Detecting gaits with 98% reliability
Updated: Jun 1, 2019
Setting up the optimal training program for your horse requires you to know the intensity of a training session. For anybody not riding with heart rate equipment, the time spend in a gait plus the number of transitions can be a good alternative for calculating the intensity of training.
If you would keep track of your training efforts for 3 weeks, we might also say something about the assumed fitness level of your horse and compare this with what other riders are doing. Perhaps in the future, this will give us an idea about the optimal training program for our horses.
It all starts with riders keeping track of the time spent per gait this is the foundation of any training program. To make this as easy as possible the IPOS training application has an algorithm that can detect in which gait you are riding. Simply put your smartphone on your horse and that’s it. The app will keep track of your training program for you.
Since at IPOS we believe in science, we wanted to test (validate) our algorithm, before releasing it. We wanted to make sure that if the App detects trot you were indeed really trotting. A second question we wanted to look into is where can you put the smartphone for it to detect the gaits correct. Can you put it in your pocket? Even if you are in rising trot? What if you are lunging the horses? Can you just stick it under a girth and it will still know what the horse is doing?
Time to put the app to the test. Our intern Anke Maas has measured 5 horses ranging from 148 – 180 cm in height. The horses all performed a standardizes trial consisting of all gaits on a straight line and on the circle in both directions. The trial was repeated 3 times with the smartphone in three different positions.
Under the girth, while lunging
In a specially designed saddle pad
In the jacket of the rider
The results are amazing! The application was able to detect the gait of the horse correct in 98,2% of the time. This means that when you are riding for an hour the application will be off by maximal 1 minute. We will keep on working to further perfect this, but for now, this is a very acceptable score and definitely a work of art by our data scientists and app developer.
We found an even a higher reliability of 98,7% if you would put the smartphone in the saddle pad or under the girth. Probably because in these locations the phone is more fixated and can detect the movements better compared to when you would put the smartphone in your pocket (reliability 97,4%). The app performed evenly good riding on the straight line or in a left or right lead circle.
If you like to know more about how we have validated our Gait Detection Algorithm or you want to use the app in scientific research you can use our scientific paper as a reference:
Other references you might find interesting:
Braganca, F.M., Bosch, S., Voskamp, J.P., Marin-Perianu, M., van der Zweeg, B.J., Vernooij, J.C.M., van Weeren, P.R., Back, W. (2016). Validation of distal limb mounted inertial measurement unit sensors for stride detection in Warmblood horses at walk and trot.