Tennis Warehouse: Zepp Tennis Sensor Playtest

[julian]

Protocol

So, I got the new version. Here is the "improved" 3D serve feature:

How can they get it so wrong??

Good morning,

I am not sure about a protocol to deal with Zepp Labs.
If a bug is reported do you expect an e-mail stating that it was fixed?
Is it possible that the bug reported by you is still waiting to be fixed?
Maybe a fix "for you" did not make a release?

I have clicked:
https://play.google.com/store/apps/details?id=com.zepp.ztennis&hl=en

Please go to the bottom of the page.

It says:
"What's New
Improvements to swing capture algorithms for improved serve and ball spin tracking. Also, fixed crash bugs."

Please see as well
https://itunes.apple.com/us/app/zepp-tennis/id738455328?mt=8

The version "reported" is 1.2.2

I have tried to compare your picture above with your picture in post #314

Did you get any numbers for spin?

 

[gzhpcu]

Spin numbers are better now. Had 6/10 for this one.

Zepp has a support page where you email your question. They say thanks they will work on it...

If you look carfully at the graphic, it looks like an underhand serve....

 

[gzhpcu]

I am beginning to think Zepp bit more than they can chew. Tennis serve swing is not a golf swing...

 

[julian]

Holding racket steady 1-2 seconds before a serve

Gzhpcu,
I receive the following note from Zepp Labs
----->
Currently with the 3D serve mode users are recommended to hold their racket steady for 1-2 seconds prior to use. This will allow the sensor to have a starting place for the swing it is going to capture.
------>
I have thought it maybe of interest to you
Regards,
Julian

 

[gzhpcu]

The latest app update was an algorithm update targeted toward better tracking in Play Tracking mode for serves vs. forehands. Thank you for including the picture as well as the video, I am unsure why the device is registering what looks to be a serve of a lefty, can you confirm whether you ever switch the settings in your app (possibly to let a friend use it) to a left handed user? Also, when serving to registering higher quality data we recommend holding the racket still for 1-2 seconds prior to serving. This allows the sensor to get a sense of where the starting point is with a users serve.

Zepp emailed me this today...

 

[julian]

Bouncing Matilda?

Zepp emailed me this today...

It is NOT obvious whether bouncing a silly ball is allowed

 

[gzhpcu]

It is NOT obvious whether bouncing a silly ball is allowed

I would think you can bounce the ball, but you have to keep the racket steady afterwards...

 

[gzhpcu]

OK, tested as recommended, looks much better. Next week will test if the tracking mode captures topspin on the forehand better...

 

[julian]

Any chance of a snapshot of today?

OK, tested as recommended, looks much better. Next week will test if the tracking mode captures topspin on the forehand better...

A little snapshot,please

 

[julian]

Zepp support

OK, tested as recommended, looks much better. Next week will test if the tracking mode captures topspin on the forehand better...

It would be a good idea to check with Zepp whether there are any EXTRA
recommendation related to testing topspin on the forehand side.
I did NOT think about "1-2 seconds wait rule" before
so maybe there are some extra rules in the case of testing forehand.
As you probably know from my E-mail a snapshot od www.tennisnow.com
got disqualified by the support of Zepp is done/performed incorrectly.
Julian W.Mielniczuk
USPTA pro
Catholic Memorial head tennis coach (boys')

 

[gzhpcu]

A little snapshot,please

sent you an email...

 

[julian]

the paper by Elliot

sent you an email...

hi,
I have found by accident today a paper by Elliot
https://secure.ausport.gov.au/__dat...380883/No._27_Stroke_Production_in_Tennis.pdf
please see page #12
If you have some problems with an access please do as follows:
google "elliot serve secure"
google will produce a list of documents
click a document described above
If you know how to insert Figure 5 I would appreciate it very much.
There is a logical connection between this paper and YOUR trajectory (sent by E-mail).

Switching gears:

couple of comments about YOUR trajectory provided in your E-mail of 12:28 of EST of July 17th:

1.Problems of your post #401 are gone

2.The trajectory fits "the screen"

3.The contact point is described by 2 parameters:

a.the height
b.how much "inside" the contact is

Basically I cannot "complain" to Zepp that trajectories are too low.
My very rough measurement is that your contact point is around 8 feet high when using your trajectory.

When using iyour video your contact point is around 9 feet.
I believe that ONE FOOT got "lost" somewhere

4.I like the spin 3/10 much more than 10/10
(btw: did you get a copy of my E-mail to sent to the Zepp support on the subject of 10/10 spins ?)

