Accept the licences of the 77 datasets to access EEGDash Corpus v0
This repository is publicly accessible, but you have to accept the conditions to access its files and content.
EEGDash Corpus v0 is a collection of 77 public EEG datasets. The collection has no licence of its own: each dataset keeps its own licence, listed with a link in LICENSES.csv and written into the dataset's buckets in the store. By requesting access you agree to: (1) use every dataset in the store under its own licence; (2) credit the corpus and every dataset you use (CITATION.bib); (3) keep the conditions that apply where you use the data they cover: non-commercial use for SparrKULee (nm000238, CC BY-NC 4.0), share-alike for the twelve CC BY-SA 4.0 datasets, the terms of the GPL-3.0 for Shin2017A (nm000267), or leave those buckets out; (4) not redistribute the store, in whole or in part, without these terms and an equivalent access gate. This summary is not legal advice; the licence texts govern.
Log in or Sign Up to review the conditions and access this dataset content.
EEGDash Corpus v0
77 public electroencephalography (EEG) datasets, served through EEGDash from OpenNeuro and NEMAR, converted once with the preprocessing recipe of the REVE EEG foundation model into training-ready 10-second samples, in one Zarr store. Every dataset keeps its own licence; access is gated so that every user accepts them.
At a glance
| quantity | value |
|---|---|
| datasets | 77 (56 from OpenNeuro, 21 from NEMAR) |
| subjects | 7,272 (counted per dataset) |
| sessions (recordings kept) | 41,133 |
| hours at 200 Hz | 14,074.7 |
| samples (10 s, 2,000 time points) | 5,045,111 |
| tokens per epoch (samples × channels × 11 patches of 200 points, stride 180) | 6,206,191,189 (6.21 B) |
| buckets (one per dataset and exact channel set) | 632 |
| channels per bucket | 6–256 (111.8 duration-weighted mean) |
| native sampling rates | 200–10,000 Hz |
| files in the store | 44,801 |
| size | 3.60 TB on disk (3,356.8 GiB, Blosc-zstd, lossless); 4.51 TB as uncompressed float32 |
One dataset dominates: PEERS (ds004395) holds 64.7 % of the hours and 73.0 % of the tokens.
By paradigm
| paradigm | datasets | subjects | hours | % of hours | tokens | % of tokens |
|---|---|---|---|---|---|---|
| memory & cognition | 5 | 510 | 9,658.9 | 68.6 | 4.79 B | 77.2 |
| development (rest & tasks) | 9 | 2,192 | 1,694.1 | 12.0 | 0.85 B | 13.7 |
| perception & emotion | 20 | 598 | 671.1 | 4.8 | 0.20 B | 3.2 |
| BCI (MI & P300) | 15 | 584 | 626.8 | 4.5 | 0.13 B | 2.1 |
| clinical routine EEG | 1 | 2,413 | 489.0 | 3.5 | 0.04 B | 0.7 |
| resting & consciousness | 9 | 542 | 318.6 | 2.3 | 0.08 B | 1.2 |
| auditory & language | 6 | 198 | 312.2 | 2.2 | 0.07 B | 1.1 |
| seizure | 1 | 14 | 141.0 | 1.0 | 0.02 B | 0.3 |
| learning & reward | 9 | 155 | 119.8 | 0.9 | 0.03 B | 0.4 |
| SSVEP | 2 | 66 | 43.2 | 0.3 | 0.01 B | 0.2 |
| total | 77 | 7,272 | 14,074.7 | 100 | 6.21 B | 100 |
The paradigm of each dataset is in the table at the end of this card. Memory & cognition is dominated by PEERS; the nine Healthy Brain Network releases (children and adolescents, resting state and a task battery) are the development group; clinical routine EEG is NMT (normal and abnormal routine recordings); seizure is Siena.
