affects human vision), and flight turbulence with shifting and non-uniform wind patterns (which the players can't predict with precision, either). Human vision is far from infallible concerning all these variables, either. The main difference is that a human judge on the court is that the human judge is given a preferential viewing angle to final impact, whereas the camera system has far more eyes working in harmony, at a far higher speed, combined with a rigorous application of linear algebra to minimize the error term in the coherent reconstruction of the _entire_ flight path of the ball. The human brain would incorporate more Bayesian analysis based on all distributions and patterns in all the balls previously witnessed (spanning the entire range of variables) producing the kind of accuracy we're used to having, for what it's worth. However, the problem with the human mind is that it's not correlation-free with respect to random, recent events. If you fought with your spouse the previous night, you get a different error function. You get caught up in the emotions of the match or the day, you get a different error function. You mentally remind yourself to factor these things out so as to remain maximally impartial, you get a different error function (which might be less biased, or it might not). If you're still kicking yourself at a bad call three points ago, you get a different error function. There's _heaps_ of psychological research showing that we're rather bad at calibrating to a fully unbiased center. We're so bad at this, that we've learned to substitute the _performance_ of an unbiased arbitrator for the real thing. In almost all sports, referees keep a mental account of which party _appears_ to have received the short end of the stick on previous judgement calls, and then subtly tilts the scale back toward restitution. The NHL is terrible for this, because it's a target rich environment. There are technical obstruction penalties on nearly every hard-fought play. Awarding a 2-minute powerplay is a big deal and easily impacts the final score. Awarding a slew of 2-minute powerplays back and forth is just as bad, even if completely balanced, because it changes the sport from 5-on-5 to alternate 5-on-4, suitable to a completely different set of players (often one team is better 5-on-5 while the other team prefers alternating 5-on-4 and 4-on-5). There's an entire unspoken science of "game management" on the part of the referees (usually unspoken, except for now and again an embarrassing open mic). It's such a dark and unseemly underbelly that the league hands out $10,000 fines like dinner mints to coaches who state the obvious after the game. In tennis, the players fall under the known illusion of control. They don't actually know the minor details of wind turbulence on the court, which easily affects their shot placement on the centimeter scale. When they hit that close to the line, it's no longer skill, it's luck of the gusting breeze (or tiny convection cell on the court during a hot, windless day). If their balls are being called out on less than a 5 mm margin, they should simply be hitting their balls a few cm shorter on average (or they should simply accept that this is a percentage situation, and you can't win them all). What finally matters is that machine vision is repeatable and impartial for the vast majority of common circumstances. This has been achieved already to a very high standard. The few exceptions to the exceptional performance of this system are hardly changing the long-term outcome of tennis as a sport. If the two players are so evenly matched that one player prevails over the other only because one player was on the right side of more 5 mm line calls, the idea of a final winner and loser is more of a coin flip than a demonstrated superiority. The entire postmodern culture wishes not to believe that "equity" often hangs in a highly material way on the last decimal point. But the truth of the matter is that all systems with chaotic elements are capable of going left or right on a decimal point far beyond human apprehension. It's a fundamental property of chaotic systems: somewhere in the configuration space, left and right are jammed together so tightly that no possible perception can pry them apart, except to sit back and watch what happens. The infinite recursive pattern of the Mandelbrot set is trying to tell you something, if you're more into telling than yelling (at his level of nuance, most of the population sides with yelling). Even the most elite tennis athlete needs to give it a rest in the implied complaint that they can hit a ball 100 MPH in outdoor conditions to the accuracy of a quarter of a beer cap twenty-five yards away. They are psychologically wired to want (and need) to believe this, but it's simply not true. Finally they need to chill out, STFU and accept an impartial outcome from an electronic deity of linear algebra as implemented to this exotic standard.
When ever I’m watching a tennis match and there’s a challenge I always try to guess if the balls in or out and I’m right about 30 percent of the time lol 😅.
I think it's basically reliable. I am a bit sceptical when there is a camera pointed right at the spot where it hit but they won't show that shot and only show the Hawkeye generated animation. The Hawkeye camera is around 350 fps, but there are video cameras that can shoot up to 1,000,000 frames per second. In matches where multiple people are seeing Hawkeye being off, It should be an option to have 1 super camera verifying it's accuracy. This would be a whole process of what shot to measure, what angle would give the least optical distortion etc...so don't just say it can't be done without justification.
The issue with replacing line umpires at top tournaments is that you remove a lot of the incentive for people to get into umpiring in the first place. The unintended consequence of that might be a shortage of umpires at lower levels of tennis.
2 ปีที่แล้ว
Very like postmen, of the kind of jobs that makes no sense anymore.
Bro, you’re videos are really good, you speak well and you’re able to make it not boring, nice editing and nice “hosting”! Keep up man, loving ur channel
they might as well build some touch sensitive lines to accompany hawk eye, as the lines are where the arguments stem from. that way you get your 3d hawk eye visualisation and also the line sensors tell you if they were triggered, in case its very close
7:27 As commented previously.. this is Bernard Tomic, not Djokovic.. Djokovic was never sponsored by Nike.. Shoes always Adidas, clothes Sergio Tacchini, then Uniqlo and now Lacoste.. and Bernard Tomic has one of the weirdest forehands btw
As soon as the naked eye could see on TV the inaccuracies of the calls of the naked eye on court between Serena and Jennifer, it's a no brainer to get a system that would contest the call on court, which is the least accurate.
The linesperson told the umpire and the supervisors that Serena said "fuck you" (She didn't actually, she said "you better be glad that I don't shove this ball down your fucking throat"). When Serena heard the linesperson repeat this, she misheard her and exclaimed "I didn't say I would kill you, are you serious?!" That's when one of the supervisors brought her in close to say that the linesperson was actually alleging that she said the f-word, and you can see her admitting to it before finding issue with the course of action taken. All of this clearly recorded and most of it is audible with the on-court mics so it's shocking that it still gets routinely misreported to this day.
