Pengendalian Pendingin Ruangan (Air Conditioner) Berdasarkan Deteksi Jumlah Manusia Mengunakan Metode YOLO-V7 pada Orange Pi 5

Abstract

This study aims to develop an automatic air conditioning (AC) control system based on human detection using the YOLOv7 method on a single-board computer, Orange Pi 5. The background of this research is need to optimize AC usage in public spaces such as classrooms and meeting rooms, which are often used inneficiently. The methods used in this study include the utilization of a pre-trained YOLOv7 model trained on the COCO dataset and filtered specifically to detect humans from images captured every 10 minutes using an IP Camera. The system sends commands from the Orange Pi 5 to an Arduino UNO to control the AC operation, including turning it on, off and adjusting the temperature based on the number of people detected in the room. Temperature settings are made according to the standards set by the Indonesian Minister of Health Decree No. 1405/Menkes/SK/XI/2992, which recommends room temperature ranging from 23 ° to 26 ° C. The research results show that pre-trained YOLOv7 model effectively detects humans. This system also successfully controls the AC automatically based on the detected number of people detected and adjust it when the room tempereture exceeds the AC setting. Moreover the AC is turned off when the system detect no people in the room beyond a specified time limit.