Advancing non-human primate welfare: An automated facial recognition system for unrestrained cynomolgus monkeys

Abstract

Cynomolgus monkeys (Macaca fascicularis) are vital in biomedical research, particularly for drug development and studying neurological diseases. However, accurately identifying individuals in group housing environments remains a significant challenge. This paper presents a near real-time facial recognition system tailored for cynomolgus monkeys, utilizing a fine-tuned Detectron2 model for face detection, followed by eigenface-based classification with Support Vector Machine (SVM) and radial basis function (RBF) kernel. The system achieved an accuracy of 97.65% in 10-fold cross-validation and identified individuals in under 1 minute under ideal conditions. This method eliminates the need for invasive identification techniques, potentially reducing stress and improving animal welfare, and has the potential to reduce the need for individualized housing or specialized enclosures. Additionally, as the system reduces the time and labor required for identifying monkeys, it might benefit research facilities with high turnover rates. This method could improve identification in non-human primate research while minimizing stress associated with traditional techniques.