📝 SummarXiv

Abstract

This paper introduces an open-source framework, Video2EEG-SPGN-Diffusion, that leverages the SEED-VD dataset to generate a multimodal dataset of EEG signals conditioned on video stimuli. Additionally, we disclose an engineering pipeline for aligning video and EEG data pairs, facilitating the training of multimodal large models with EEG alignment capabilities. Personalized EEG signals are generated using a self-play graph network (SPGN) integrated with a diffusion model. As a major contribution, we release a new dataset comprising over 1000 samples of SEED-VD video stimuli paired with generated 62-channel EEG signals at 200 Hz and emotion labels, enabling video-EEG alignment and advancing multimodal research. This framework offers novel tools for emotion analysis, data augmentation, and brain-computer interface applications, with substantial research and engineering significance.