5.the number 60 KILOMETERS per hour is too low

6.I need to check against your video whether you are leaning inside of the court

7.all trajectories produced at my club do have incorrect contact points outside of a court.

Please let me know whether you have any questions/comments about my analysis

 

[julian]

Comments about a Zepp sensor

please see comments below the video
https://www.youtube.com/watch?v=V4argW5PQ20

 

[gzhpcu]

please see comments below the video
https://www.youtube.com/watch?v=V4argW5PQ20

Note, however, that the comments are 7 months old...

 

[gzhpcu]

There is a discrepancy in the speed given by 3D Serve and Tracking Mode. My average serve speed on tracking mode is in the high 80's in 3D serve mode in the 50-60's....

 

[julian]

Zepp Support to the rescue?

There is a discrepancy in the speed given by 3D Serve and Tracking Mode. My average serve speed on tracking mode is in the high 80's in 3D serve mode in the 50-60's....

Please try to talk to them.
I am on a tennis court
Too noisy to think

 

[gzhpcu]

There is a discrepancy in the speed given by 3D Serve and Tracking Mode. My average serve speed on tracking mode is in the high 80's in 3D serve mode in the 50-60's....

Sorry, I made a mistake:

Tracking mode gives a percentage, and here I am in the high 80's. High 80% of a pro player.
In 3D Serve mode it says my serve speed is between 60 and 80 mph.

The first figure seems too high, the second too low. Told Zepp...

 

[julian]

Is Zepp unhappy?

Sorry, I made a mistake:

Tracking mode gives a percentage, and here I am in the high 80's. High 80% of a pro player.
In 3D Serve mode it says my serve speed is between 60 and 80 mph.

The first figure seems too high, the second too low. Told Zepp...

I do not think Zepp will be happy about it

 

[julian]

A number from a radar gun?

Sorry, I made a mistake:

Tracking mode gives a percentage, and here I am in the high 80's. High 80% of a pro player.
In 3D Serve mode it says my serve speed is between 60 and 80 mph.

The first figure seems too high, the second too low. Told Zepp...

A number from a radar gun would help negotiating with Zepp

 

[gzhpcu]

A number from a radar gun would help negotiating with Zepp

Zepp should be testing themselves... Surely they have enough funds...

 

[julian]

Pptential speed of a ball served

please see
http://www.scribd.com/doc/234315414/The-Science-of-a-Drive
The formula with a coefficient of restitution is applicable to tennis
See as well
http://www.physics.usyd.edu.au/~cross/PUBLICATIONS/12. Unhappy.PDF
Ps
The word "potential" is mispelled in the subject line.
I cannot fix it-sorry

 

[gzhpcu]

please see
http://www.scribd.com/doc/234315414/The-Science-of-a-Drive
The formula with a coefficient of restitution is applicable to tennis
See as well
http://www.physics.usyd.edu.au/~cross/PUBLICATIONS/12. Unhappy.PDF

I think there are lots of parameters to be considered for an accurate measurement:

- racket head speed
- racket weight
- degree of racket stiffness
- type of tennis ball used
- how strongly the racket is gripped at impact
- type of string
- string tension

Anything I missed?

 

[julian]

Ball speed

I think there are lots of parameters to be considered for an accurate measurement:

- racket head speed
- racket weight
- degree of racket stiffness
- type of tennis ball used
- how strongly the racket is gripped at impact
- type of string
- string tension

Anything I missed?

I have a very loosely related comment-
I believe Zepp is using the following formula
A potential ball speed is equal 1.33 of a racket head speed.
Please note that there is no space for different parameters here.
I could be wrong about the conjecture above but I have though it was worth mentioning.
I am not sure whether there is any polite way of asking Zepp whether I am right or wrong.

 

[gzhpcu]

I have a very loosely related comment-
I believe Zepp is using the following formula
A potential ball speed is equal 1.33 of a racket head speed.
Please note that there is no space for different parameters here.
I could be wrong about the conjecture above but I have though it was worth mentioning.
I am not sure whether there is any polite way of asking Zepp whether I am right or wrong.

Don't think so.

Some examples from serves (3D serve) measured:

racket speed........ball speed potential.......factor
70.............................90........................1.38
60.............................80........................1.33
54.............................70........................1.29

 

[julian]

You ruined my day

Don't think so.

Some examples from serves (3D serve) measured:

racket speed........ball speed potential.......factor
70.............................90........................1.38
60.............................80........................1.33
54.............................70........................1.29

You ruined my day

 

[julian]

Physics of serve analysis by Zepp

Please see a device described by an US patent 0275459 A1

 

[gzhpcu]

Please see a device described by an US patent 0275459 A1

How?

