Implementation of Sleep Detector Using Histogram of Oriented Gradients and Support Vector Machine for Saving Electricity in Household Electronic Equipment

Abstract

Users of electronic devices sometimes leave their devices on while sleeping or when they are not actively using them, resulting in the electricity being wasted. The effort to save electricity on electronic equipment is needed to reduce excessive electricity consumption to save natural resources, and reduce the cost of electricity bills. This paper aims to design a system to evaluate the user's sleep state when using electronic devices. The system estimates the user's condition when he awakes, is asleep or has left the device using the Histogram of Oriented Gradients (HOG) and Support Vector Machine (SVM) methods. In order to acquire input images for both methods, we use a camera as an auxiliary tool. This system will actively estimate the user's condition and automatically turn off the electronic device if some threshold is reached. This study analyzes the effectiveness of the HOG and SVM methods in their implementation in a sleep detection system. The experimental results show that the system can detect the user's condition in real time with an accuracy of 90%. These results are obtained from a system that has been integrated with actuators mounted on electronic devices such as televisions, lights and fans.