WBCIC-SHU Motor Imagery Dataset

Brain-computer interfaces (BCIs) provide an effective means for users to control external software applications and devices solely by decoding their brain activity, without the need for muscle engagement. A large-scale, high-quality BCI dataset can stimulate researchers from related fields to develop advanced deep learning algorithms, thereby enriching the BCI domain. Therefore, creating an EEG dataset that supports the development and research of BCI systems is crucial. This dataset, derived from the World Robot Conference Contest-BCI Robot Contest MI, focuses on upper-limb or upper-and-lower-limb motor imagery (MI) tasks across three recording sessions. Sixty-two healthy, right-handed participants (ages 17–30, 18 females) with no prior BCI experience took part in this experiment. Of these, 52 subjects completed the two-class MI experiment, while 11 subjects participated in the three-class MI experiment. This dataset offers significant potential for a wide range of BCI-related research, including the analysis of inter-session variability for individual subjects and enhancing decoding algorithm performance.

If the download of this version is slow, you can choose to download Version 3 in batches.