Mental fatigue estimation based on facial expressions during speech

In this study, we propose a method to estimate a user´s fatigue state resulting from mental load (e.g., from office work) based on facial expression information. Changes in appearance because of mental fatigue are subtle and difficult to estimate using ambient sensor information. Thus, we estimate the user´s mental fatigue based on facial expressions during speech. The experimental results show that the fatigue state can be recognized with up to an 89% accuracy. Furthermore, we confirm that our method enables highly accurate estimation of the fatigue state without dependence on the task types.