Who else agrees that if Hawk Eye Live was used on Arthur Ashe stadium this year, Djokovic might have won the whole thing!!
Why did they not use it this season
Nope
no. he was playing shit.
Totally
Let's not forget he wasn't exactly playing his best. He lost a set against Edmund and was about to lose the first set vs Carreno Busta. Sure, he probably would have turned it around but we don't really know. Thiem was always going to be a big challenge as well!
I stand by the notion (especially after that US Open match with Serena) that even if the technology is inaccurate, human line calls are just far more inaccurate to leave millions of dollars on the line for.
They will be innacurate by a small margain, yet consistent. That consistency alone would quash on court umpire drama
@@M3DIT4TE good point as well
agreed!
Exactly. People get furious when a human being with a fallible set of eyes and mind get a call wrong of whether a ball going 100 miles per hour was in or out with a margin of millimeters. Then, when you implement a computer system that is right 99% of the time and makes one slight miscalculation, they get furious again. I'll take almost never wrong over people who are wrong quite often lol.
You still win big coming 2nd... look at golf, come 4th with a missed putt that would have put you 1st? No biggie, prizemoney is still huge.
Having watched my first lot of tennis with automatic line calling at the Australian Open, I have to say that I found it a relief as the game flows so much better without the constant challenges and their delays.
Yeah the line judges should be removed completely
It helps Americans players as a lot of umpires have shown anti American biases before. I’m not american but I want everybody treated equally so I’m pro it although I miss line judges
@@krob2327
Care to share with us any examples of your 'anti-American' biases
@@tassie7325 Davis Cup Paris 1932. This is a famous example.
Regardless of whether it is "right" or "wrong", it is still better than a human because it is consistent.
But the linesmen are consistent. Consistently bad
Chris eyyyyyyy
and unbiased
The only possible issue would be wind gust’s, but even then lines people could be mistaken as well.
I like the human error aspect of sport. Just look at the state of the premier League now and it's ridiculous by-the-book handballs and millimetre offside checks after every goal - it's a shambles
removing the line judges would've helped Djoko out a tonne
Cue scene of a future descendant of Djoko hitting a robot in the servo by accident.
Not being an idiot would have, too
He’ll still get a foul for breaking the camera
You mean ignoring the rules?
poor djoko
A computer is more accurate than a human. Yes computers make mistakes and have errors. But I would take a computer calculating where a ball lands going 100+ mph over a human any day of the week.
@Heather i think it's a matter of time until the playing population understands/trusts technology and the game shifts. 2020USO had some courts with only these cameras so it should come soon
A computer is only as accurate as the human minds who programmed it
Also computers are dispassionate and will not be phased by the importance of the call or the drama of the occasion.
@@MeFreeBee and the angry man woman swinging a bat at it 😂 although to be fair, the best news for me is that there won't be any more Djokovic incidents if line judges are replaced by computers fully
@@nick2swsThat not how it work, that like saying AI can only play Chess as well as the person program it
Some of those calls in the Williams/Capriati match were like what I see in my local doubles comp. Every ball that lands on the line gets called out. Lots of revenge calls in return!
So there is a little back story to that match. Earlier in the year the Olympics were held and several top umpires were caught manipulating their Olympic credential so as to be able to see other Olympic events besides tennis for free. They were caught and a final judgement was made during the 2004 US Open. The judgement resulted in several top officials being sent home from the US Open. (Look up some news articles about it) this on the court for that match as the chair umpire was a woman who was up and coming at the time but not ready for the big time just yet but in the vacuum left by the suspension of top officials she drew this assignment. She is the one who started everything by overruling a ball to out that was called in by the line judge and was clearly in. After the ensuing mayhem she panicked and stopped overruling anything. The line judges also got nervous and began blowing calls.
"Watch as the worlds greatest tennis player winds up for a forehand"
Shows Tomic
lmaoo
non tennis fans didn't get the joke im sure, that's the reason I left a like on this video
Cult Tennis owes me a new keyboard lol
@@chloeharvey5684 That was no joke. That was an understandable mistake. They clearly thought the guy is Djokovic, and it is very difficult to tell them apart.
sorellman you didn’t get the other jokes either then I guess
@@chloeharvey5684 Dear Jesus, what jokes? Are you a writer for Bill Maher? When he gets things right, he is a social commentator. When he makes mistakes, he is a comedian now, don't you get it? Am I supposed to see error in judgement as being a joke? I thought so, but no.
a) More accurate than humans by far.
b) Unbiased.
c) Always at full concentration.
d) Noone for players to get mad at.
The $60k price tag is prohibitive to human beings, for decent tournaments, it's affordable.
@Heyzyen Funny you should say that. Have you heard about the racist soap dispensers? Basically, the developers were racially ignorant, so they didn't test the motion sensors for non-white people, so only white people could get soap. So whenever you create something, it's possible that your own biasis slip in there. If there was a program using machine learning to get the Hawk Eye effect and it's only fed shot by men, it might not be accurate for shots by women (who are generally a bit slower but with more spin).
Never underestimate the ability of humans to create flawed stuff.
@@DrZaius3141 it calculates spin
@@DrZaius3141 yeah - but i presume hawkeye does not track the Players - i reckon it kind of ignores them, to test Hawk eye you would after all use some sort of serving machines for precision.
@@Zwangsworkaholic I should clarify: I don't think at all that Hawk Eye in its current form has any bias, I was just trying to point out how it could be feasible. Any automated system has certain parameters it's optimized for. To get back to the bathroom with another example: it's still not uncommon that mirros in women's bathrooms are placed at heights where many women can't see themselves. The height is optimized for men and noone put a second thought in it.
@@DrZaius3141 An AI trained on a limited dataset is not "sexist" or "racist" because it doesn't work properly on a completely different dataset. It's literally impossible for a computer to be sexist or racist, even if someone programmed a computer to say "women should stay in the kitchen" the computer still wouldn't be sexist, the programmer would be.
It’s not about fundamental accuracy in a vacuum, but about being better then a human. So long as a computer is even slightly better then a human it will be chosen.
