A Cloud-Based AIoT Application in Smart Building

Abstract

This research article presents a design and implementation of an Internet of Things (IoT) application for detecting people in smart buildings. Most occupancy detection systems use motion sensors to detect people indoor and often requires a highly stable internet connection to ensure reliable communication between devices. To lessen network-related issues, we propose a cloud-based platform that is suitable for the smart building environment. An IoT application that consists of image processing system, control system, and statistics display system is developed and deployed on an on-premise cloud. The images captured from surveillance cameras are processed by artificial intelligence with Intel’s pre-trained model in order to detect people in each zone of the building. Based on the detection results, the control system automatically adjusts lighting and air conditioning in rooms to preserve energy, as well as opens or closes building gates. The application also sends out safety alert in case detection occurs outside of the specified time via LINE Notify. In addition, the platform is designed to scale to support more zones in the future. The experiment results showed that our system was able to reduce overall energy usage hours of the building by 31.47 percent.