Dresden 2017 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
BP: Fachverband Biologische Physik
BP 19: Posters - Computational Biophysics
BP 19.3: Poster
Dienstag, 21. März 2017, 14:00–16:00, P1A
Automated tracking of Adelie penguins — •Alexander Winterl1, Daniel Zitterbart1,2, Sebastian Richter1, Richard Gerum1, and Ben Fabry1 — 1Department of Physics, Biophysics Group, Friedrich-Alexander-University Erlangen-Nuremberg, 91052, Erlangen, Germany — 2Alfred Wegener Institut, Helmholtz Zentrum für Polar und Meeresfoschung, 27568 Bremerhaven, Germany
Decision-making processes of colony-forming birds when they approach or leave their home colony e.g. during foraging are currently poorly understood and remain largely unexplored. To identify rules that govern such processes in Adelie penguins, we recorded high-resolution time-lapse images of a colony near Dumont d'Urville, Antarctica, during January and February 2015. We then developed an automated tracking software to follow the movements of all birds outside the colony within the field of view. To correctly allocate crossing tracks when two penguins pass each other, we model the penguin movements by an auto-regressive random walk that assumes that penguins do not abruptly change their speed and direction. 12 ambiguous tracks were selected and analyzed by human observers who watched the full time-lapse image sequence. Compared to this "ground truth", the assignment quality of the algorithm (25% error) was slightly inferior to human observers (17% error) when only the center of mass positions of the crossing tracks are provided. We conclude that reliable automated tracking of passing and crossing penguins requires the analysis of additional image information such as body posture and head orientation.