EEGdash›NeMAR›NM000105

Iss. 105 · 100 subjects · 100 recordings · CC-BY-NC 4.0

Dataset Brief · FRL Discrete Gestures

NM000105: emg dataset, 100 subjects#

FRL Discrete Gestures: Hand Gesture Recognition from Surface Electromyography

Citation: Patrick Kaifosh, Thomas R. Reardon, CTRL-labs at Reality Labs (2019). FRL Discrete Gestures: Hand Gesture Recognition from Surface Electromyography. 10.82901/nemar.nm000105

100-participant EMG dataset — FRL Discrete Gestures: Hand Gesture Recognition from Surface Electromyography.

Data & curation Patrick Kaifosh · Thomas R. Reardon · CTRL-labs at Reality Labs
Year 2019 · Distributed via NeMAR
Funding Meta Reality Labs

EMG · 16 ch2000 HzBIDS 1.11.0Task · discretegestures

Layer 01Study

What was asked

Hypothesis, independent & dependent variables, paradigm, cohort, and the editorial caveats around what the recordings can and cannot answer.

Layer 02Signal · BIDS

What was recorded

Sidecars, channels & electrodes, coordinate system, event semantics, and quality stats from the NEMAR pipeline when available.

Layer 03Training · ML

What you can train on

Recommended access modes — MNE Raw, braindecode windows, PyTorch DataLoader — plus the targets the metadata makes addressable.

§ 01Access · Get started

Quickstart#

Get Started

Install

pip install eegdash

Access the data

from eegdash.dataset import NM000105

dataset = NM000105(cache_dir="./data")
# Get the raw object of the first recording
raw = dataset.datasets[0].raw
print(raw.info)

Query & Filter

Filter by subject

dataset = NM000105(cache_dir="./data", subject="01")

Advanced query

dataset = NM000105(
    cache_dir="./data",
    query={"subject": {"$in": ["01", "02"]}},
)

Iterate recordings

for rec in dataset:
    print(rec.subject, rec.raw.info['sfreq'])

Cite This Dataset

If you use this dataset in your research, please cite the original authors.

BibTeX

@dataset{nm000105,
  title = {FRL Discrete Gestures: Hand Gesture Recognition from Surface Electromyography},
  author = {Patrick Kaifosh and Thomas R. Reardon and CTRL-labs at Reality Labs},
  doi = {10.82901/nemar.nm000105},
  url = {https://doi.org/10.82901/nemar.nm000105},
}

§ 02Study · The README

About This Dataset#

Dataset: discrete_gestures - Discrete hand gestures from wrist-based surface electromyography

Task: Nine discrete hand gestures (pinches and swipes) Participants: 100 subjects Sessions: 100 total (1 per subject) Publication: Kaifosh et al., 2025 - “A generic non-invasive neuromotor interface for human-computer interaction” (Nature)

This dataset captures wrist-based sEMG signals during prompted discrete hand gestures for navigation and activation tasks. The goal is to enable gesture-based computer control without cameras or visible hand movements, with applications in AR/VR, mobile interfaces, and accessibility.

discrete_gestures: Discrete Hand Gesture Detection from EMG

Overview

Key research objectives: - Generic models that work across users without calibration - Discrete gesture classification with high accuracy - Real-time gesture detection for interactive systems - Robustness to electrode placement variability
View full README
discrete_gestures: Discrete Hand Gesture Detection from EMG

Overview

Key research objectives: - Generic models that work across users without calibration - Discrete gesture classification with high accuracy - Real-time gesture detection for interactive systems - Robustness to electrode placement variability

Dataset Details

Participants

Sample size: 100 participants Demographics: Not available (age, sex, handedness marked as n/a) Recording side: Dominant wrist (assumed right-handed, varies by participant) Sessions: 1 session per participant

Hardware

Device: sEMG Research Device (sEMG-RD) Configuration: Single wristband (dominant wrist) Channels: 16 Sampling rate: 2000 Hz Bit depth: 12 bits Dynamic range: ±6.6 mV Bandwidth: 20-850 Hz Connectivity: Bluetooth Electrode type: Dry gold-plated differential pairs

Gestures

Nine discrete gestures: Thumb swipes (4): - Left swipe - Right swipe - Up swipe - Down swipe

Pinches (4): - Index-to-thumb pinch - Middle-to-thumb pinch - Ring-to-thumb pinch - Pinky-to-thumb pinch

Activation (1): - Thumb tap

Recording Protocol

Participant dons sEMG-RD on dominant wrist

Gesture prompter displays gesture cue (scrolling left-to-right)

Participant performs prompted gesture

Randomized order with randomized inter-gesture intervals

Multiple repetitions of each gesture type

Session duration: Varies by participant Total gestures: 1900 prompted gestures across all participants Stage boundaries: 16 recording stages per session

Data Contents

Files per Session
sub-XXX/ses-XXX/emg/
├── sub-XXX_ses-XXX_task-discretegestures_emg.edf
├── sub-XXX_ses-XXX_task-discretegestures_emg.json
├── sub-XXX_ses-XXX_task-discretegestures_channels.tsv
├── sub-XXX_ses-XXX_task-discretegestures_events.tsv
└── sub-XXX_ses-XXX_electrodes.tsv
Channel Configuration