In addition people are far less likely to get angry at a computer for a seemingly bad call, rather then a human.
@@The_SUN1234
1. I would definitely pay to watch robots compete in sports events
2. I wasn’t even advocating robots competing, but adjudicating.
3. What? Are you pro or anti anger? I genuinely can’t tell
What about robotic players? I mean where is the line drawn here?
You might scoff at the notion but so did people 30 years ago at the thought of robot umpires and linesmen
@@BoleDaPole i would love to see a robotic Novak Djokovic take on the real one 😂
As a baseball fan that somehow stumbled into tennis TH-cam, we absolutely need something like this already. Umpires are absolute garbage. Great video
While you are correct for calling balls and strikes-and heck, you could use hawkeye with a snickometer to get force plays and foul tips-there are some things that you’d need an umpire for. A computer doesn’t know what a balk is.
And to top it of, calling strikes requires depth, especially on sharp breakingballs.
@@GabbieGreenEyes Why does a computer not know what a balk is?
4go101 for example if someone spits on a ball or uses tar to have better grip it’s considered a balk . You can simply decide if a pitch is a pick off or a balk .
Humans have a deeper understanding in context sensitive things . For most things in baseball especially you can use modern slow motion cameras to support almost every decision making .
@@Mrterminus this is what they do right now. Albeit the umps are doing most of the things and just reviews it if a call got challenge.
7:26 world Greatest tennis player got me lmao
Went straight to the comments after hearing that 😂
Was waiting for someone to catch it!
Ikr he is da best tennis player
Haha..was it Borna Coric?
@@RamachandranS20 bernard tomic the goat
Some history from cricket's side - Hawk Eye's first use in cricket back in 2001 was also for entertainment's sake. It wasn't until 2008 i.e. after tennis implementation that it was used to make official umpiring calls. Compared to tennis Hawk Eye calls in cricket have to be more predictive. The reason is because ball never reaches or touches the surface the wicket - the surface to which Hawkeye needs to predict the path to. You can see the example at 1:54 - red is the actual movement and blue is the Hawkeye prediction line to the three yellow lines (wicket). In cricket this is called LBW or Leg Before Cricket calls.
Since its introduction it has as being quite controversial. Earlier Hawk Eye rules stated that if the distance between wickets and the ball impact was more than 2.5 meters Hawk eye cannot be used. This was very controversial because people couldn't understand why Hawk eye failed to predict distances above 2.5 meters. Aside from that Hawk Eye used to often misinterpret the path when a cricket ball was spinning. It has led to some hilariously wrong calls. You can find videos on YT about that.
While the technology has gotten better with time it is still controversial for the "umpire's call" seen in 1:54. The reason is that in cricket Hawk Eye needs to make an actual prediction. So, if a ball partially touching the yellow lines or wickets there is no 100% certainity of that reallly happening . So, if there is some degree of uncertainity the decision is more aligned to the on-field (court) umpire's call than Hawk eye's call. Some say Hawk Eye's prediction has gotten better and it is 100% reliable but rules in cricket still don't allow that.
That said, Hawk Eye had a huge impact in the way cricket is played. Cricketers used to often thrust out their legs to stop the ball from hitting the wickets. While LBW was a rule to stop people from doing that it was still a misused tactice. With Hawk Eye people need to be extra careful about using their legs and this has had a huge impact in the way cricket is played.
I feel that Hawk-Eye would work best in tandem with full replay review for visual confirmation - adding replay like in the ATP Cup this year would also help check double bounces, lets, etc. which Hawk-Eye can't do
But, which one is correct if they disagree. Unless the camera is able to run at say 1000FPS, Hawkeye is better
Depending on the angle and velocity of the ball, as well as the rotation, some balls will slightly "skid" or "skim" along the ground at the moment of impact, rather than bouncing directly up or down. This is more pronounced on grass surfaces seen in your example at 7:25. This skidmark covers a larger area on the ground than the ball bouncing directly on its axis
Dude am binge watching your entire channel, this is next level stuff!
Great minds think alike!!
Hawkeye was originally developed to determine LBW outs in cricket. The main function of the technology was to predict the ball's trajectory.
It takes the forecast approach - a lot like your daily weather report.
@ yeah... except that Hawkey predicts split seconds into the future, after having observed the trajectory for seconds.
I also watched this video.
Where is the rest of the transcript from this video?
@ bdw your so called daily weather report requires super computers to predict weather.
Players cannot argue with a computer. However, I would like to see the lines themselves use some kind of touch sensitive sensor, linked to the ball coating.
bigrobsydney that might be easy to do, like a conductive paint
I fear there would be false positives if the ball bounced close to the line and the sensors picked up the bounce of the ball even if it didn't touch the line
@@joshuas3897 they can adjust sensitivity and see how much pressure was applied to it
@@PokeLoomerTM Potentially
YES. Super frustrating that something that LOGICAL and way more accurate is still not implemented !
Worked along side Hawkeye many times. There is a LOT of work that goes into it and it's a huge team. Every single camera is configured and represented in a 3d simulation. Their smallest setup requires more people than our own crew. I highly respect their work and I understand technology can be flawed, but they are working hard to do deliver accuracy. I genuinely have nothing but respect for them and glad to work alongside them. Hoping they can bounce back after the lost of so many tournaments.
In the early '80s, Vic Braden was using hi-speed video to check the accuracy of lines calls. I seem to remember that the consensus was that human lines persons were pretty inaccurate on close line calls.
Vic’s show was awesome. He also used the videos to show that almost all players stop watching the ball at about 3 ft in front. Which makes Fed even more amazing - he watches it all the way to his strings.
Hawk Eye is just great. Very accurate, fast to implement and fun. Great addition to tennis
Damn, this is some quality youtube content!!
Thank you!
I think the problem with Hawk Eye being able to see that a ball was in when no human could have ever seen it like in the one example you showed is that actually changes what is considered in and out. Tennis players have always known these sorts of shots to be out and played accordingly. It’s almost as if the net was suddenly a centimeter higher.
