CapuchinAI uses facial recognition to study monkeys in Costa Rica

Researchers from Emory University and the Georgia Institute of Technology have successfully deployed an AI-driven system called CapuchinAI in Costa Rica’s Taboga Forest Reserve to study primate cognition in natural environments.

The autonomous station consists of a ruggedized wooden box equipped with a touchscreen, camera, Raspberry Pi computer, and a motorized food dispenser. Using a YOLO-based facial recognition algorithm, the system can identify individual capuchin monkeys with 97 percent accuracy. This allows researchers to tailor cognitive tasks to specific individuals and prevents dominant monkeys from monopolizing the food rewards.

During a pilot study involving 16 capuchins, several monkeys learned to interact with the screen to receive dried banana pieces. Observations noted diverse interaction methods, with some monkeys using their fingers while others used their lips to press the screen. The project aims to move cognitive research out of controlled laboratory settings and into the unpredictable natural habitats where primate brains originally evolved.