Publication Date: 2023/12/30
Abstract: Smart control systems are broadly implemented by means of controlling remotely or automatically, however approaches the usage of imagery are nevertheless not often discovered. The goal of this research is to integrate a smarthome with a control platform that controls lights using images as a source of automated control. As input, images are obtained and the Haar-cascade classification algorithm is applied to calculate the number of human objects within the room. however, a light sensor is also used as a parameter in acquiring the light value in the room. If there's a lack of value due to the presence of a human object sitting in a chair, the system will provide extra lighting and regulate the room lighting value to the suitable room light standard. The control and settings interface is made in web form, where several types of control are available. communication between devices inside the system uses the MQTT protocol. From the research outcomes, this control system platform is successful in running and controlling lamp lighting based on image detection input. There are differences in the level of detection accuracy that is influenced by the distance between the camera and the object, apart from that the quality of the camera sensor in capturing images additionally affects detection overall performance. At the same time, objects tend to be easily detected, are objects that do not use facial accessories such as mask, glasses and so on. It was additionally found that the average reduction in light value inside the room became 0.83 lux for each 2-3 humans sitting in the room. This research answers a control system solution for a smart home with an indoor image detection approach. similarly research is needed to precisely calculate the comparison of the accuracy of object detection with other image detection algorithms.
Keywords: Control System, OpenCV, MQTT.
DOI: https://doi.org/10.5281/zenodo.10443824
PDF: https://ijirst.demo4.arinfotech.co/assets/upload/files/IJISRT23DEC1264.pdf
REFERENCES