By source
| source | datasets | subjects | hours | mean channels | tokens | % of tokens |
|---|---|---|---|---|---|---|
| OpenNeuro: Healthy Brain Network, releases 1–9 | 9 | 2,192 | 1,694.1 | 129 | 0.85 B | 13.7 |
OpenNeuro: PEERS (ds004395) |
1 | 364 | 9,110.8 | 126 | 4.53 B | 73.0 |
| OpenNeuro: other datasets | 46 | 1,549 | 1,726.8 | 83 | 0.57 B | 9.2 |
| NEMAR: other datasets | 21 | 3,167 | 1,542.9 | 42 | 0.26 B | 4.1 |
| total | 77 | 7,272 | 14,074.7 | 112 | 6.21 B | 100 |
By licence
| licence | datasets | subjects | hours | % of hours | tokens | % of tokens |
|---|---|---|---|---|---|---|
| CC0 1.0 | 52 | 2,137 | 11,211.0 | 79.7 | 5.19 B | 83.6 |
| CC BY 4.0 | 11 | 400 | 475.1 | 3.4 | 0.08 B | 1.3 |
| CC BY-SA 4.0 (Healthy Brain Network releases 1–9, NMT, Sosulski2019, AlexMI) | 12 | 4,626 | 2,203.7 | 15.7 | 0.89 B | 14.4 |
GPL-3.0 (Shin2017A, nm000267) |
1 | 29 | 29.0 | 0.2 | 0.003 B | 0.06 |
CC BY-NC 4.0 (SparrKULee, nm000238) |
1 | 80 | 155.7 | 1.1 | 0.04 B | 0.6 |
The per-dataset list, with licence, paradigm, authors, source DOI, archive page, hours, subjects and buckets: LICENSES.csv.
Licence model
The collection has no licence of its own: it is the union of its datasets' licences. Nothing is relicensed. Each
bucket carries its dataset's licence in its Zarr attributes (licence as an SPDX id, licence_url, source = the
dataset's page on OpenNeuro or NEMAR); the root group carries licence: "other", licences (dataset id → SPDX id,
77 entries) and licence_file: "LICENSES.csv". Using the whole store therefore means non-commercial use (SparrKULee),
share-alike for the twelve CC BY-SA datasets, the GPL's terms for Shin2017A, and attribution for all 77. The
permissive subset is the buckets whose licence is CC0, CC BY or CC BY-SA: 75 datasets, 98.7 % of the hours and
99.3 % of the tokens. Four NEMAR copies carry a licence label that differs from the upstream deposit (Lee2019 MI, ERP,
SSVEP and Cho2017, CC0 1.0 at GigaDB); the store and LICENSES.csv follow the deposits. No legal assurance is given:
check the licence of each dataset you use.
Changes to the source datasets
This section is the notice that the attribution and share-alike licences ask for: what was done to every dataset between its archive and this store. Nothing in it changes any dataset's licence.
Source. Every dataset was read through EEGDash, as the raw BIDS files of its archive: OpenNeuro for the 56 ds
datasets (53 of them served from NEMAR's re-host, EEGDash id on…, the same files), NEMAR for the 21 nm datasets.
The version is the one the archive served at download (OpenNeuro's versioned DOI in LICENSES.csv where the
catalogue records it).
