Brain-Download-Datasets

Datasets

The dataset contain EEG signals recorded from EMOTIV Insight 5-channel headset of four different experiments. All the following experiments except for Baseline were conducted by visually stimulating the subject's brain with a random image presentation. The image sequencing was done using the Brain Download software.

1. Baseline: EEG signals of calmed mind with eyes opened and closed.
2. Image-Blank: Simulated thinking “something” and “nothing” on the subject’s brain by randomly presenting images and blank screens.

3. Left-Right_Arrows: Showed left and right directed arrows on the left and right edges of the screen respectively and the subject was instructed to directly look at them without moving the head. Since the image sequence is randomized, a reference mark was presented at the center of the screen after each arrow image.

4. Visual_Q&A: Simultaneously displayed a YES or NO answerable question about the presented image and instructed the subject to think about the answer.

Scripts

• After extracting the .zip files in the Dataset directory, you can plot a random record (.csv file) using plot_random_full.py.
• You can split the records (.csv files) into smaller chunks of watching one stimulus by using the splitter.py script. The separated chunks and their labels will be saved as .npy files.
• You can plot some random chunks in a saved .npy data file by using the plot_random_split.py.

Experiment Details

All the data samples from the above mentioned experiments were collected using one male subject (25 years old). The following table shows the details of the performed experiments.

Property	Image-Blank	Left-Right	Visual_Q&A
Number of classes	2 (Image, Blank)	3 (Center, Left, Right)	2 (Yes, No)
Number of images in the image-set	500	320	250
Image resolution	1200x800 pixels	1896x790 pixels	1200x800 pixels
Screen time for an image	2 seconds	2 seconds	2 seconds
Image presentations per record(.csv file)	20	20	20
Number of records	500	300	250
Total stimuli observed	10,000	6,000	5,000