Exciting new research into Icelandic gaits paves the way for a riding app that can recognise all five Icelandic gaits with up to 98% accuracy. TöltSense founder, and IHSGB member, Torben Rees has been working with Icelandic startup HorseDay and the University of Iceland to develop the algorithm. According to Torben:
“Our classifier is the first in the world to recognise all five gaits, including flying pace. With the sole exception of Equilab, which identifies tölt quite well, other equestrian apps out there can only label your activity as walk, trot, or canter.”
TöltSense itself uses data streamed from a network of sensors worn on the legs of the horse. This is the basis for highly accurate gait classification and real-time analysis. Taking this technology as a starting point the team have utilised machine learning techniques to develop a mobile phone algorithm they call the ‘gait classifier’. The group has recently submitted their research to a peer-reviewed scientific journal, and the manuscript is available here at the preprint server Research Square.
In addition, Torben has conducted a thorough validation of the TöltSense system which is available here. The research shows that TöltSense exceeds 99% agreement with a panel of independent sport judge. Also, the accuracy of hoof-on and hoof-off timings is equivalent to frame-by-frame analysis of standard video.