Killing-Time Detection from Smartphone Screenshots

Yu-Chun Chen, Kuei-Chun Kao, Yu-Jen Lee, Faye Shih, Wei-Chen Chiu, Yung-Ju Chang

Abstract: Finding good moments to deliver interruptions has drawn research attention. Since users have attention surplus at these moments, killing-time is considered one such a kind of moment. However, detection on killing-time has been under researched. In this paper, we propose a screenshot-based killing-time detection with deep learning. Our model achieves an accuracy 79.71%, recall 90.24%, precision 84.51%, and AUROC 65.50%. This suggests that using screenshots to detect users’ kill time behavior on smartphones is a promising approach. It may be worthwhile to investigate how the fusion of screenshots and sensor data can further improve detection.