Researchers at the University of Minnesota have developed a motion-capture system that allows for the markerless 3D tracking of free-moving macaques (Bala et al., 2020). The new technology, called OpenMonkeyStudio, can accurately detect the poses and social interactions of two individuals as they move unconstrained within a 9' × 8' × 8' enclosure. This new deep learning system is an important advance over existing behavior tracking systems, which either rely on markers or have only 2D tracking capabilities. Macaques commonly show discomfort when fitted with jackets or bodysuits for tracking purposes; this causes them to move in constrained and unnatural ways. Other small, spherical markers are difficult to attach to macaques because of the animals’ long, thick fur; in addition, macaques are naturally curious and tend to remove any markers with their agile hands. While markerless 2D tracking systems are appropriate for some species, such as worms or mice, macaques adopt a much wider range of distinctive poses by moving through space in a fundamentally three-dimensional way. This new technology, which uses multiple cameras that reconstruct a full set of 3D movements using 13 joints as body landmarks, is thus less invasive and can yield more accurate data when used to track macaques for research purposes.
AWI Quarterly Issue