High Frequency Event-based Eye Tracking Towards Mental Health Diagnosis

Eye movement analysis is a key technique for understanding information processing in virtual reality (VR) and psychological therapy. Then the dynamics of the eye movement overtime is an important reference for mental health diagnosis. However, it is challenging for conventional CMOS/CCD cameras to capture the detailed change of fast moving eye balls. To bridge the gap, in this paper, we propose to use event cameras, which are novel sensors that record pixel-level brightness changes at high frequencies, to achieve over kilohertz frequency eye tracking. This enables us to collect rich and precise data on eye movements. To demonstrate its potential on providing high temporal resolution of eye movement for mental health diagnosis, we design a deep learning-based pipeline to accurately extract pupil region of human eyes recorded by event cameras and calculate a number of key parameters including pupil region, blink rate and pupil trail.