Comparing Manual and Automated Spatial Tracking of Captive Spider Monkeys Using Heatmaps

Area of Science:

• Animal behavior and welfare science
• Computer vision applications in zoology
• Primate ethology

Background:

• Traditional animal welfare assessments rely on manual observation, which is labor-intensive, prone to bias, and limited in temporal scope.
• Quantifying enclosure use and activity is crucial for supporting naturalistic behaviors and enhancing animal Quality of Life (QoL).

Purpose of the Study:

• To compare the efficacy of manual tracking (ZooMonitor) with automated pose estimation (SLEAP) for monitoring animal behavior.
• To validate computer vision technology as a reliable tool for zoo-based welfare assessments.

Main Methods:

• Manual observation and automated pose estimation (SLEAP) were used to track enclosure use and activity in black-headed spider monkeys.
• Spatial heatmaps and activity estimates were generated and compared between the two methods.
• Data were collected over six non-consecutive days on a mother-son pair at Aalborg Zoo.

Main Results:

• Both manual tracking and automated pose estimation showed strong agreement in identifying core activity areas (83-99% overlap, Pearson's r = 0.93-1.00).
• Comparable estimates of active time were obtained, with no significant differences between methods across days (p = 0.952).
• Automated pose estimation demonstrated reliability and scalability for monitoring enclosure use and activity.

Conclusions:

• Computer vision technology, specifically automated pose estimation, provides a dependable and efficient method for animal welfare assessments in zoos.
• This technology reduces the need for time-consuming manual observations, improving consistency and scalability.
• Automated monitoring enhances the ability to support naturalistic behaviors and improve animal QoL.