Abstract - In this paper, we propose an HCI system based on EOG signals, which includes designing experiments using OpenBCI to collect EOG signals from multiple directions of the user and the corresponding visual direction categories, constructing and training an EOG-visual direction classification model. According to the control strategy corresponding to the line-of-sight direction category, the user could control the HCI device.

Clinical Relevance - This system can reduce the software and hardware costs of HCI while achieving a simple and effective interaction method. It also creates the possibility for patients with neurodegenerative diseases to control the computer.