mmk
Hall of Fame
Sure, just ask David Goffin about the call Cédric Mourier made at Monte Carlo this year.
In/Out Device
- Thread starter Gut4Tennis
- Start date
-
- Tags
- in/out
therecanbeonlyone
Semi-Pro
I actually just got the May newsletter that they send out to people who donated. They won an award for this product at a start-up event called Vision Tank, which they seem very excited about. Also, they have settled on the hardware design and started production of the plastic parts. They say more information about the summer release will be provided in the next (June) newsletter. Color me excited.
Steve Joseph
Banned
Full of technology in all sports, which makes the sport clean by eliminating error of judgement
Karma Tennis
Legend
Maximus XXIV
Rookie
Any updates yet? I see myself as an early adopter of this.
jm1980
Talk Tennis Guru
He's gone awfully quiet
Maximus XXIV
Rookie
This is my fear. Getting from prototype to production is hard and I really want this to happen, for many reasons.
According to tweets from yesterday, no significant delays.
https://twitter.com/inout_tennis
https://twitter.com/inout_tennis
Maximus XXIV
Rookie
Good to see. Sounds like late summer though. I wonder if I can go to the front if I drive down there...
D
Deleted member 23235
Guest
thx for the update.
but sounds like, unless you were a donor (presuming kickstarter or the like)... it won't be avail to the general public until batch #2+
which is ok, probably want to wait for reviews before buying.
Maximus XXIV
Rookie
We will see. That could also be marketing to get more people to donate.
Is kickstarter the way to go about this then? Or is it the best way to keep ownership? Is the market there for something like this? I would think so, but maybe not....
jm1980
Talk Tennis Guru
He doesn't have a kickstarter... It's all on his website
I'll believe it when I see it, although summer technically started just a couple of days ago
Maximus XXIV
Rookie
I am not an expert at all, but I do not think there is pent up demand for this. No one I have talked to seems very interested. My hope is that people will see it as a tool to allow you to concentrate on play instead of where the ball is landing. This is the way pros play and I think it is a good thing.
Another issue is trust. If opponents do not want it, then it is useless for calling balls.
I am also hoping that it will help with analysis. I played with some of the sensors on the market but found them to be of little more use than gamification. The video synched up with line calls will help. Now add a synched racquet sensor and you can go right to mis hits and watch the video or out balls and watch the video. It also needs to be able to track players for doubles. A simple fob could do that.
I think there is a lot of potential but I also think it needs to be sold. I am not sure many are waiting for this to enter their lives.
Maximus XXIV
Rookie
He does seem to have a version that looks consumer ready. I think it will be late summer or fall though. Though I really do want this to happen. Maybe he will sell out to a Wilson or Head though. I could see this being beneficial for analysis in many ball sports.
I think the big thing is using tennis to get things down, because once the technology is down, there are a number of sports that could benefit from this I would think. So the market is probably larger than we are thinking.
Considering the things that get funding, I have a hard time believing this couldn't get more than a token amount, unless he's either a good techie with poor business planning skills, or he's burned through all the cash for one reason or another.
Maximus XXIV
Rookie
I am thinking more on the sales side than funding side. He is an inventor by profession and seems to be able to self fund. He does live in one of the most expensive towns in the country.
The question is will enough people but at $200 to make it a successful product. That is not impulse territory so people need a good reason. I see this as the hurdle.
JEDI MASTER
Professional
I just got an update:
July update: Production can now start!
the second week of Wimbledon. All donors will receive a notification email and one
unit will be reserved for a week. You will have to pay for the unit ($199) and your
credit card will be charged. Shipping is scheduled around US Open. Donors who will
order quickly might be able to buy a second unit. Quantity is limited because there
is way more demand than availability for this first batch
July update: Production can now start!
- Order during the second week of Wimbledon,
- MEET the In/Out device during WTA Stanford Bank of the West Classic,
- SHIP around or after US Open.
the second week of Wimbledon. All donors will receive a notification email and one
unit will be reserved for a week. You will have to pay for the unit ($199) and your
credit card will be charged. Shipping is scheduled around US Open. Donors who will
order quickly might be able to buy a second unit. Quantity is limited because there
is way more demand than availability for this first batch
Maximus XXIV
Rookie
If there is way more demand than supply, how will anyone be able to order a second unit? It sounds like he is over promising or trying to increase revenues. They also seem to not be manufactured yet, so production capacity may be negotiable. I do not really like how wishy washy this all is.
eelhc
Hall of Fame
Well 2 units can be used in concert for Doubles. The guy seems legit so my guess is that he is "rewarding" the early adopters who seeded his tooling with $25 a favor. I'm one of these guys and I will probably order 2 units.
Maximus XXIV
Rookie
I will buy one and make sure it works well enough before getting a second. I have too many tennis tech items in my trash heap to jump that far on the wagon this early.
Maximus XXIV
Rookie
Based on a reply from them, it seems that non donors will be able to order in Q4 as well.
