Virginia Bear Mange Study
Utilizing Tri-Axial Accelerometers to Classify Behaviors of Free-Ranging American Black Bears
Tracking wildlife is becoming increasingly common among wildlife managers and researchers. A major advancement in this field is the use of GPS collars that allow us to see where animals travel and how they move across different landscapes. Many of these collars are also equipped with small devices called tri-axial accelerometers. These accelerometers measure movement in three different directions. They track motion up-and-down (X-axis), side-to-side (Y-axis), and front-to-back (Z-axis).
These accelerometers can record detailed movement patterns multiple times each second. However, determining exactly what an animal is doing based solely on these movement patterns is challenging, especially for elusive animals that are difficult to observe in the wild. To address this issue, we are utilizing data from camera collars to validate these behavior models using video footage collected during the Virginia Appalachian Carnivore Study.
As part of the Virginia Appalachian Carnivore Study, camera collars equipped with built-in accelerometers were deployed on 15 American black bears. These collars recorded short video clips and detailed movement data, with the videos labeled for the bears' behaviors second-by-second. We are developing supervised classification models to identify different behavioral readings from these triaxial accelerometer outputs based on the video data in this dataset.
A goal of this research is to enable wildlife managers and researchers with similar datasets to process their data through our model, obtaining behavior classifications for their collared individual American black bears and gaining a better understanding of their behavior linked to locational information.
Are you a wildlife manager or researcher?
Do you have collars with tri-axial accelerometers?
If you think you may have tri-axial accelerometry data that could be used in this model, or want to learn more about possibly using this classification model with your collar data, please fill out this Google form (or scan this QR Code) with your contact information and I will reachout to discuss with you more!
