The Tiger Tracker is a first-of-its-kind system that will autonomously monitor and log the behaviors of animals in captivity. This system will use computer vision and machine learning to collect information on the tiger’s behavior throughout the day and night. This information will give the animal care managers at the zoo faster feedback on the tiger’s well-being and give researchers a fast, robust data collection method. Given this data, researchers and caretakers will be able to quickly quantify how adjustments to the environments affect animals’ behaviors and their overall well-being.
The Tiger Tracker will use the thermal animal tracking technology developed in the Intelligent Camera Trap project to characterize the tiger’s behavior while capturing videos. As it records this footage, the Tiger Tracker will recognize and flag videos of interesting tiger behaviors. At first the system will sort out ‘active’ and ‘inactive’ clips, corresponding to motion or lack thereof. Once this basic characterization is working, we can move on to higher level behavior classifications, like pacing recognition. It is important for researchers and caretakers to quickly identify negative behaviors like pacing, since it is a strong indicator of anxiety in an animal. The sooner researchers can spot this behavior, the sooner they can take action to mitigate the source of stress. After classifying pacing behavior, we can move up to even more sophisticated analytics, like documenting where the animal goes and how it spends time in its enclosure throughout the day.