Was it raised?
They should probably just trim the last mm of the line to make more "realistic" predictions/calls
The reason why at 7:30 the ball looked as if it bounced out, is because it DID bounce while it was out. The ruling being the way it was, was based on it being a grass court. The ball grazed some white painted grass before it bounced(out of bounds).
Based on that ruling, there can be slightly wider boundary margins on grass courts, especially with low-angled high-velocity shots, be it as few and far between as shots like that may happen.
yea there is no way that it being in was the fair call, it might've been the right one but not a fair one
I'm so glad that the line calling technology is being widely adopted.
I've always felt it was absurd to have all of those people staring down the lines when we know that that the human eye is incapable of judging the ball accurately at those speeds. Prior to this technology it would've made as much or more sense to simply have the chair umpire make the calls.
Hopefully linespeople will soon be a thing of the past.
why dont you play tic tac toe wityh your computer? let's see who wins? how exciting
@@indielounge1776Irrelevant. Hawkeye isn't an opponent.
"the world's greatest tennis player" about Tomic - that sarcasm took me by surprise :D kudos to You CULT TENNIS
You make an assumption. No sarcasm there. It was a mistake. Tomic and Djokovic look very much alike, especially from that angle.
I don’t see the sarcasm. Tomic is the goat
@Warrior Son I'd say is between him and Kyrgios :)
Lol. Who is Tomic
Tomic is the greatest tennis player in his family, though sister Sara ain’t too shabby. GOAT = Greatest Of All Tomics.
Now we know why Serena had a short fuse at future US Opens lol
After computers replace line judges, they will have T 1000s by the courtside in case she threatens the cameras
I really love these documentary style videos, just binge watched them all. You deserve way more subscribers than you have.
What I think is interesting about the introduction of technology into referee calls of various kinds is that it will require getting specific with rules, finally. So many professional sports, with millions and millions of dollars at stake, have vague rules that are up to the subjective judgment of refs. That's the way it has had to be to cope with the limitations of humans, but no more. We can get be specific down to the millisecond over what counts as a traveling call, then let the computers and cameras call them. We can be specific down to the precise positions of hands and distance to the ball about what counts as pass interference, and let computers and cameras call it.
keep the videos coming!
Will do!!
There is 1 major inaccuracy in this video. While Hawkeye cameras are used to view the baseline in a Hawkeye live setup, they do NOT “track footfaults as well” as stated in the video. The review official watches the baseline on a screen and if he/she sees a footfault, they hit a button and a voice says “footfault “over the PA system. So there still could be footfault controversy with Hawkeye live.
This is my new favorite channel! Great videos!
Very nice of you to say :) Thanks for watching!
Hawk-eye shows shots that are "in" that would be called "out" 99% of the time in a rec game.
because computer calculations understand that where a ball bounced from is not necessarily where the ball made contact with the ground. in other words, they can account for skid.
Wow, your videos are VERY well made -- thoroughly researched, good pace, enjoyable editing, and all the right topics that satisfy a curious itch.
Keep it up, buddy! Love your work!
Cheers from Korea
Hawk Eye operates with 10-12 "slow" cameras with 150 frames per second per camera (the video mentions 340 fps -perhaps Hawk Eye has improved). In contrast, Foxtenn ("Real Bounce" system) works with 40 high speed cameras with 2500 (!!) frames per second per camera (already used in quite many ATP tournaments). Hawk Eye interpolates/calculates/estimates the ball trajectory within the frames and shows a computer generated image on the basis of the calculations. Therefore it is sensitive to shocks like wind and trembles, etc. Its expected systematic error is 2,6 mm. Foxtenn shows the real video image of the ball's path and due to its extra fine time resolution, its systematic error is basically zero. Think for a moment to understand why the difference in resolution is so important: a ball flying at 126 km/h travels 35 cm in 0.01 second (!) or 23 cm between Hawk Eye frames. For 90 km/h it is 25 cm or 17 cm. So in the case of Hawk Eye, some wind gust after the last frame taken before landing can easily divert the ball with a few mm from the expected path.
Yes. It's a probability of in or out, not a certainty. The faster frame rate brings the system error down.
You deserve way more subscribers! Great Video!
Thanks so much!! Glad you enjoyed it
I feel that when we get to the point where we start asking questions like "what actually counts as the ball touching the ground", we may as well just flip a coin.
It's like trying to measure sprints to the millisecond, where it's just about how we define crossing the line.
If only Chris Evert had this in 1988 during her Wimbledon semi-final with Navratilova!
same for VAR and goal line systems in football ;) people were opposed strongly but there is no turning back now :)
4:48 cool animation. excellent channel you'll grow for sure. I dont even like tennis yet enjoy the content!
In cricket, even now in 2020 after 19 years of advancement, the use of Hawk Eye has limitations, and isn't as accurate due to the more inconsistent nature of a cricket pitch, the unique spin and swing nature of each ball as it is worn out through a match, and the effect of the seam of the ball in a bounce. If I recall, something like more than half the ball has to be impacting the stumps in a hawk eye replay for it to be considered confidently hitting.
This is of course a bit different from tennis where the ball does eventually bounce after the estimation, whereas in cricket it remains an estimation and is used to predict the path of a ball if there had been no other action or impedance against it.
Great vid! loved it.
Hopefully computers can save a couple dozen line judges from being threatened with execution by Serena.
Line judge jobs getting phased out though 😢
Only ball kids will be there on court now.
@@gmediagroupss they'll find an electric gimmick that traces and catches balls.... and only after the point has ended. ;)
@@fridbertaugust I do not want to see machine that uses powerful vacuum technology to suck the balls in.
@@elaineoh3984 right, tennis would loose it’s feeling. I at least hope Wimbledon would stick to tradition and have people on court and no machines.
Tennis is better with Hawkeye. Gone are the challenges and arguments with line judges. Players get to see the call immediately and then move on. It doesn’t need to be perfect, just consistent.