 

[julian]

I meant please see a patent application

How?

I meant : "please see a patent application"
I was hoping thay you will google using the patent number I provided.
I have seen a device like that used for tennis but I do not have a link.
Don B. may know something on this subject

 

[gzhpcu]

I meant : "please see a patent application"
I was hoping thay you will google using the patent number I provided.
I have seen a device like that used for tennis but I do not have a link.
Don B. may know something on this subject

I did Google. No hits other than your post.

 

[julian]

Please see a document

I did Google. No hits other than your post.

Please see a document
http://www.google.com/patents/US20110275459

 

[gzhpcu]

How does this apply to the Zepp sensor?

 

[julian]

Zepp sensor

How does this apply to the Zepp sensor?

A Zepp sensor can handle serves as described by the patent.
I am not sure whether a Zepp sensor can handle correctly other serves like ones
with J hook tosses.
More coming tomorrow.
Some coaches jumping on a Babolat waggon.

 

[julian]

I expanded my post number 412

I expanded my post number 412.
Please see item number 7.

 

[gzhpcu]

Please see a document
http://www.google.com/patents/US20110275459

Sorry, but it looks like a kind of silly device to me. A long-winded description for a very banal and trite device. Still don't see what Zepp could do with this...

 

[julian]

Physics of a Zepp sensor

Sorry, but it looks like a kind of silly device to me. A long-winded description for a very banal and trite device. Still don't see what Zepp could do with this...

I realized that what Zepp does for a tennis serve is not very far away from
a golf swing (in terms of physics/mechanics/mathematics).
I believe that I have sent you an e-mail stating the post below.
However let me repeat my e-mail.

The tennis model is one of two:
a tennis ball suspended from a ceiling on a string
or
a tennis ball sitting on a stick inserted perpendicularly to a ground.

The only thing "missing" is that Zepp has to calculate a height of a "stick" or
a distance from ground in the case of a hanging ball.
In both cases a ball is stationary before a contact.

So a golf ball got replaced by a tennis ball.
A golf stick got replaced by a tennis racket.

A Zepp tennis sensor knows a location of a ball via knowledge of time elapsed
before a ball got hit.
In reality a tennis ball is not stationary but in our case it can be assumed to be
stationary
because a speed of a ball is much smaller than a racket head speed.
It is assumed that a toss is perpendicular to a ground.
I hope that this post explains my "interest" in a patented device.

I believe that a Zepp model has some limitations/assumptions.
Two of them got stated above.
Those two assumptions are the only "added" value of this post.
I am not sure whether the second one applies in the case of your toss as presented in
your video.

 

[gzhpcu]

I would have thought the Zepp sensor works as follows:

It tracks swings, not keeping all swing data. It does not know when a serve is going to be hit. It detects a serve based on the arm position and ball impact. From this it keeps the tracking position prior to the serve and the tracking data after the impact. This data is kept for the sensor to send to the app for a given serve. Keeping racket still for 1-2 seconds prior to serving, allows it to detect the correct serve starting point, otherwise it might go to far back.

 

[julian]

Please check my calculations

Sorry, I made a mistake:

Tracking mode gives a percentage, and here I am in the high 80's. High 80% of a pro player.
In 3D Serve mode it says my serve speed is between 60 and 80 mph.

The first figure seems too high, the second too low. Told Zepp...

Say that the fastest serve is 140 miles per hour for a speed of a ball.
Should we expect 112 miles per hour (not kilometers) for a potential ball speed
in the 3D mode ?
Because 80 percents of 140 miles per hour is 112 miles per hour.
Please advice whether I make any sense.

 

[julian]

Tracking mode results

Tracking mode results for forehand and backhand should depend on a speed incoming ball.
I believe it is a weakness of a Zepp sensor.
More on this subject 2morrow.
The problem does not exist for serve because a speed of incoming ball is zero or very small
(see post number 435).

 

[gzhpcu]

Zepp wrote me this in respect to the discrepancy between tracking mode and 3D serve speed measurements:

 

[julian]

I do not understand

Zepp wrote me this in respect to the discrepancy between tracking mode and 3D serve speed measurements:

I do not understand why kph,not mph
Please see post number 442 as well

 

[julian]

Doubts

Zepp wrote me this in respect to the discrepancy between tracking mode and 3D serve speed measurements:

It put a lot of doubts about usefulness of a Zepp tennis sensor for teaching.
Let me summarize three points:
1.an issue of power calculations (as above)
2.low contact point
3.a contact point outside of a court
Number 2 and 3 disagree with videos.