What was done to every recording, in order:
| # | step | detail |
|---|---|---|
| 1 | read | the recording file (BrainVision, EEGLAB, EDF or BDF) with MNE-Python through EEGDash |
| 2 | recordings skipped | sampled below 200 Hz (none in v0); 10 s or shorter (16); unreadable (11); fewer than 3 channels with a position (3) |
| 3 | channels kept | channels typed EEG in the dataset; when a recording types none as EEG, every channel whose name resolves to a scalp position. Non-EEG channels (EOG, ECG, EMG, respiration, triggers, …) are removed |
| 4 | positions | each kept name matched, case-insensitively, against the position bank: the public REVE bank (brain-bzh/reve-positions, 543 names) plus MNE's biosemi160 and biosemi256 templates. Rules: a leading EEG is ignored; BioSemi ring labels (A1…H32) go to the 128-, 160- or 256-channel template by the rings the recording carries; a zero typed for the letter O (P03) is read as PO3 when the exact name fails; the same electrode twice is kept once. An EEG channel the bank lacks keeps its own name and is placed from the recording's electrodes.tsv when that file, fitted to the bank on at least 8 shared electrodes (rotation, uniform scale, translation), fits within 20 mm RMS; otherwise it is dropped |
| 5 | channel names and order | the bank's spelling (the dataset's own for channels placed from electrodes.tsv), sorted alphabetically |
| 6 | band-pass | 0.5–99.5 Hz, MNE's default zero-phase FIR filter; the upper edge is 0.45 × the native rate when that is lower, i.e. 90 Hz for the three datasets recorded at 200 Hz (nm000146, nm000181, nm000267) |
| 7 | resampling | to 200 Hz, polyphase (none for native 200 Hz) |
| 8 | values | microvolts, float32 |
| 9 | statistics | per channel, the mean and SD of the whole processed recording, stored beside the samples |
| 10 | samples | non-overlapping 10-s windows (2,000 points) from the first point; the remainder under 10 s is not stored as a sample, nor the last full window of a recording that is an exact multiple of 10 s (REVE's loader cuts this way) |
| 11 | buckets | the recordings of one dataset with the same channel set appended into one group; Blosc-zstd compression, lossless |
Not done. No normalisation of the stored values (the samples are band-passed microvolts: REVE z-scores each sample with its session's mean and SD and clips at 15 SD at load time; another model can apply its own rule to the same bytes), no filtering beyond the band-pass, no notch filter, no re-referencing (the reference is as recorded), no clipping, no artefact or bad-segment rejection, no interpolation of missing or bad channels. Events, annotations, participant tables and stimulus files are not copied: each session's source file (BIDS path) is kept in the bucket attributes and the manifest, so they can be joined from the archive.
Channels dropped. Every channel dropped for lack of a position is counted per session in the manifest
(n_dropped_no_position; EEG-typed channels only, since channels typed otherwise are removed at step 3);
n_positions_aligned counts the channels placed from electrodes.tsv. The most frequent dropped names: EOGvu,
x_EMGl, x_GSR, x_Respi, x_Pulse, x_Optic (1,060 sessions each), EMG1–EMG4 (432), HEO, VEO (371),
VEOG, HEOG (202), EXG2–EXG6: non-scalp channels. Three datasets lose scalp electrodes the bank does not know
and the dataset does not place: ds004356 (EP1, EP2; 5.9 % of its channel-hours, other channels included),
ds005697 (CB1, CB2; 5.4 %), ds004279 (CB1, CB2; 3.1 %). The per-dataset effects are tabulated at the end
of this card.
Layout
| path in the repository | content |
|---|---|
README.md |
this card, including the change notice |
LICENSES.csv |
one row per dataset: ids, name, licence and its URL, paradigm, authors, source DOI or URL, archive, hours, subjects, buckets |
CITATION.bib |
the corpus and the 77 datasets, keyed by dataset id |
eegdash-corpus-v0-buckets.csv |
one row per bucket: bucket, dataset, shard, n_samples, n_sessions, n_chans |
eegdash-corpus-v0-manifest.csv |
one row per session: source file (BIDS path), bucket, channels, duration, native rate, channels dropped, subject, session, run, task |
eegdash-corpus-v0-SHA256SUMS |
SHA-256 of every file of the store |
eegdash-corpus-v0.zarr/ |
the store (Zarr v3) |
The store: one group per bucket, named 0 … 631.