Wow I have never heard of that serena match they should literally get these line judges out of there
Im pretty sure one of the line judges must have had a personal vendetta with serena
unown unown they certenly made it seem like so
I genuinely don’t like and have never liked tennis or watching it but I still watch your videos and it baffles me
I did the math and on a 150mph serve Hawk eye’s camera take a picture of a frame every 8 inches the ball moves. Which is pretty large especially relative to the ball’s diameter. I think having the camera’ be better will dramatically reduce the likelihood of a wrong call
Isn't it also true, that it is still vastly superior to line judges. Plus, changing the rules to giving the benefit of the doubt the ball being out will make things a lot easier. The system is still largely superior to human judgement. I feel the data set generated can predict the parabola of a tennis really well.
Not sure about this math...how many frames per second does it take?
Nah. Predictive analytics such as counter battery fire does not work when type and amount of spin is variable.
@@Blaisem 150mph = 2640 inches per second, which at 340fps will cause the ball to move around 7.76in per frame.
You know why they don’t want to even have it on clay courts for even “fun reviews”, similar to how it was years ago, is because with the actual clay marks, you will easily see just how much room for error 2.6mm really is. That 2.6mm is 3D spacial error, which is spherical… meaning that the ball itself could flex or wobble its shape enough to throw the calculations off. A calculated “in” because of its estimation may actually be an out once it actually hits the court and vice versa. A good example of this is that hard court call that Hawkeye had wrong and you could clearly see the fuzz mark with no other ones around… and that was way more than 2.6mm off in error. If they truly believe it’s that good, let it prove itself on clay courts too. Heck, if anything else, they could use clay reality to calibrate it even more. I would put down money that it would make errors on clay that we could actually see, but obviously they don’t want to do that and lose credibility. 🙄
I never see TH-cam videos on tennis other than highlights compilations of the same old points. This is a welcome change for all tennis fans and I hope you keep making awesome videos like this!
The future for tennis is probably a combination of Hawkeye and the Foxtenn one you mentioned in the last minute. Foxtenn coupled with Hawkeye will give real data to improve Hawkeye's algorithm.
On the other hand if players feel like Hawkeye f***ed up they can use the Foxtenn footage as a challenge.
No need for combination -at least for the basic service of linejudgeing (Hawk Eye has quite many additional smart services, too). Foxtenn is simply superior in accuracy. Hawk Eye operates with 10-12 "slow" cameras with 150 frames per second per camera (the video mentions 340 fps -perhaps Hawk Eye has improved).. In contrast, Foxtenn ("Real Bounce" system) works with 40 high speed cameras with 2500 (!!) frames per second per camera (already used in quite many ATP tournaments). Hawk Eye interpolates/calculates/estimates the ball trajectory within the frames and shows a computer generated image on the basis of the calculations. Therefore it is sensitive to shocks like wind and trembles, etc. Its expected systematic error is 2,6 mm. Foxtenn shows the real video image of the ball's path and due to its extra fine time resolution, its systematic error is basically zero. Think for a moment to understand why the difference in resolution is so important: a ball flying at 126 km/h travels 35 cm in 0.01 second (!) or 23 cm between Hawk Eye frames. For 90 km/h it is 25 cm or 17 cm. So in the case of Hawk Eye, some wind gust after the last frame taken before landing can easily divert the ball with a few mm from the expected path.
Man it was hard focusing on the video after seeing the Serena vs Capriati
Your content and voice are like the Core-A Gaming of tennis. I don't even watch tennis, yet I subscribed to your channel, I don't play fighting games, but I subscribed to Core-A.
Even if it calls your ball out when it was in, it will also call it in when it was out so chances are throughout a player's career they will break even.
Good point and its not like any human could benefit from those faults, because you cant hit a ball this crazy precise.
@Vlasko60 I think Cricket matured a lot more than Tennis with usage of Hawkeye. Since it is not 100% accurate, in cricket, umpires decisions are considered for marginal decisions. Instead of showing them as hitting or missing, which is akin to in or out for tennis, they are referred to as "Umpire's Call".
This is only true if those errors apear very often. Its called the law of big numbers. Basically, if you roll a dice 6 times and track the results, it would be unlikely to be evenly distributed. But if you roll dice 60 times, 600 times or more its more likely to distribute evenly between all 6 numbers.
I love how I’ve never played tennis in my life but still binge watch this channel
Great video and explanation. A minor grammatical suggestion. Your two text "it's" should read "its" without the "' ' ."
ahhh good catch!
I’m reminded of the Agassi-McEnroe match in the Wimbledon semis in 1992, in which after the whistle for out-of-bounds kept going off at the wrong time on McEnroe’s serve, Agassi came up to the ref and said, “Turn it off. Just turn it off. This things been screwing up for the whole tournament.”
Capriati in that 2004 match against Williams was incredibly dishonest; how can you not see those balls were in and steal your opponent's points. But the look on her face during the on-court interview when faced with questions about this was very telling. She knew Serena was being cheated, but was cool with it. I suppose that's how she'd have liked it for herself.
It's not the players job to make calls, it's the ref. There's no telling that Williams wouldn't have returned the favor if Capriati had been honest, so there's no incentive for her to do a job that's not hers by making calls she thinks are in or out (especially when she may not have good vision on it either.)
@@squidge903 thats on paper.. But in Human heart there is something called sportsman ship .. when you know you don;t desrve a point you don't take it.. if u r taking just because it's given .. it may not be your fault but you are not worth it..
@@Straightforward786 It's not that simple. In reality, it's often hard to say for sure if a ball was in or not, especially as a player on the baseline with the ball coming at you fast and landing near the baseline. Sure, it's often clearly out. But by definition, it's the closer calls that line judges get wrong. So, in practice, most of the time. As a player there will be times when your opponents ball is called out but you know it might have been in. But equally, you're far from certain. The only reasonable thing to do is let the call stand unless it's very clearly wrong. And, like I said, the very clearly wrong doesn't happen much - and I don't mean very clearly wrong on the replay, I mean very clearly wrong as you see it in the real time as the player. That rarely happens because nearly all incorrect calls are when the ball lands very close to the line.
