Evaluating 3D Visual Fatigue Induced by VR Headset Using EEG and Self-attention CNN

3D visual fatigue is one of the major factors that hinder the development of virtual reality contents towards larger population. We proposed an EEG-based self-attention CNN model to evaluate user's 3D visual fatigue in an end-to-end fashion. We adopted a wavelet-based convolution to extract spatiotemporal information and prevent overfitting. Besides, a self-attention layer was added to the feature extractor backbone to cope with the subject-variation problem in EEG-decoding. The proposed method is compared with four state-of-the-art methods, and the results demonstrate that our model has the best performance among all methods in subject-dependent and cross-subject scenarios.