| array | shape | dtype | content |
|---|---|---|---|
<b>/data |
(samples, 2000, channels) | float32 | the 10-s samples, µV; one chunk per sample, 128 samples per shard file, Blosc-zstd level 5 |
<b>/session |
(samples,) | int32 | the session (row of stats, entry of sessions) each sample comes from |
<b>/stats |
(sessions, 2, channels) | float32 | per-session mean and SD per channel, µV |
<b>/positions |
(channels, 3) | float32 | electrode positions, metres (MNE head frame) |
| attributes of | keys |
|---|---|
| the root | format: "reve-samples", version, sfreq: 200, window: 2000, units: "uV", band_pass_hz: [0.5, 99.5], datasets, n_buckets, n_samples, buckets (the bucket index), licence, licences, licence_file |
| a bucket | dataset, channels (names, sorted), sessions (per session: source file, duration in samples at 200 Hz, n_samples, first_sample), bucket, source_shard, licence, licence_url, source |
Loading
Access needs a Hugging Face account that has accepted the terms, and a token (hf auth login).
hf download EEGDash/eegdash-corpus-v0 --repo-type dataset --local-dir eegdash-corpus-v0 # everything, 3.6 TB
hf download EEGDash/eegdash-corpus-v0 --repo-type dataset --local-dir eegdash-corpus-v0 \
--include "eegdash-corpus-v0.zarr/zarr.json" "eegdash-corpus-v0.zarr/0/**" # the root and bucket 0
import numpy as np
import zarr # zarr >= 3
root = zarr.open_group("eegdash-corpus-v0/eegdash-corpus-v0.zarr", mode="r")
print(root.attrs["n_samples"], root.attrs["licences"]["ds004395"]) # 5045111 CC0-1.0
b = root["0"]
print(b.attrs["dataset"], b.attrs["licence"], b.attrs["source"], len(b.attrs["channels"]))
x = b["data"][0] # (2000, C) float32, µV: one 10-s sample
mean, sd = b["stats"][int(b["session"][0])] # its session's per-channel mean and SD
z = np.clip((x - mean) / sd, -15, 15) # REVE's normalisation, applied at load
pos = b["positions"][:] # (C, 3), metres
# the permissive subset (CC0, CC BY, CC BY-SA), from the root's index without opening each bucket
permissive = {"CC0-1.0", "CC-BY-4.0", "CC-BY-SA-4.0"}
keep = [str(r["bucket"]) for r in root.attrs["buckets"] if root.attrs["licences"][r["dataset"]] in permissive]
Held out
The corpus is meant to pretrain models that are then evaluated on OpenEEGBench, so the benchmark's sources were kept out of it. Of the eight public tasks, seven come from datasets EEGDash serves and were never in the corpus; the conversion refused them by id. ISRUC is not in EEGDash's catalogue.
| task | classes | what it tests | source | nearest paradigm in the corpus |
|---|---|---|---|---|
| BCIC-IV-2a | 4 | motor imagery | nm000139 |
BCI (MI & P300) |
| PhysioNet-MI | 4 | motor imagery | ds004362 |
BCI (MI & P300) |
| BCIC-2020-3 | 5 | imagined speech | nm000113 |
auditory & language (Arabic inner speech) |
| Mumtaz | 2 | depression vs control, resting EEG | nm000114 |
resting & consciousness; clinical routine EEG |
| Zyma | 2 | mental arithmetic load | nm000109 |
memory & cognition |
| FACED | 9 | emotion while watching video clips | nm000112 |
perception & emotion |
| CHB-MIT | 2 | seizure detection, paediatric scalp EEG | nm000110 |
seizure (adult, Siena); development (children, no seizures) |
| ISRUC | 5 | sleep staging | not in EEGDash | none: the corpus holds no sleep recordings |
Integrity
eegdash-corpus-v0-SHA256SUMS holds the SHA-256 of every file of the store (44,801 lines, sha256sum format, paths
relative to the repository root). From the root of a full download:
sha256sum -c eegdash-corpus-v0-SHA256SUMS
The file's own SHA-256 is 0cba6a7b2106d8f9853c04d07f647256c32aa0c51ae4f8bc0953eaa882e32e54, at repository revision
v1.0.