Maximus XXIV
Rookie
Did anyone order last week?
eelhc
Hall of Fame
yep...
therecanbeonlyone
Semi-Pro
I did
JEDI MASTER
Professional
yup....
Tennis_Monk
Hall of Fame
Yes
tennisaddict03
Rookie
I'm so excited to receive it!
pnw_hardcourt
Rookie
I have a pre-sale, just paid for it the other day.
What I'm really excited for is the stats - to see what percentage I hit shallow, deep, down-the-line, or cross-court etc. Combined with the ball trajectory info like speed, spin, and net clearance, I can finally analyze my game properly and do some coaching of myself to improve my strategy etc. This is impossible to do with every other tennis sensor because they don't have court awareness!
Heck even from a spectator's standpoint. If my local tournaments had them, I could see stats like total points won, % first/second serve etc and it would be soo much more interesting than just looking up who won/lost whatever round..
What I'm really excited for is the stats - to see what percentage I hit shallow, deep, down-the-line, or cross-court etc. Combined with the ball trajectory info like speed, spin, and net clearance, I can finally analyze my game properly and do some coaching of myself to improve my strategy etc. This is impossible to do with every other tennis sensor because they don't have court awareness!
Heck even from a spectator's standpoint. If my local tournaments had them, I could see stats like total points won, % first/second serve etc and it would be soo much more interesting than just looking up who won/lost whatever round..
West Coast Ace
G.O.A.T.
I hope this pans out - great idea.
The tolerance does make me wonder if some people will refuse to allow it to be used to call shots in matches. And maybe a 2nd/3rd version, with better cameras, will improve the accuracy to the point of being almost as good as Hawkeye.
Like others, I'm more interested in the stats to see where my shots are landing. Will be great to see how minor changes in grip, stroke, etc affect the outcome of shots.
The tolerance does make me wonder if some people will refuse to allow it to be used to call shots in matches. And maybe a 2nd/3rd version, with better cameras, will improve the accuracy to the point of being almost as good as Hawkeye.
Like others, I'm more interested in the stats to see where my shots are landing. Will be great to see how minor changes in grip, stroke, etc affect the outcome of shots.
onehandbh
G.O.A.T.
I've noticed that some of my opponents' visual margin for error when calling my shots that land on the lines is +10 mm also, which means they call it out.
D
Deleted member 23235
Guest
there was a study not to long ago, tested people's ability to call lines...
people could be off by as much as 4inches, especially when you're on the other side of the court (ad side) calling a sideline on the deuce side, while still moving.
in the end it's not really about how much of a ball was in or out, it's about the consistency of the calls.
graycrait
Legend
The only reason I want one of these because in our little tennis community we have this one family (and they are "good" players, D1 kids, USTA state champs, etc) but they all call really bad line calls. When non-family members are their partners they even look askance at this particular family members' calls. Apparently "Dad" has taught all of them to call any ball in question out, unless it is their ball and then it has to be "in' and then an argument ensues. I call all questionable balls in just for the chance to take a swat at another ball, but that results in some raised eyebrows from my partners and sometimes my opponents. I want it for lines calls #1, all the other stuff #2.
ElMagoElGato
Semi-Pro
Competitive players tend to cheat a lot. They should be punished and be ashamed. I also see innocent cheaters who never can track the ball but imagine it instead. This device simply helps them.
eelhc
Hall of Fame
This is what I purchased it for and "oh by the way" it can make line calls. This thing will basically end the market for the buttcap tennis sensors.
eelhc
Hall of Fame
Here's my approach...
- I call it as I see it
- I NEVER challenge my opponents line calls. It's their call:
- If they're cheaters, challenging them accomplishes nothing
- If they're not cheaters and called it as they saw it, challenging them makes me look I am intimidating/bullying my opponent
- If I disagree, or on close calls, I may ask nicely "how much was that out by?" or "what was that? a couple of inches?". Something that establishes that it's my opponents call to make.
- I don't appreciate getting challenged on my calls. Particularly those guys who think they can see a moving ball that's 4 inches out all the way from the opposite baseline better than their opponent who is looking straight down that baseline? Once that happens it's "game on" and I get extra motivation to beat the guy.
I live in an area with a small enough tennis community so that everyone knows who the cheaters are. It doesn't help them. I would never go to a doctor, lawyer, accountant, contractor, etc, etc and privately would never recommend those people to my family/friends. People who are willing to cheat in something as unimportant as a recreational tennis match might certainly have integrity issues so that they can't be trusted in things that really matter. This includes those people who retaliate bad calls with bad calls of their own.
Karma Tennis
Legend
On courts where the ball leaves a ball mark there is a clear process that can be followed to allow a call to be queried and challenged.
I've seen many a "poor" Call (deliberate or otherwise) easily turned over thanks to a clear Ball Mark.
Of course, there is the issue of finding the right mark. But that's another story.