Total channels: 16 (EMG0-EMG15) Channel naming: Unique identifiers (EMG0-EMG15) Electrode naming: E0-E15 (physical positions) Reference: Bipolar (differential sensing) channels.tsv columns: - name: Channel identifier (EMG0-EMG15) - type: EMG - units: V - signal_electrode: Physical electrode name (E0-E15) - reference: bipolar

electrodes.tsv columns: - name: Electrode identifier (E0-E15) - x, y, z: 3D coordinates (percent units, no decimals)

Events

events.tsv contains: - Gesture prompts: Timestamped prompts for each gesture

type: gesture_X (where X is the gesture name)

latency: Sample index when gesture was prompted

gesture_type: Specific gesture (e.g., “index_pinch”, “thumb_swipe_left”)

Stage boundaries: Recording session phases - type: stage_boundary - stage_name: Stage identifier

Total events: 1916 (1900 gesture prompts + 16 stage boundaries)

Coordinate System

Single coordinate system (no space entity):
EMGCoordinateSystem: Other
EMGCoordinateUnits: percent
X: USP → RSP (0-100%)
Y: Right-hand rule perpendicular (0-100%)
Z: Radial offset (constant 10%)
Anatomical landmarks: - RSP: Radial Styloid Process - USP: Ulnar Styloid Process

Note: Right-handed coordinate system for dominant wrist

Signal Processing

Preprocessing Applied

High-pass filtering: 40 Hz cutoff

Clock drift correction: Time synchronization

Irregular sampling handling: Resampling when deviation >1% (up to 9290% deviation detected)

Signal Characteristics

Gesture patterns: - Patterned activity across channels corresponding to flexor/extensor muscles - Fine differences across gesture instances - Channel activity correlates with muscle positions (Fig. 1 in paper)

Baseline Performance

Published Results (Kaifosh et al., 2025)

Offline Classification (held-out participants): - Accuracy: >90% for gesture classification - False-negative rate improves with more training data - Generic models trained on hundreds of participants

Closed-loop Performance (n=24 naive test users): - First-hit probability: Median improvement from 0.74 (practice) to 0.82 (evaluation block 2) - Gesture completion rate: Median 0.88 gestures/second (evaluation block 2) - Baseline comparison: Gaming controller achieves 1.45 completions/second

Model architecture: 1D convolution → LSTM layers Learning effects: Participants improve from practice to evaluation blocks

Representation Analysis

Network learns: - First layer filters resemble motor unit action potentials (MUAPs) - Deeper layers progressively separate gesture categories - Invariance to nuisance variables (participant ID, electrode placement, signal power)

Confusion Matrix

Common confusions (from paper): - Index and middle holds sometimes released too early - Similar gestures (e.g., adjacent finger pinches) occasionally confused - Swipe directions generally well-separated

Note: Some errors are behavioral (wrong gesture performed) not just decoding errors

Use Cases

Machine Learning

Time series classification: Discrete event detection

Generic modeling: Out-of-the-box cross-user generalization

Representation learning: Physiologically-grounded features

Real-time prediction: Low-latency gesture detection

Applications

Grid navigation: Discrete movement in 2D space

Menu selection: Activation gestures for UI elements

Game control: Gesture-based game inputs

AR/VR interfaces: Hands-free navigation

Accessibility: Alternative input modality

Known Issues and Limitations

By Design

Single wrist: Dominant hand only (not bilateral)

Handedness unknown: Assumed right-handed, varies by participant

Gesture novelty: Users needed coaching to learn effective gestures

No demographic data: Age, sex, handedness not collected

Technical

Electrode placement: Single session per user (less cross-session data than emg2qwerty)

Signal amplitude: Varies with gesture force

Hardware unavailable: sEMG-RD not commercially available

Data Quality

Irregular sampling: High deviation detected (up to 9290%), resampling applied

Behavioral errors: Not all errors are decoder errors (some user mistakes)

Comparison to Baselines

Nintendo Joy-Con controller: - Median: 1.45 completions/second - sEMG decoder: 0.88 completions/second (66% slower)

However: sEMG doesn’t require hand-encumbering device

BIDS Format
Pernet, C.R., et al. (2019). EEG-BIDS, an extension to the brain
imaging data structure for electroencephalography.
Scientific Data, 6(1), 103.
Access and Contact

Original data: Part of Meta Reality Labs neuromotor interface research BIDS conversion: Custom MATLAB tools using EEGLAB BIDS plugin Data curator: Yahya Shirazi, SCCN (Swartz Center for Computational Neuroscience), INC (Institute for Neural Computation), UCSD Contact: See Nature paper for corresponding authors

License

Research and educational use. See original publication.