Citation
Cite the corpus:
@inproceedings{eegdash_corpus_v0,
title = {{EEGDash Corpus}: A Training-Ready Public Corpus for {EEG} Foundation Models},
author = {Shahi, Kuber and Aristimunha, Bruno and Delorme, Arnaud},
booktitle = {7th International Workshop on Machine Learning for EEG Signal Processing (MLESP 2026), with IEEE BIBM 2026},
year = {2026}
}
and every dataset you use: CITATION.bib has one entry per dataset, keyed by its id (ds…, nm…), plus REVE, whose
recipe the corpus follows.
Per dataset
Paradigm, sessions kept, native rates, channels kept, EEG-typed channels dropped for lack of a position, and sessions with
channels placed from electrodes.tsv (from eegdash-corpus-v0-manifest.csv).
| dataset | short name | paradigm | licence | sessions | native rate (Hz) | channels kept | EEG-typed channels dropped per session (mean) | sessions with channels placed from electrodes.tsv |
|---|---|---|---|---|---|---|---|---|
ds001787 |
Meditation | resting & consciousness | CC0-1.0 | 40 | 256 | 64 | 0.0 | 0 |
ds002680 |
Go/no-go | perception & emotion | CC0-1.0 | 350 | 1,000 | 31 | 0.0 | 350 |
ds002721 |
Affective music | perception & emotion | CC0-1.0 | 185 | 1,000 | 19 | 0.0 | 0 |
ds002778 |
Parkinson rest | resting & consciousness | CC0-1.0 | 46 | 512 | 32 | 8.0 | 0 |
ds003004 |
Imagined emotion | perception & emotion | CC0-1.0 | 34 | 256 | 134–229 | 4.8 | 0 |
ds003690 |
Rest & auditory | resting & consciousness | CC0-1.0 | 375 | 500 | 61 | 3.9 | 375 |
ds003775 |
SRM rest | resting & consciousness | CC0-1.0 | 152 | 1,024 | 64 | 0.0 | 0 |
ds003825 |
THINGS-EEG1 | perception & emotion | CC0-1.0 | 50 | 1,000 | 63–123 | 0.2 | 0 |
ds003846 |
Prediction error | learning & reward | CC0-1.0 | 50 | 500 | 63 | 0.0 | 0 |
ds003885 |
Object motion 1 | perception & emotion | CC0-1.0 | 24 | 1,000 | 123 | 4.0 | 0 |
ds003887 |
Object motion 2 | perception & emotion | CC0-1.0 | 24 | 1,000 | 123 | 4.0 | 0 |
ds003969 |
Meditation vs think | resting & consciousness | CC0-1.0 | 392 | 1,024, 2,048 | 64 | 0.0 | 0 |
ds004043 |
Feature attention | perception & emotion | CC0-1.0 | 20 | 1,000 | 63 | 0.0 | 0 |
ds004147 |
Task value | learning & reward | CC0-1.0 | 12 | 1,000 | 31 | 0.0 | 0 |
ds004148 |
Test-retest rest | resting & consciousness | CC0-1.0 | 900 | 500 | 61 | 0.0 | 0 |
ds004152 |
Drum trainer | learning & reward | CC0-1.0 | 21 | 1,000 | 31 | 0.0 | 0 |
ds004262 |
Feedback | learning & reward | CC0-1.0 | 21 | 1,000 | 31 | 0.0 | 0 |
ds004264 |
Steer the ship | learning & reward | CC0-1.0 | 21 | 1,000 | 31 | 0.0 | 0 |
ds004279 |
