High Classification Rates for Continuous Cow Activity Recognition Using Low-Cost GPS Positioning Sensors and Standard Machine Learning Techniques

  • Torben Godsk
  • Mikkel Baun Kjærgaard
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6870)


In precision livestock farming, spotting cows in need of extra attention due to health or welfare issues are essential, since the time a farmer can devote to each animal is decreasing due to growing herd sizes and increasing efficiency demands. Often, the symptoms of health and welfare state changes, affects the behavior of the individual animal, e.g., changes in time spend on activities like standing, lying, eating or walking. Low-cost and infrastructure-less GPS positioning sensors attached to the animals’ collars give the opportunity to monitor the movements of cows and recognize cow activities. By preprocessing the raw cow position data, we obtain high classification rates using standard machine learning techniques to recognize cow activities. Our objectives were to (i) determine to what degree it is possible to robustly recognize cow activities from GPS positioning data, using low-cost GPS receivers; and (ii) determine which types of activities can be classified, and what robustness to expect within the different classes. To provide data for this study low-cost GPS receivers were mounted on 14 dairy cows on grass for a day while they were observed from a distance and their activities manually logged to serve as ground truth. For our dataset we managed to obtain an average classification success rate of 86.2% of the four activities: eating/seeking (90.0%), walking (100%), lying (76.5%), and standing (75.8%) by optimizing both the preprocessing of the raw GPS data and the succeeding feature extraction.


Global Position System Global Navigation Satellite System Global Position System Receiver Segmentation Strategy Global Position System Position 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Agouridis, C., Stombaugh, T., Workman, S., Koostra, B., Edwards, D.: Examination of GPS Collar Capabilities and Limitations for Tracking Animal Movement in Grazed Watershed Studies. In: ASAE Annual International Meeting, pp. 27–30 (July 2003)Google Scholar
  2. 2.
    Galileo - a european global navigation satellite system, (Online; accessed 13-01-2011)
  3. 3.
    Gonzalez, L., Tolkamp, B., Coffey, M., Ferret, A., Kyriazakis, I.: Changes in feeding behavior as possible indicators for the automatic monitoring of health disorders in dairy cows. Journal of Dairy Science 91(3), 1017 (2008)CrossRefGoogle Scholar
  4. 4.
    Huzzey, J., Veira, D., Weary, D., von Keyserlingk, M.: Prepartum behavior and dry matter intake identify dairy cows at risk for metritis. Journal of Dairy Science 90(7), 3220–3233 (2007)CrossRefGoogle Scholar
  5. 5.
    Ito, K., von Keyserlingk, M., LeBlanc, S., Weary, D.: Lying behavior as an indicator of lameness in dairy cows. Journal of Dairy Science 93(8), 3553–3560 (2010)CrossRefGoogle Scholar
  6. 6.
    Kjærgaard, M.B., Blunck, H., Godsk, T., Toftkjær, T., Christensen, D.L., Grønbæk, K.: Indoor positioning using GPS revisited. In: Floréen, P., Krüger, A., Spasojevic, M. (eds.) Pervasive Computing. LNCS, vol. 6030, pp. 38–56. Springer, Heidelberg (2010)CrossRefGoogle Scholar
  7. 7.
    Mobile action, (Online; accessed 20-12-2010)
  8. 8.
    Mohr, M., Krustrup, P., Bangsbo, J.: Match performance of high-standard soccer players with special reference to development of fatigue. Journal of Sports Sciences 21(7), 519–528 (2003)CrossRefGoogle Scholar
  9. 9.
    Nadimi, E., Søgaard, H., Bak, T.: ZigBee-based wireless sensor networks for classifying the behaviour of a herd of animals using classification trees. Biosystems Engineering 100(2), 167–176 (2008)CrossRefGoogle Scholar
  10. 10.
    Phillips, C.: Cattle behavior and welfare. Blackwell Science Ltd., Malden (2002)CrossRefGoogle Scholar
  11. 11.
    Robert, B., White, B., Renter, D., Larson, R.: Evaluation of three-dimensional accelerometers to monitor and classify behavior patterns in cattle. Computers and Electronics in Agriculture 67(1-2), 80–84 (2009)CrossRefGoogle Scholar
  12. 12.
    Schwager, M., Anderson, D., Butler, Z., Rus, D.: Robust classification of animal tracking data. Computers and Electronics in Agriculture 56(1), 46–59 (2007)CrossRefGoogle Scholar
  13. 13.
    Spencer, M., Lawrence, S., Rechichi, C., Bishop, D., Dawson, B., Goodman, C.: Time-motion analysis of elite field hockey, with special reference to repeated-sprint activity. Journal of Sports Sciences 22(9), 843–850 (2004)CrossRefGoogle Scholar
  14. 14.
    Weka api, the university of waikato, (online; accessed 05-01-2011)
  15. 15.
    Zheng, Y., Liu, L., Wang, L., Xie, X.: Learning transportation mode from raw gps data for geographic applications on the web. In: WWW 2008: Proceeding of the 17th International Conference on World Wide Web, Beijing, China, pp. 247–256. ACM, New York (2008)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Torben Godsk
    • 1
    • 2
  • Mikkel Baun Kjærgaard
    • 2
  1. 1.Knowledge Centre for AgricultureAarhus NDenmark
  2. 2.Department of Computer ScienceAarhus University, IT-parkenAarhus NDenmark

Personalised recommendations