Qlipp v Sony – Are Tennis Sensors Accurate?

Sony Smart SensorQlipp SensorTennis sensors are a relatively new addition to tennis. The Qlipp Sensor is the latest to hit the market while the Sony Smart Sensor has been around a little longer. We decided to get a handle on whether the output from these devices is accurate.

Why Qlipp and Sony and not the Others?

Two simple answers: Session Log and Sensor Location.

One thing that’s really impressive about both the Qlipp and Sony sensors is that they allow you to zoom in on one shot (or a group of shots) within a larger time period e.g. a hit session or a match.

Both the Babolat Play racket and the Zepp Tennis Sensor show a session log as a whole heap of averages. This can be valuable over time but the future of this market is in the ability to allow the user to zoom in on a time period or individual shot.
We used a Yonex Ai 98 tennis racket for our Qlipp v Sony sensor test
The second reason is that the Qlipp device is located in the string bed, while all the others obtain their data from the butt of the racket.

Equipment

We used a Yonex Ai 98 racket strung with Technifibre Razor Code. Balls were fed from a Tennis Tutor Ball Machine in oscillation mode.

Method

Pretty simple really. Hit 100 forehands and backhands. A large percentage of shots were played with topspin. There were a few slice forehands and several slice backhands. Overall, a good mix of shots that you would expect during a normal hit session or match.

For this test we were only interested in shot and spin identification. That is, whether the sensors correctly identified a forehand or backhand, as well as whether slice or topspin was used. We’ll worry about serves, volleys and other stuff in coming weeks.

Note that Qlipp has a “flat” reading for spin. We’re not fans for reasons contained in this article. For the purposes of this exercise, however, we forgave the flat reading. We would agree that varying degrees of spin were used for each shot, but it was always topspin or slice. Flat readings were not counted as mistakes.

The video capabilities of the sensors were not used. The video you can see is straight from an iPad. Beside the iPad, we placed an Android phone which was connected via Bluetooth to the Qlipp Sensor. This meant that the automated voice calling out numbers is the Qlipp app calling its ball speed for each shot.

Qlipp and Sony apps showing different readings for shot 89This was done for two reasons. Firstly, Qlipp has no internal memory so data has to be captured live. The other is so that we could identify any shots that may have been missed and/or appeared in the session log but weren’t called by the app (both happened).

Ball speeds were called out in kilometres per hour.

The Video

The video can be seen on this page or on Youtube. As the video runs through it will run in slow motion each time the Qlipp Sensor makes a mistake. The output of both the Qlipp and Sony apps will be shown to illustrate the mistake made by the Qlipp sensor.

We’ve identified the shot played and the spin used on the Sony app by using the curtain function at the bottom of screen. All shots are color coded. In the example shown to the right, the shot is an orange dot, which is a topspin forehand. Qlipp shows the same shot as a serve.

 

Results

The table below shows our session log as identified from the video.

All incorrect shot readings or incorrect spin readings are highlighted in red. Correct readings are in green. All the flat readings have no background.

Overall, Qlipp was correct 69 out of 100 times. Sony got absolutely everything right. There’s no simpler way to say it.

If you’ve read our review of the Qlipp sensor, you will know we are fans of the concept for many reasons. The problem is, at this stage, a concept is all this device is. Also, this isn’t an isolated problem limited to this hit session or this particular Qlipp device. We’ve tested 3 different Qlipps with different players and different rackets, all with similar accuracy results to this one.

70%-odd may not sound too bad. For a prototype in development, it’s a good start. But if around 1 in 3 shots are misreads, the whole data pool is out by such a margin as to make the illustrations inside the app worthless. The app can only be as good as the data being fed to it.

Until the accuracy issues of the Qlipp sensor are resolved, Sony Smart Sensor is the only sensor worth owning.

