Prototipe Sistem Deteksi Burung Menggunakan ESP32-Cam dan Algoritma YOLO

Abstract

The agriculture and plantation sectors play a crucial role in the economies of many countries, including Indonesia, but often experience productivity declines due to pests. Technologies such as IoT and artificial intelligence now offer modern and more effective solutions to these problems. This research develops an automated system for repelling pests based on the ESP32-CAM camera module utilizing the YOLO (You Only Look Once) object detection algorithm and OpenCV image processing. The system is designed to accurately recognize and repel birds in real-time and can be accessed and controlled remotely through an IoT platform. Previous studies have used various sensors, including ultrasonic and PIR sensors, but these often suffer from limitations in accuracy and flexibility. The YOLO algorithm was chosen for its ability to provide superior speed and detection accuracy. The research results show that the developed bird detection has a precision value of 0.998, a recall of 0.981, and an F1 Score of 0.989, These metric values indicate that the system is very reliable in detecting birds. High precision guarantees minimal false detections, while high recall ensures that almost all birds are detected. The combination of both, reflected in the F1-score, indicates a good balance between precision and recall. Thus, this system can make a significant contribution to improving pest control efficiency and agricultural productivity.