Digital Screen Detection Using a Head-mounted Color Light Sensor

The immense growth of digital technology has led people to live in a multi-screen environment. If used for a prolonged time, digital screens (DS) can be considered harmful to the eyes leading to such health issues as computer vision syndrome (CVS). In this paper, we study the problem of Digital Screen Detection (DSD), using a new generation of small color light sensors that can observe four different components of the visible light, including two components of the blue light. To this aim, the sensor outputs are analyzed and different supervised machine learning algorithms are applied and compared to obtain a proper algorithm for DSD providing a trade-off between accuracy and cost. The proposed technique has been tested on 10 subjects performing experiments in three common office tasks. The results show that if the features are selected appropriately, high accuracy, of more than 86 %, can be achieved using Naïve Bayes and Random Forest classifiers.