EmoX

Citation Author(s):
Shuangyou
Li
Northeast Electric Power University
Submitted by:
Shuangyou Li
Last updated:
Fri, 01/10/2025 - 08:38
DOI:
10.21227/bnar-c233
License:
0
0 ratings - Please login to submit your rating.

Abstract 

With the continuous advancement of technology, small and portable physiological sensors that can be worn on the body are quietly integrating into our daily lives, and are expected to greatly enhance our quality of life. In order to further enrich and expand the emotional physiological signals captured by portable wearable devices, we utilized the 14-channel portable EEG acquisition device Emotiv EPOC X, and with emotional video clips as the stimulus source, we collected two sets of emotional EEG signals from two groups of 10 participants each, named EmoX1 and EmoX2. In these two sets of data, we also additionally collected EEG signals in the resting state, which have been proven to play an important role in the preprocessing stage of emotional signals. In addition, each dataset contains emotional label data, providing strong label support for emotion classification based on artificial intelligence. For specific information, please refer to the attached PDF file. We look forward to our research findings being of help to you.

Instructions: 

EmoX1: The dataset contains EEG signals of 10 subjects (5 males and 5 females), which are collected by a 14 - channel Emotiv EPOC X device while subjects watched 15 emotion stimulus clips. And these signals are stored in folders ending with “raw_signal”. To help subjects return to a neutral emotional state, before each emotion data collection, we play a piece of calming music, and record the EEG signals as baseline data, subsequently which are stored in folders ending with “background.” All EEG signals are recorded at a sampling rate of 128Hz and processed using a band-pass Hamming sinc linear-phase FIR filter. After watching the video clips, subjects are asked to rate the arousal, valence, and dominance values on a scale from 1 to 5. These ratings are saved as emotion labels in a file named “label.csv”. 

EmoX2: The data collection process and equipment are the same as in EmoX1, with the only difference being the gender composition of the subjects (this time consisting of 6 males and 4 females). Additionally, we selected a different set of 15 emotion stimuli. The stimuli clips in EmoX1 were sourced from short video platforms, reflecting more everyday scenarios, while the EmoX2’s are from movie platforms, giving them a more formal tone.

Dataset Files

LOGIN TO ACCESS DATASET FILES

Documentation

AttachmentSize
File EmoX.pdf185.06 KB