Shot #ShotSonyQlippSpinSonyQlipp
1ForehandForehandForehandTopspinTopspinTopspin
2BackhandBackhandBackhandTopspinTopspinFlat
3BackhandBackhandBackhandTopspinTopspinFlat
4BackhandBackhandBackhandTopspinTopspinTopspin
5ForehandForehandForehandTopspinTopspinTopspin
6ForehandForehandForehandTopspinTopspinTopspin
7ForehandForehandForehandTopspinTopspinTopspin
8BackhandBackhandBackhandTopspinTopspinFlat
9BackhandBackhandBackhandTopspinTopspinTopspin
10BackhandBackhandBackhandSliceSliceSlice
11ForehandForehandForehandTopspinTopspinSlice
12ForehandForehandForehandTopspinTopspinTopspin
13ForehandForehandVolleyTopspinTopspinTopspin
14BackhandBackhandBackhandTopspinTopspinTopspin
15BackhandBackhandBackhandTopspinTopspinTopspin
16BackhandBackhandBackhandTopspinTopspinTopspin
17ForehandForehandServeTopspinTopspinTopspin
18ForehandForehandForehandTopspinTopspinSlice
19ForehandForehandVolleyTopspinTopspinTopspin
20BackhandBackhandBackhandTopspinTopspinTopspin
21BackhandBackhandBackhandTopspinTopspinTopspin
22BackhandBackhandBackhandTopspinTopspinTopspin
23ForehandForehandVolleyTopspinTopspinTopspin
24ForehandForehandForehandTopspinTopspinFlat
25ForehandForehandForehandTopspinTopspinSlice
26BackhandBackhandBackhandTopspinTopspinTopspin
27BackhandBackhandBackhandTopspinTopspinTopspin
28BackhandBackhandBackhandTopspinTopspinTopspin
29ForehandForehandForehandTopspinTopspinTopspin
30ForehandForehandForehandTopspinTopspinTopspin
31ForehandForehandServeTopspinTopspinSlice
32BackhandBackhandBackhandTopspinTopspinTopspin
33BackhandBackhandBackhandTopspinTopspinTopspin
34BackhandBackhandBackhandTopspinTopspinTopspin
35ForehandForehandForehandTopspinTopspinTopspin
36ForehandForehandForehandTopspinTopspinFlat
37ForehandForehandForehandTopspinTopspinSlice
38BackhandBackhandBackhandTopspinTopspinFlat
39BackhandBackhandForehandTopspinTopspinTopspin
40BackhandBackhandBackhandSliceSliceSlice
41ForehandForehandForehandTopspinTopspinTopspin
42ForehandForehandForehandTopspinTopspinTopspin
43ForehandForehandForehandTopspinTopspinTopspin
44BackhandBackhandBackhandTopspinTopspinSlice
45BackhandBackhandBackhandSliceSliceSlice
46BackhandBackhandBackhandSliceSliceSlice
47BackhandBackhandForehandTopspinTopspinSlice
48BackhandBackhandForehandTopspinTopspinFlat
49BackhandBackhandBackhandSliceSliceFlat
50ForehandForehandForehandTopspinTopspinTopspin
51ForehandForehandForehandTopspinTopspinTopspin
52ForehandForehandServeTopspinTopspinTopspin
53BackhandBackhandBackhandTopspinTopspinFlat
54BackhandBackhandBackhandTopspinTopspinTopspin
55BackhandBackhandBackhandSliceSliceTopspin
56ForehandForehandForehandSliceSliceTopspin
57ForehandForehandMissedTopspinTopspinMissed
58ForehandForehandForehandTopspinTopspinFlat
59BackhandBackhandBackhandTopspinTopspinFlat
60BackhandBackhandServeTopspinTopspinFlat
61BackhandBackhandBackhandTopspinTopspinFlat
62ForehandForehandForehandTopspinTopspinFlat
63ForehandForehandServeSliceSliceFlat
64ForehandForehandForehandTopspinTopspinTopspin
65BackhandBackhandBackhandTopspinTopspinSlice
66ForehandForehandForehandTopspinTopspinTopspin
67BackhandBackhandForehandTopspinTopspinFlat
68BackhandBackhandBackhandTopspinTopspinFlat
69BackhandBackhandBackhandTopspinTopspinFlat
70ForehandForehandForehandTopspinTopspinFlat
71ForehandForehandForehandTopspinTopspinFlat
72ForehandForehandForehandTopspinTopspinFlat
73BackhandBackhandMissedSliceSliceMissed
74BackhandBackhandForehandTopspinTopspinTopspin
75BackhandBackhandBackhandTopspinTopspinSlice
76ForehandForehandForehandTopspinTopspinTopspin
77ForehandForehandForehandTopspinTopspinFlat
78ForehandForehandForehandTopspinTopspinTopspin
79BackhandBackhandBackhandTopspinTopspinFlat
80BackhandBackhandBackhandTopspinTopspinFlat
81BackhandBackhandForehandSliceSliceSlice
82ForehandForehandForehandTopspinTopspinFlat
83ForehandForehandForehandTopspinTopspinSlice
84ForehandForehandForehandSliceSliceFlat
85BackhandBackhandBackhandTopspinTopspinSlice
86BackhandBackhandBackhandTopspinTopspinFlat
87BackhandBackhandBackhandTopspinTopspinTopspin
88ForehandForehandForehandTopspinTopspinFlat
89ForehandForehandServeTopspinTopspinTopspin
90ForehandForehandForehandTopspinTopspinTopspin
91BackhandBackhandBackhandTopspinTopspinSlice
92BackhandBackhandServeTopspinTopspinTopspin
93BackhandBackhandBackhandTopspinTopspinTopspin
94ForehandForehandForehandTopspinTopspinTopspin
95ForehandForehandForehandTopspinTopspinTopspin
96ForehandForehandForehandTopspinTopspinFlat
97BackhandBackhandServeTopspinTopspinTopspin
98BackhandBackhandBackhandSliceSliceSlice
99BackhandBackhandBackhandTopspinTopspinFlat
100ForehandForehandForehandTopspinTopspinTopspin
AccuracySony: 100/100Qlipp: 81/100Sony: 100/100Qlipp: 84/100
Overall AccuracySony: 100/100Qlipp: 69/100