Spanish sentences | auditory & language | CC0-1.0 | 60 | 1,000 | 62 | 2.0 | 0 |
ds004284 |
Neuroforecasting | learning & reward | CC0-1.0 | 18 | 1,000 | 129 | 0.0 | 0 |
ds004315 |
Mood induction | perception & emotion | CC0-1.0 | 50 | 500 | 59–63 | 6.4 | 0 |
ds004324 |
Toon faces | perception & emotion | CC0-1.0 | 26 | 500 | 28 | 0.0 | 0 |
ds004356 |
Music vs speech | auditory & language | CC0-1.0 | 24 | 10,000 | 32 | 2.0 | 0 |
ds004357 |
Features-EEG | perception & emotion | CC0-1.0 | 16 | 1,000 | 123 | 4.0 | 0 |
ds004395 |
PEERS | memory & cognition | CC0-1.0 | 6,480 | 250, 500, 512, 1,000, 1,024, 2,048 | 125–256 | 0.1 | 0 |
ds004408 |
Natural speech | auditory & language | CC0-1.0 | 380 | 512 | 128 | 0.0 | 0 |
ds004477 |
Pandemic scenario | learning & reward | CC0-1.0 | 9 | 2,048 | 64 | 15.0 | 0 |
ds004561 |
Agency illusion | learning & reward | CC0-1.0 | 23 | 10,000 | 62 | 0.0 | 0 |
ds004572 |
Hypnosis | resting & consciousness | CC0-1.0 | 516 | 1,000 | 58 | 0.0 | 0 |
ds004582 |
Fake faces | perception & emotion | CC0-1.0 | 73 | 10,000 | 59 | 0.0 | 0 |
ds004603 |
Context trajectory | perception & emotion | CC0-1.0 | 37 | 1,024 | 64 | 0.0 | 0 |
ds004706 |
Spatial memory | memory & cognition | CC0-1.0 | 297 | 2,048 | 128 | 0.0 | 0 |
ds004816 |
RSVP attention 1 | perception & emotion | CC0-1.0 | 20 | 1,000 | 63 | 0.0 | 0 |
ds004817 |
RSVP attention 2 | perception & emotion | CC0-1.0 | 20 | 1,000 | 63 | 0.0 | 0 |
ds004840 |
Music therapy | auditory & language | CC0-1.0 | 51 | 256, 512, 1,024 | 8 | 1.8 | 0 |
ds004902 |
Sleep deprivation | resting & consciousness | CC0-1.0 | 218 | 500, 5,000 | 61 | 0.0 | 0 |
ds004951 |
Braille letters | learning & reward | CC0-1.0 | 23 | 1,000 | 63 | 0.0 | 0 |
ds005089 |
Selective attention | perception & emotion | CC0-1.0 | 33 | 1,000 | 63 | 0.0 | 0 |
ds005095 |
Sternberg | memory & cognition | CC0-1.0 | 48 | 1,000 | 63 | 0.0 | 0 |
ds005121 |
Task not stated | memory & cognition | CC0-1.0 | 39 | 512 | 58 | 0.0 | 0 |
ds005189 |
Recollection | memory & cognition | CC0-1.0 | 30 | 1,000 | 62 | 0.0 | 0 |
ds005262 |
Arabic inner speech | auditory & language | CC0-1.0 | 186 | 250 | 8 | 0.0 | 0 |
ds005273 |
Seen vs unseen | perception & emotion | CC0-1.0 | 32 | 1,000 | 61 | 2.0 | 0 |
ds005505 |
HBN R1 | development (rest & tasks) | CC-BY-SA-4.0 | 1,342 | 500 | 129 | 0.0 | 0 |
ds005506 |
HBN R2 | development (rest & tasks) | CC-BY-SA-4.0 | 1,404 | 500 | 129 | 0.0 | 0 |
ds005507 |
HBN R3 | development (rest & tasks) | CC-BY-SA-4.0 | 1,812 | 500 | 129 | 0.0 | 0 |
ds005508 |
HBN R4 | development (rest & tasks) | CC-BY-SA-4.0 | 3,342 | 500 | 129 | 0.0 | 0 |