To me the fact that it predicts where the shot should go instead of where it might actually go is a problem. It might be accurate most of the time, but if there is a slight breeze down the middle of the court, it would be enough to move the ball a millimeter, which is the difference from a ball being in or out. It makes more sense to me to just have those cameras focus on every inch of all the lines (and from varying angles) to determine if it’s out or not.
That's really interesting, to be honest I thought it would be more accurate. Before Hawkeye I used to think that one day the actual lines would be "smart" with the ability to sense contact with the ball/ players feet. Thanks for posting
That's a nice idea
I love that idea! I find it interesting how different minds work. I would not have looked at it from that aspect. To me, those examples of the camera seeing the exact bounce and skid and everything would be a good idea. It is apparent that the technology exists now, but maybe those cameras have to be so close to get that level of accuracy? I don’t know, but I do like your idea
That idea is already possible today with the use of fiber optic cabling.
The key here is even if the computer is inaccurate, it is consistent. If its precision is high, that becomes the new standard and any pro athlete can and should adapt. One could argue that a line judge is less consistent due to inherent biases. Thus the precise, if not perfectly accurate, behavior of the computer system is superior to the imprecise (and still not perfectly accurate) human judge.
what about touch sensitive lines, I guess the challenge then would be to make the ball behave the same on the lines
One thing that always confuses me is that the rules of tennis state that the ball has to TOUCH the line, but hawkeye will call IN when the shadow of the ball OVERHANGS the line. Like if the very, very edge of the ball touches the line on a hawkeye replay it's "IN" despite the fact that that part of the ball can't possibly have touched the line. Doesn't make any sense to me...
This is what the MLB needs to do with Umps. Too many umps decide to make themselves the star of the game over the players on the diamond
Your channel is very underrated. You should deserve more subs👍
I like the system where the line judges male the calls, with Hawkeye being the challenge
The question is - does being "accurate" helps or hurts the spirit of the game? "You cannot be serious" moment is legendary. It would not exist if McEnroe had to argue with a computer... I'm leaning more toward being human and imperfect instead of having sterile, 100% accurate robot.
I think the way it is used now - for a limited number of "challenges" is a good compromise. Replacing human judges completely would probably be a huge mistake.
Ultimately the fundamental purpose of sport is to determine the best player, inaccurate calls are a hindrance to that.
Can’t they replace the ground lines with lasers that trigger when tripped?
Then any motion that blocks the laser would trip it. Bugs, leafs, trash, rocks... even a large enough amount of dust can trigger a laser sensor. Also, what if a player's foot is already blocking the lasers line? The laser would also have to be above the ground, not on the ground so it would actually get trigger before hitting the ground saying the ball is in when it could just barely be out.
@@TheMusicman1945 then ban leaves and bugs. Duh. Jk
I agree that lasers alone would not work but perhaps in addition to the cameras. Might be a lot of extra work and cost for little gain. I suspect you would need laser on each side in case there is a foot and a ball at the same time. AI would have to figure out which trip to use and hopefully Biggest issue trying make sure there timing is the same too because at 320 fps if they are out of sync a little bit it could cause trouble. Ball hits at .002. Was the trip at .001 the ball or something else. I think it would make system WAY more expensive and in a little if any gain. I do like your thinking behind it though.
It'd be less expensive to just install a layer of sensors in the ground.
This video is so good honestly.
Hawk eye is a gift to all other sports from the great sport of Cricket.
Line call arguments are a fun aspect of the sport. Keep line judges and hawkeye. Together they make the sport more dramatic
The issue with Hawkeye is: If it breaks during a match (which is certain to happen with thousands of matches each year), then how do any line calls get made? You would have to have some line judges on reserve in case that happens.
what do you mean "break"? how can a program break at all? i suppose game stops for a while and programmers will get it working again i think. it would not be an issue at all.
one viewer - what I mean is at some matches, players will challenge a call. Then, the umpire is told Hawkeye isn’t working. Maybe it can be fixed, but I don’t think the tournament would pause the match to fix it. No telling how long it might take to get it working again.
@@tberry79 it should be tested and optimised before using it regularly.
@@saberswordsmen1 and yet, that system would be million times better than human eye. as i said, it should be maximally optimaized before using it and i think it should work fine.
@@saberswordsmen1 yeah, maybe you guys are right, back up plan would not hurt now that i think about it.
Amazing production value and super informative. Love the sarcasm and certain "shade" throwing thrown in lol.
got to keep viewers on their toes!
@@CULTTENNIS haha no doubt. Just came across your channel and I subbed immediately. Keep doing what you're doing and this channel will grow huge. Just got done watching the Coria video. Classic footage!
Wow, as a South African im shocked an American heard about cricket!
Guessing he's done his research...
In the example at 7:50, the ball is clearly out. What, are we taking into account how hairy the ball happens to be? Also, an ‘average’ accuracy of 2.6mm doesn’t give the full picture. What is the spread of accuracy? And the worst case?
First off, well produced video.
Second:
8:43 is NOT a foot fault. Also, Serena might not have actually foot faulted. Too close to tell, but saying she would "get away" with it, is really snide of you.
I’m glad someone said this.
I agree with hawk-eye technology. It is amazing that 10 cameras are focusing on the movement of a single ball, speed and bounce, and mathematically can predict whether in or out. Other factors that influence the ball movement, such as humidity, temperature, shades and wind gusts are components of mathematical model. It is more accurate of course than human eyes, considering that players now can hit the balls with much velocity. May be putting more cameras (say 20) can achieve 100% accuracy. Cult Tennis, Great presentation. Thank you.
I wonder at what point you have to face the reality that a white line on grass is "Blurry".
When Hawkeye zooms in twice, u know it's a big one
Similar controversy to adding cameras for soccer goals. It's pretty darn accurate.
The truly weird part is incorporating the deformation of the ball into the call and ignoring that a spherical object can only make initial contact with the ground at a single point, meaning if only the deformation of the ball touches the line, the ball was in contact with the ground outside the line before that and the call should be Out.