 

4 Replies to "Qlipp v Sony - Are Tennis Sensors Accurate?"

  • Eric
    September 14, 2016 (12:53 am)
    Reply

    Great job on this comparison. I own the Sony, Babolat Pop and as of a couple days ago the Qlipp. I like the reporting of the Sony App, and after some modifications, I was able to mount it on a babolat racket. Unfortunately, the Sony comes off in my hand while playing. So I bought the Qlipp with great hopes, but am a little disappointed after seeing this report. Hopefully they continue to tweek the sensor and provide updates to get better results.
    Thank you for your efforts.

    • Nick
      Nick
      September 29, 2016 (10:48 pm)
      Reply

      Hi Eric, I really like the potential of the Qlipp and have said as much on numerous occasions. Have to agree with you on the Sony as well. While the data produced is accurate, the positioning of the sensor in the cap isn’t ideal. For those with a little overhang on the grip, especially on serve, it unfastens itself far too easily.

      Thanks for your feedback,
      Nick

  • Martin
    September 24, 2016 (2:45 am)
    Reply

    Good comparison review and great test methodology.
    I like the Qlipp mounting method over the Sony but I would not accept the error rate of the Qlipp having been made aware of it now. I am a future buyer of this tech (obviously looking around now) and these are the sort of benchmark tests that help a lot with selecting the right unit.
    A feature I’d like to see in one is realtime spin rate voice feedback similar to the shot speed voice. I think it’s harder for a player to see spin differences more so than ball speed.

    • Nick
      Nick
      September 30, 2016 (5:09 am)
      Reply

      Agree on all counts Martin. What would be good is if the voice feedback could be user defined. That is, if the user would like the spin rate, select that. If you’d prefer the sweet spot measurement, select it. I wouldn’t think it would be too hard to vary the voice output. But that’s a far off possibility for the Qlipp at this stage. The data has to be right first.
      Thanks for your feedback,
      Nick


Got something to say?

Some html is OK