Last edited:
D
Deleted member 23235
Guest
regarding "call it as i see it".... this the crux of the problem.
a person "sees" in two ways... with their eyes and with their brain.
the "brain seeing" is everyone's ability to automatically fill in the gaps. that's how optical illusions work (the mind creates a perception of reality that differs from the physical reality)
basically it's a persons mind trying congruently make sense of the world.... and how it aligns with the way they think the world works
when you add in the fact that a person wants to win, and they need to a ball to be out,... guess what, their mind fills in the blanks as appropriate.
so unless there is a fair number of "i didn't clearly see a gap between the ball, i can't call it out" type calls (which always comes up at least once a set), i'm skeptical that my opponent is truly giving their best effort to distinguish between perception and reality.
best match i ever played (that i lost) was where we were basically giving each other points every set. it was high quality hitting where were overruling each other's "i didn't see it" calls, and calling the ball out on each other
another common scenario is, let's say i'm pulled wide to the ad court, return cc, and they hit dtl,... near the line. often i'm way out of position to call a ball that's even 2-3in out... so i'll defer and ask "did you see that" since they are in the better position to make the call. if they say they didn't see it, i can't call it out
Last edited by a moderator:
TennisRookie
Rookie
For machine learning to take place, the machine would have to get feedback on every call. If it gets no feedback, it won't know if it's wrong, so it can't learn. What may be possible is that the inventor trained the ai on a bunch of test cases so that the device has already "learned".
eelhc
Hall of Fame
He is getting his AI from an open source autonomous driving SW with machine learning.
https://inout.tennis/en/vision.htm
As that SW gets better, so does the line calling/match analysis.
TennisRookie
Rookie
Wait, is this device connected to the internet? Because if not, it can't really update it's software. Also, line detection is definitely an algorithm that doesn't require neural networks (it doens't need to learn), not sure about the other two.
eelhc
Hall of Fame
??? why does it have to have network connectivity to update software? I assume it is though since it connects to your phone/tablet via the app (guessing bluetooth). But even if it doesn't have connectivity, the SD card can be used to upgrade the code.
Also... I wouldn't expect a $200 device to have machine learning but the developer would be making code improvements as well as the open source autonomous driving SW.
TennisRookie
Rookie
Considering that your average smartphone uses neural networks for speech recognition, it wouldn't be too much of a shock if this also used it.
I thought that the connection to your phone/tablet was to get the data from the device, not to update the device. The reason I was talking about machine learning in the first place was because someone asked.
Phones record audio and send it to the cloud for speech recognition. Even with the network latency, that's the faster option than trying to do the work locally. Machine learning requires a lot of computational power to do effectively, and phones aren't nearly there. The In/Out almost certainly uses rules developed by machine learning to perform its analysis, meaning the AI happens elsewhere.
It's impressive enough that a $200 device can do realtime video analysis. I don't think it needs to learn to provide its value.
TennisRookie
Rookie
Sure phones send audio to the cloud for speech recognition, but that is a more difficult task for machine learning than say, recognizing a yellow tennis ball on a blue blue background.
Besides, I have used speech recognition on several raspberry pi projects, it's done locally, and it's pretty effective. And raspberry pi's have less computational powers than modern iphones. Does it work as efficiently as Siri? No, not quite, but it's definitely effective.
"The In/Out almost certainly uses rules developed by machine learning to perform its analysis, meaning the AI happens elsewhere.
What do you mean? Rules developed by machine learning? Do you mean that it uses neurons trained by machine learning? And when you say the AI happens elsewhere, do you mean that the neurons were trained elsewhere?
mmk
Hall of Fame
Unless there is some monthly subscription fee, how is that going to be done? And how would it work on an indoor court with poor/no cell connection? Or maybe I don't understand what you mean.
A neural network takes a wide range of inputs and figures out how it should weight each one to best correlate with the answer (also provided in the training data). As a very simple example, if you feed a neural network data like this:
Name,FavoriteColor,FavoriteColorIsRed
Alex,Green,No
Brian,Red,Yes
Chris,Red,Yes
...
Zeke,Blue,No
...it should correlate between "Red" in the second column and "Yes" in the third. That should result in a weight near zero for the first column and a near maximum weight on the second column, meaning the network has learned to ignore the name when determining Yes/No.
Real examples are much more complex, and are usually looking for correlations that aren't obvious. But to train an AI, you might give it thousands or millions of videos of tennis balls landing, instructing it whether each represents in or out. With the right setup, the software will figure out what properties of that video are most important to determining the difference between in and out. Since you're feeding a large number of inputs, training takes a while. However, the weighting and rules that are developed through that process can be shipped in a much simpler device, because applying fixed rules is much easier than analyzing all the data needed to develop them.
Humans can jump-start the process to speed it up. In this case, it probably makes sense to focus on the ball and its transition between downward and upward motion (i.e. the bounce). Otherwise, left to its own devices, a neural network might decide that balls land in when there are pine trees in the background, or something else you didn't notice in your training data.