Totally agree. In cricket it is 'not out' (umpire's call) if less than 50% of the ball is touching the stumps/bails.
Cricket balls have no deformation. In tennis, 10% would do away with stupid edge cases.
I think it is generally accepted that the interpretation of the point of contact is not the instantaneous first point of contact, but the entire area the ball contacts on that first bounce. So though the very first point of touch could be out, if the ball flattens to impact onto the line as well then it is in.
Weird tennis would set itself up like that with the impossibility of trying to "see" the complete deformation of the ball as it strikes the ground. Seems to me if any part of the ball touches out of bounds, before any part of the ball touches inbounds, the ball is out. Like in every other sport with boundaries. But then I don't play tennis. ¯\_("/)_/¯@@mattc3581
I am not a tennis fan but the quality of the video is keeping me interested. Great job!
this pandemic taught me one thing, to learn programming as soon as you can.
Best up and coming tennis channel by a mile. Keep doing what you’re doing man!
Part of the reason that Hawkeye isn't used on Clay is that the spot of the ball is visible. You'll notice that the balls always produce a small amount of clay spray every time that they hit. Watch the close ups. That clay dust can mimic the ball with Hawkeye. The ball could be a fraction of an inch out but still register as in. If they can ever come up with Hawkeye that recognizes the ball and not the clay dust then they'll use it, but I just don't see that happening.. More importantly, and I've played on clay so have seen it.
That's why they're not used on clay.
@Jurtez Lupa well I have heard that it was partially due to dust on the French Open broadcast a few times.
It doesn't really matter if it's off by 5mm or 5cm, what matters is it's unbiased towards players and off by a consistent amount. If you're not hitting the ball because you think it's going to be 1mm out of bounds, you're making a bad judgement call. You should be hitting every ball that has a good chance of being in. 1mm out of bounds is still a very well placed shot.
I would prefer an actual camera system that shows where the ball actually hit.
Here is the deal, if a ball is hit with 100 miles per hour, a camera that takes a picture 340 times per second, is only accurate for 13cm.
100 mph => 160 kmh => 44m/s => 44m/s *(1/340)s => 0.13m. Considering the two images beofre impact and after impact is pretty much what the system does, only more sophisticated.
For fun if you want to be as precise as 3.6mm you need 12356 images per second, which is bonkers. (Funny number 12345).
@@Schnorzel1337 no, that's not how it works. If you see a truck move at a steady speed from a to b, you don't need to have recorded its position at every single point to know where it would be at a certain time
@@patrickbeart7091 Good that was never the question but good. @Nazraq S suggested a "real camera setup" that shows where it actually hits, which is not possible at all. If you want to get decent results just for the camera you would need 12356 images per second so the first image where the ball touches the ground captures the ball at most 3.6mm from the initial impact. That part obviously ignores friction and everything else.
@@Schnorzel1337 ahh I see, sorry for the confusion
@@patrickbeart7091 For context if you had a truck moving at 100 mph then the truck would be blurred across in the direction of travel, making the truck not only appear longer than it actually was, it also blurs the edges and distorts it.
Looking at a tennis ball moving at 100mph at 300fps looks like a pill, when it hits the ground it deforms and skids and all of those actions are, consistently, declared as being part of the contact and determines whether it's in or not.
To achieve such a system you'd need cameras that either play at insane framerates, or you'd have to combine THIS system to predict where it would hit with a camera capable of a 10000+ of a second exposure time to achieve a "picture" of the ball actually hitting, which adds another problem, the 10000+ frame time would likely need to be a series of shots to get all of the frames where it's contacting. So, no, a system that actually monitors the contact of the ball is just a technological hurdle that nobody is really willing to foot the bill for.
I have never ever considered hawk eye to be wrong. It would be highly audacious and stupid of me to sit infront of my screen, consider a ball out, have hawk-eye disagree, and then say hawk-eye was wrong... the truth is unless there's a great replay, its impossible to tell many times.
But people really do underestimate a ball brushing the surface and landing beyond the line, which was greatly exhibited in the Tomic example.
All in all hawk-eye has a what, proposed 99.7% accuracy? Far greater than the human eye, OBVIOUSLY. And its few malfunctions of not working are one out of million. In fact, ive only seen the machine not work entirely once (Azarenka example). The other times it just couldnt appear as visual.
HAWKEYE should be used on clay, just Google the amount of controversy from players that's disagree whether the ball clipped the line.
As the game becomes more precise maybe people will learn the tendencies of what they can get away with on each court and change their strategy with that knowledge
The only question to ask is "What is the most accurate calling system, the human eye or the electronic Hawkeye?" Since the whole point of calling balls in tennis is to determine which hit is in or which is out, then whichever system does this best is therefore best for the game. Make the accuracy determination and you have your answer.
I'm appalled at the overall quality of this discussion thread, so far as I read. Integrating multiple observations of a moving body in a clever way goes all the way back to Gauss, who invented least squares for the purpose.
The real issues here are optical distortion in the camera lens, imperfections in the court surface (e.g. drainage curve), synchronization of the camera feeds, distortion of the shape of the ball after impact with the racquet (which likely oscillates in the air for a few meters as it sorts itself out), non-uniformity of the ball (e.g. moisture off the court surface, scuff marks), motion blur in frame capture, variable illumination, optical turbulence (which also affects human vision), and flight turbulence with shifting and non-uniform wind patterns (which the players can't predict with precision, either).
Human vision is far from infallible concerning all these variables, either.
The main difference is that a human judge on the court is that the human judge is given a preferential viewing angle to final impact, whereas the camera system has far more eyes working in harmony, at a far higher speed, combined with a rigorous application of linear algebra to minimize the error term in the coherent reconstruction of the _entire_ flight path of the ball. The human brain would incorporate more Bayesian analysis based on all distributions and patterns in all the balls previously witnessed (spanning the entire range of variables) producing the kind of accuracy we're used to having, for what it's worth.
