Related Experiment Video
Updated: Aug 28, 2025
07:05
Behavioral Disturbances: An Innovative Approach to Monitor the Modulatory Effects of a Nutraceutical Diet
Published on: January 3, 2017
9.0K
Identification of reindeer fine-scale foraging behaviour using tri-axial accelerometer data.
Heidi Rautiainen1, Moudud Alam2, Paul G Blackwell3
1Department of Animal Nutrition and Management, Swedish University of Agricultural Sciences, Uppsala, Sweden. Heidi.rautiainen@slu.se.
Movement Ecology
|September 20, 2022
Summary
Hidden Markov models accurately classify reindeer behavior using collar-mounted accelerometers, offering a new tool for monitoring animal welfare and environmental responses in large herbivores.
Area of Science:
- Animal behaviour
- Ecological monitoring
- Machine learning applications
Background:
- Understanding animal behavior is crucial for survival and welfare assessment.
- Fine-scale behavioral monitoring provides insights into environmental responses.
- Collar-attached sensors offer a non-invasive method for tracking animal movements.
Purpose of the Study:
- To apply supervised machine learning algorithms for classifying reindeer fine-scale behavior.
- To evaluate the performance of Random Forests, Support Vector Machines, and Hidden Markov Models.
- To assess the generalizability of these models using leave-one-subject-out cross-validation.
Main Methods:
- Utilized collar-attached acceleration sensors on 19 reindeer.
- Implemented Random Forests, Support Vector Machines, and Hidden Markov Models.
- Classified behavior into seven categories: grazing, browsing (low/high), inactivity, walking, trotting, and other.
Main Results:
- Hidden Markov Models (HMMs) successfully classified all predefined reindeer behaviors with reasonable accuracy.
- Random Forests (RF) achieved the highest overall accuracy (85%) using 5-s windows.
- HMMs demonstrated superior performance in predicting individual behaviors and identifying rare activities like trotting and high browsing.
Conclusions:
- Hidden Markov Models are effective for remote monitoring of reindeer behavior.
- This approach can be extended to other large herbivore species.
- Quantifying fine-scale behavior aids in understanding responses to environmental changes.
