篇名 | Smart Occupancy Detection System Based on Long Short-Term Memory Units |
---|---|
卷期 | 31:5 |
作者 | Asif Husnain 、 Tae-Young Choe |
頁次 | 159-175 |
關鍵字 | deep learning 、 lighting control 、 LSTM 、 occupancy detection 、 sensor system 、 EI 、 MEDLINE 、 Scopus |
出刊日期 | 202010 |
DOI | 10.3966/199115992020103105012 |
Smart lighting is a system designed to consume lighting energy efficiently. Occupancy detection is one of the key functionalities for a smart lighting system or home automation. The previous researches expect that room occupancy can be monitored by some sensors, i.e. remote thermal sensor arrays and pyroelectric sensors. Unfortunately, they cannot detect occupancy of the entire room with the limited number of sensors. In order to detect occupancy even in the offrange area of thermal sensors, we proposed a deep learning based occupant detection system comprising a 4×4 thermal sensor array and a PIR sensor. The proposed system is focuses on the occupancy detection of the whole room instead of the occupancy detection in front of the sensor area only. The deep learning module consists of Long Short-Term Memory (LSTM) units in order to achieve robust occupancy detection. The proposed system can memorize the sequence of human movements and detects occupancy of the room with high accuracy. The performance of the proposed system is compared with several state-of-the-art machine learning techniques and achieves 95.62% accuracy on test data set.