However, the problem with the human mind is that it's not correlation-free with respect to random, recent events. If you fought with your spouse the previous night, you get a different error function. You get caught up in the emotions of the match or the day, you get a different error function. You mentally remind yourself to factor these things out so as to remain maximally impartial, you get a different error function (which might be less biased, or it might not). If you're still kicking yourself at a bad call three points ago, you get a different error function.
There's _heaps_ of psychological research showing that we're rather bad at calibrating to a fully unbiased center. We're so bad at this, that we've learned to substitute the _performance_ of an unbiased arbitrator for the real thing. In almost all sports, referees keep a mental account of which party _appears_ to have received the short end of the stick on previous judgement calls, and then subtly tilts the scale back toward restitution.
The NHL is terrible for this, because it's a target rich environment. There are technical obstruction penalties on nearly every hard-fought play. Awarding a 2-minute powerplay is a big deal and easily impacts the final score. Awarding a slew of 2-minute powerplays back and forth is just as bad, even if completely balanced, because it changes the sport from 5-on-5 to alternate 5-on-4, suitable to a completely different set of players (often one team is better 5-on-5 while the other team prefers alternating 5-on-4 and 4-on-5). There's an entire unspoken science of "game management" on the part of the referees (usually unspoken, except for now and again an embarrassing open mic). It's such a dark and unseemly underbelly that the league hands out $10,000 fines like dinner mints to coaches who state the obvious after the game.
In tennis, the players fall under the known illusion of control. They don't actually know the minor details of wind turbulence on the court, which easily affects their shot placement on the centimeter scale. When they hit that close to the line, it's no longer skill, it's luck of the gusting breeze (or tiny convection cell on the court during a hot, windless day). If their balls are being called out on less than a 5 mm margin, they should simply be hitting their balls a few cm shorter on average (or they should simply accept that this is a percentage situation, and you can't win them all).
What finally matters is that machine vision is repeatable and impartial for the vast majority of common circumstances. This has been achieved already to a very high standard.
The few exceptions to the exceptional performance of this system are hardly changing the long-term outcome of tennis as a sport. If the two players are so evenly matched that one player prevails over the other only because one player was on the right side of more 5 mm line calls, the idea of a final winner and loser is more of a coin flip than a demonstrated superiority.
The entire postmodern culture wishes not to believe that "equity" often hangs in a highly material way on the last decimal point. But the truth of the matter is that all systems with chaotic elements are capable of going left or right on a decimal point far beyond human apprehension. It's a fundamental property of chaotic systems: somewhere in the configuration space, left and right are jammed together so tightly that no possible perception can pry them apart, except to sit back and watch what happens. The infinite recursive pattern of the Mandelbrot set is trying to tell you something, if you're more into telling than yelling (at his level of nuance, most of the population sides with yelling).
Even the most elite tennis athlete needs to give it a rest in the implied complaint that they can hit a ball 100 MPH in outdoor conditions to the accuracy of a quarter of a beer cap twenty-five yards away. They are psychologically wired to want (and need) to believe this, but it's simply not true.
Finally they need to chill out, STFU and accept an impartial outcome from an electronic deity of linear algebra as implemented to this exotic standard.
When ever I’m watching a tennis match and there’s a challenge I always try to guess if the balls in or out and I’m right about 30 percent of the time lol 😅.
I think it's basically reliable. I am a bit sceptical when there is a camera pointed right at the spot where it hit but they won't show that shot and only show the Hawkeye generated animation. The Hawkeye camera is around 350 fps, but there are video cameras that can shoot up to 1,000,000 frames per second. In matches where multiple people are seeing Hawkeye being off, It should be an option to have 1 super camera verifying it's accuracy. This would be a whole process of what shot to measure, what angle would give the least optical distortion etc...so don't just say it can't be done without justification.
The issue with replacing line umpires at top tournaments is that you remove a lot of the incentive for people to get into umpiring in the first place. The unintended consequence of that might be a shortage of umpires at lower levels of tennis.
Very like postmen, of the kind of jobs that makes no sense anymore.
Bro, you’re videos are really good, you speak well and you’re able to make it not boring, nice editing and nice “hosting”! Keep up man, loving ur channel
Please do a video on Serena Williams!
I would love to see that
they might as well build some touch sensitive lines to accompany hawk eye, as the lines are where the arguments stem from. that way you get your 3d hawk eye visualisation and also the line sensors tell you if they were triggered, in case its very close
7:25 you just made yourselfe some enemies mate
Bernie, top 3 tennis player of all time.
7:27 As commented previously.. this is Bernard Tomic, not Djokovic.. Djokovic was never sponsored by Nike.. Shoes always Adidas, clothes Sergio Tacchini, then Uniqlo and now Lacoste.. and Bernard Tomic has one of the weirdest forehands btw
LMAO ARIGHT 5:45
7.25 heres a footage of the worlds greatest tennis player ..... wait whaaat??? Bernard tomic🤣🤦♂️
Lukas Helmhart
Djokovic, Federer and Nadal on suicide watch,
As soon as the naked eye could see on TV the inaccuracies of the calls of the naked eye on court between Serena and Jennifer, it's a no brainer to get a system that would contest the call on court, which is the least accurate.
8:48 The woman later admitted she might have "misheard" Serena sooooooo
Serena threatened to shove the ball down her throat, this can be heard if you watch the interaction in full. Wasn’t a threat to kill, but close enough
"The woman later admitted she might have misheard Serena for proper amount of money." (I fixed it for ya)
Santi Dieguez it’s not close. At all.
Michał no.
The linesperson told the umpire and the supervisors that Serena said "fuck you" (She didn't actually, she said "you better be glad that I don't shove this ball down your fucking throat"). When Serena heard the linesperson repeat this, she misheard her and exclaimed "I didn't say I would kill you, are you serious?!" That's when one of the supervisors brought her in close to say that the linesperson was actually alleging that she said the f-word, and you can see her admitting to it before finding issue with the course of action taken. All of this clearly recorded and most of it is audible with the on-court mics so it's shocking that it still gets routinely misreported to this day.