STABLE - Smart Behaviour Learning for Horses
| Reference number | |
| Coordinator | Högskolan i Halmstad - Center for Applied Intelligent Systems Research (CAISR) |
| Funding from Vinnova | SEK 126 665 |
| Project duration | November 2018 - September 2019 |
| Status | Completed |
| Venture | Personal mobility between societal sectors |
| Call | Funding for staff exchange and artificial intelligence (AI) |
Important results from the project
The goal was to be able to model horse behaviour and to predict anomalies. Goal was reached by using the GRAND method to model movement in similar time periods in the past. Anomalies are detected by comparing current movement to historic movement. The GRAND algorithm calculates a deviation score which can be used to generate an alarm if it falls outside the user-defined boundaries.
Expected long term effects
We expected the following results 1) A way to quantify the level of normal behaviour for a horse, and device a method to detect deviations from the normal level. This goal has been reached. 2) Provide a working prototype. The system is being incorporated into the Vidiquus system at the moment. 3) A paper describing the system and results. This is being worked on, but is challenging because it is hard to evaluate the found anomalies without expert knowledge.
Approach and implementation
The plan was to model horse behaviour using webcam images, and use an anomaly detection algorithm to detect deviations from normal behaviour. Movement data provided by Videquus AB was used to train and test the deviation detection algorithm. The movement data was generated from the images from the webcams monitoring the horses in their stables.