 

[julian]

70-80 kph?

Zepp wrote me this in respect to the discrepancy between tracking mode and 3D serve speed measurements:

Is 70-80 kps for a racket head speed or a ball speed?

 

[gzhpcu]

Is 70-80 kps for a racket head speed or a ball speed?

They say racket head speed in kph.

I tested the tracking mode in the new version today, and now it no long misinterprets my forehands as slices. Looking good.

Next Tuesday I will be training with a friend, Javier Molina, ex-ATP player (http://www.atpworldtour.com/Tennis/Players/Mo/J/Javier-Molina.aspx. I will have him hit serves with the sensor and see the 3D serve results...

 

[julian]

Do u plan to video tape Molina?

They say racket head speed in kph.

I tested the tracking mode in the new version today, and now it no long misinterprets my forehands as slices. Looking good.

Next Tuesday I will be training with a friend, Javier Molina, ex-ATP player (http://www.atpworldtour.com/Tennis/Players/Mo/J/Javier-Molina.aspx. I will have him hit serves with the sensor and see the 3D serve results...

Do you plan to video tape Molina to compare with trajectories?
It is interesting case to compare because he is 5 feet 10 inches tall,
shorter than you.

 

[julian]

Does he know his speed of serve?

They say racket head speed in kph.

I tested the tracking mode in the new version today, and now it no long misinterprets my forehands as slices. Looking good.

Next Tuesday I will be training with a friend, Javier Molina, ex-ATP player (http://www.atpworldtour.com/Tennis/Players/Mo/J/Javier-Molina.aspx. I will have him hit serves with the sensor and see the 3D serve results...

A silly question-please provide his "real speed of serve" if you have it

 

[julian]

Please send a trajectory from Tuesday

They say racket head speed in kph.

I tested the tracking mode in the new version today, and now it no long misinterprets my forehands as slices. Looking good.

Next Tuesday I will be training with a friend, Javier Molina, ex-ATP player (http://www.atpworldtour.com/Tennis/Players/Mo/J/Javier-Molina.aspx. I will have him hit serves with the sensor and see the 3D serve results...

Please send a trajectory from Tuesday

 

[julian]

TW review of a Zepp sensor published

A TW review of a Zepp tennis sensor published
http://www.tennis-warehouse.com/productvideo.html?pcode=ZEPP1
I almost had a heart attack.

A top view of a 3D serve mentioned
The word cool used many times.

Chris Edwards got 41 miles per hour for a racket head speed in the 3 D mode.
56 miles per hour of a potential ball speed.
I believe it is too low.
Majority of bloggers claim to serve 100-120 miles per hour.
What is wrong with Chris Edwards?

It begs the question whether accuracy of Zepp calculations is like 20 percents?

 

[julian]

How to use a sensor for forehands

Please see all the content of
https://www.youtube.com/watch?v=2sPpLasvPMA
The video shows three different forehands.
It is easy to identify which one is the quickest.
So there is a capability of correlating technique and racket head speed.
It is possible that maybe I am a primitive person but it is what I need for teaching.
It works as follows:
I make 3 demonstrations and I get 3 numbers from a sensor.
Hopefully an increasing sentence of numbers.
I can deal with variations and statistics if they are getting in my way.

Next I feed balls to my student to get 3 demonstrations from a student.
Next I get 3 numbers of my student from a sensor.
Hopefully an increasing sequence.

It is not obvious whether it is science-fiction or not.
I do not care about social media and numbers of Joe Shmoe from Paris.
I am trying to teach some "trends"/patterns of biomechanics.
PS I recovered a bit from my heart attack .

 

[Backhanded Compliment]

I just ordered one... hopefully I can use the same device on both apple and android devices. Curious as to how well it will work... I often do serve practice alone as a kind of meditation so even if the feedback isnt 100% calibrated correctly I will still get comparative data. Since I generally use one stick for allmatches and other matched ones for practice and as a backup this should work nice.

have they recalibrated the apps?

 

[mad dog1]

I just ordered one... hopefully I can use the same device on both apple and android devices. Curious as to how well it will work... I often do serve practice alone as a kind of meditation so even if the feedback isnt 100% calibrated correctly I will still get comparative data. Since I generally use one stick for allmatches and other matched ones for practice and as a backup this should work nice.

have they recalibrated the apps?

yes, i tried it and it works on both devices!