Automated pose estimation in primates.

Area of Science:

• Primatology
• Behavioral Biology
• Biomedical Research

Background:

• Quantifying primate behavior is crucial for various scientific fields, including primatology, psychology, and biomedicine.
• Traditional methods for behavioral quantification have limitations in terms of information richness, cost, and variability.
• Previous approaches included low-information measures (e.g., preference, reaction time) and high-cost, variable ethogram-based categories.

Purpose of the Study:

• To review the technological advancements in primate behavior measurement.
• To introduce and define the concept of behavioral imaging.
• To advocate for the development of analytical techniques for analyzing large-scale behavioral data.

Main Methods:

• Utilizing digital video cameras and automated pose tracking software to capture high-resolution primate full-body position over time.
• Inferring behavioral states (e.g., eating, sleeping, mating) from pose data.
• Situating these technological advances within the historical context of behavior study.

Main Results:

• Recent technological advances enable affordable, scalable, and rigorous quantification of primate behavior.
• Behavioral imaging provides high spatial and temporal resolution of primate pose, allowing for detailed behavioral state inference.
• This approach facilitates the analysis of 'big behavior' data.

Conclusions:

• Behavioral imaging represents a significant revolution in measuring primate behavior.
• Investment in analytical and research techniques is needed to leverage this new era of big behavior data.
• Primate centers and zoos are poised to play a more central role in future behavioral research.