ds005509 |
HBN R5 | development (rest & tasks) | CC-BY-SA-4.0 | 3,324 | 500 | 129 | 0.0 | 0 |
ds005510 |
HBN R6 | development (rest & tasks) | CC-BY-SA-4.0 | 1,227 | 500 | 129 | 0.0 | 0 |
ds005511 |
HBN R7 | development (rest & tasks) | CC-BY-SA-4.0 | 3,098 | 500 | 6–129 | 0.0 | 0 |
ds005512 |
HBN R8 | development (rest & tasks) | CC-BY-SA-4.0 | 2,318 | 500 | 129 | 0.0 | 0 |
ds005514 |
HBN R9 | development (rest & tasks) | CC-BY-SA-4.0 | 2,881 | 500 | 129 | 0.0 | 0 |
ds005586 |
Numerosity | perception & emotion | CC0-1.0 | 23 | 1,000 | 60 | 0.0 | 0 |
ds005594 |
Alphabetic decision | perception & emotion | CC0-1.0 | 16 | 1,000 | 64 | 0.0 | 0 |
ds005620 |
Anaesthesia | resting & consciousness | CC0-1.0 | 202 | 5,000 | 62 | 2.3 | 0 |
ds005697 |
Perceive/imagine | perception & emotion | CC0-1.0 | 45 | 1,000 | 62 | 3.5 | 0 |
nm000123 |
Kalunga2016 | SSVEP | CC-BY-4.0 | 30 | 256 | 8 | 0.0 | 0 |
nm000138 |
AlexMI | BCI (MI & P300) | CC-BY-SA-4.0 | 8 | 512 | 16 | 0.0 | 0 |
nm000146 |
Weibo2014 | BCI (MI & P300) | CC0-1.0 | 10 | 200 | 60 | 0.0 | 0 |
nm000158 |
Liu2024 (stroke) | BCI (MI & P300) | CC-BY-4.0 | 100 | 500 | 29–30 | 1.5 | 0 |
nm000172 |
Schirrmeister2017 | BCI (MI & P300) | CC-BY-4.0 | 28 | 500 | 128 | 0.0 | 0 |
nm000173 |
Ofner2017 | BCI (MI & P300) | CC-BY-4.0 | 300 | 512 | 61 | 0.0 | 0 |
nm000181 |
NMT | clinical routine EEG | CC-BY-SA-4.0 | 2,413 | 200 | 21 | 0.0 | 0 |
nm000215 |
BI2014b | BCI (MI & P300) | CC-BY-4.0 | 38 | 512 | 32 | 0.0 | 0 |
nm000216 |
BI2015a | BCI (MI & P300) | CC-BY-4.0 | 129 | 512 | 32 | 0.0 | 0 |
nm000217 |
BI2015b | BCI (MI & P300) | CC-BY-4.0 | 176 | 512 | 32 | 0.0 | 0 |
nm000232 |
THINGS-EEG2 | perception & emotion | CC-BY-4.0 | 319 | 1,000 | 63 | 0.0 | 0 |
nm000238 |
SparrKULee | auditory & language | CC-BY-NC-4.0 | 659 | 1,024 | 64 | 0.0 | 0 |
nm000244 |
BI2014a | BCI (MI & P300) | CC-BY-4.0 | 64 | 512 | 16 | 0.0 | 0 |
nm000245 |
Cho2017 | BCI (MI & P300) | CC0-1.0 | 52 | 512 | 64 | 0.0 | 0 |
nm000250 |
Dreyer2023 | BCI (MI & P300) | CC-BY-4.0 | 1,040 | 512 | 27 | 0.0 | 0 |
nm000266 |
Sosulski2019 | BCI (MI & P300) | CC-BY-SA-4.0 | 2,120 | 1,000 | 31 | 3.0 | 0 |
nm000267 |
Shin2017A | BCI (MI & P300) | GPL-3.0 | 174 | 200 | 30 | 0.0 | 0 |
nm000273 |
Lee2019SSVEP | SSVEP | CC0-1.0 | 108 | 1,000 | 62 | 0.0 | 0 |
nm000323 |
Lee2019ERP | BCI (MI & P300) | CC0-1.0 | 432 | 1,000 | 62 | 2.0 | 0 |
nm000338 |
Lee2019-MI | BCI (MI & P300) | CC0-1.0 | 432 | 1,000 | 62 | 2.0 | 0 |
nm000385 |
Siena | seizure | CC-BY-4.0 | 41 | 512 | 29–31 | 0.0 | 0 |
- Downloads last month
- 4