Design And Development of An AI-Driven Solution For Rhesus Macaque

Abstract

In much of the farming land of rural Nepal, wild animals, most notably the Rhesus Macaque, a widely distributed monkey species found in Nepal, pose a significant issue. These animals frequently find their way into the farmland and destroy the farm crops, endangering the efforts and earnings of the farm- ers. The research aims to create a smart solution that uses Artificial Intelligence-based detection and ultrasonic-based repellents. The system enables the use of the AI-powered cameras to scan and monitor the target area and identify the Rhesus Macaque in real-time using the YOLO algorithm, trained using a custom dataset of monkey images, and upon detection, use an ultrasonic deterrent system to repel the Rhesus Macaque. The detection model was trained on the dataset collected from the Pashupatinath and Swayambhunath area, as the farmland shares similar species, to enhance the detection performance of the model under local settings. Upon identification of the species, the detection model activates the ultrasonic deterrent subsystem. The YOLO based detection model achieved a precision of up to 1.00, an F1-score of 0.97, and a recall of 0.99. The 555 Timer IC-Based ultrasonic generator was validated at the prototype stage, with its generated output frequency measured and verified using an oscilloscope.