Automated markerless pose estimation in freely moving macaques with OpenMonkeyStudio

TLDR

The rhesus macaque is a key model species across neuroscience, psychology, ethology, and medicine, yet existing tracking approaches fail to provide sufficient precision for freely moving behavior. The study aims to enhance the macaque model by enabling precise measurement of behavior in freely moving conditions through the development of OpenMonkeyStudio. OpenMonkeyStudio employs 62 machine‑vision cameras surrounding a 2.45 m × 2.45 m × 2.75 m enclosure, using multiview image streams to augment data via 3D reconstruction and train a robust view‑invariant deep neural network. The resulting view‑invariant system surpasses prior 2D markerless trackers, enabling fully automatic 3D pose inference on unconstrained natural motion and accurate recognition of actions and social interactions.

Abstract

Abstract The rhesus macaque is an important model species in several branches of science, including neuroscience, psychology, ethology, and medicine. The utility of the macaque model would be greatly enhanced by the ability to precisely measure behavior in freely moving conditions. Existing approaches do not provide sufficient tracking. Here, we describe OpenMonkeyStudio, a deep learning-based markerless motion capture system for estimating 3D pose in freely moving macaques in large unconstrained environments. Our system makes use of 62 machine vision cameras that encircle an open 2.45 m × 2.45 m × 2.75 m enclosure. The resulting multiview image streams allow for data augmentation via 3D-reconstruction of annotated images to train a robust view-invariant deep neural network. This view invariance represents an important advance over previous markerless 2D tracking approaches, and allows fully automatic pose inference on unconstrained natural motion. We show that OpenMonkeyStudio can be used to accurately recognize actions and track social interactions.