Citation
Kaifosh, P., Reardon, T.R., & CTRL-labs at Reality Labs. (2025).
A generic non-invasive neuromotor interface for human-computer interaction.
Nature, 645(8081), 702-711. https://doi.org/10.1038/s41586-025-09255-w
Data Curator

Yahya Shirazi SCCN (Swartz Center for Computational Neuroscience) INC (Institute for Neural Computation) University of California San Diego

Version History

v1.0 (2025-10-01): Initial BIDS conversion

BIDS Version: 1.11 | EMG-BIDS: BEP-042 | Updated: Oct 1, 2025

§ 03Cohort · Participants

Cohort#

Dataset Statistics#

Channel counts: 16 ch (n=100 recordings)

Sampling frequencies: 2000.0 Hz (n=100 recordings)

Total recording duration: 63 h

§ 04Signal · Electrodes & trace

Signal · Electrodes & live trace#

Fig. 01 Signal & montage 16 ch · EMG · 2000 Hz · 100 subjects, 100 recordings

Live trace viewer — sub-021 · ses-000 · task-discretegestures

Showing one representative recording out of 100 subjects and 100 recordings in this dataset. Browse the full set on OpenNeuro; drop any other _emg.{set,edf,bdf,vhdr} file onto the viewer (or pass ?emg=<url>) to inspect it.

Electrode layout — EMG · 16 sensors — 16 channels

NEMAR Processing Statistics#

The plots below are generated by NEMAR’s automated EEG pipeline. The histogram shows pipeline success for data cleaning and ICA decomposition, the percentage of data frames and EEG channels retained after artefact removal, line noise per channel (RMS, dB), and the age/gender distribution of participants.

HED event descriptors word cloud

§ 05Manifest · BIDS tree

Manifest#

File Explorer#

Browse the BIDS file structure of this dataset. Records are fetched on demand from the EEGDash catalog the first time you open the explorer.

Recordings—

Files—

Subjects—

Modalities—

Click to load file structure…

§ 06API · Programmatic access

API Reference#

Signature

eegdash.dataset

class

eegdash.dataset.NM000105(cache_dir, query=None, s3_bucket=None, **kwargs)

Bases: EEGDashDataset

Author (year)Kaifosh2025

Canonical—

Importable asNM000105 · Kaifosh2025

Sourceeegdash/dataset/registry.py · [source ↗]

class eegdash.dataset.NM000105(cache_dir: str, query: dict | None = None, s3_bucket: str | None = None, **kwargs)[source]#

FRL Discrete Gestures: Hand Gesture Recognition from Surface Electromyography

Study:: nm000105 (NeMAR)
Author (year):: Kaifosh2025
Canonical:: —

Also importable as: NM000105, Kaifosh2025.

Modality: emg. Subjects: 100; recordings: 100; tasks: 1.

Parameters:

cache_dir (str | Path) – Directory where data are cached locally.
query (dict | None) – Additional MongoDB-style filters to AND with the dataset selection. Must not contain the key dataset.
s3_bucket (str | None) – Base S3 bucket used to locate the data.
**kwargs (dict) – Additional keyword arguments forwarded to EEGDashDataset.

data_dir#

Local dataset cache directory (cache_dir / dataset_id).

Type:: Path

query#

Merged query with the dataset filter applied.

Type:: dict

records#

Metadata records used to build the dataset, if pre-fetched.

Type:: list[dict] | None

Notes

Each item is a recording; recording-level metadata are available via dataset.description. query supports MongoDB-style filters on fields in ALLOWED_QUERY_FIELDS and is combined with the dataset filter. Dataset-specific caveats are not provided in the summary metadata.

References

OpenNeuro dataset: https://openneuro.org/datasets/nm000105 NeMAR dataset: https://nemar.org/dataexplorer/detail?dataset_id=nm000105 DOI: https://doi.org/10.82901/nemar.nm000105

Examples

>>> from eegdash.dataset import NM000105
>>> dataset = NM000105(cache_dir="./data")
>>> recording = dataset[0]
>>> raw = recording.load()

__init__(cache_dir: str, query: dict | None = None, s3_bucket: str | None = None, **kwargs)[source]#

save(path: str, overwrite: bool = False, offset: int = 0)[source]#

Save datasets to files by creating one subdirectory for each dataset:

path/
    0/
        0-raw.fif | 0-epo.fif
        description.json
        raw_preproc_kwargs.json (if raws were preprocessed)
        window_kwargs.json (if this is a windowed dataset)
        window_preproc_kwargs.json  (if windows were preprocessed)
        target_name.json (if target_name is not None and dataset is raw)
    1/
        1-raw.fif | 1-epo.fif
        description.json
        raw_preproc_kwargs.json (if raws were preprocessed)
        window_kwargs.json (if this is a windowed dataset)
        window_preproc_kwargs.json  (if windows were preprocessed)
        target_name.json (if target_name is not None and dataset is raw)

Parameters:

path (str) –

Directory in which subdirectories are created to store
-raw.fif | -epo.fif and .json files to.
overwrite (bool) – Whether to delete old subdirectories that will be saved to in this call.
offset (int) – If provided, the integer is added to the id of the dataset in the concat. This is useful in the setting of very large datasets, where one dataset has to be processed and saved at a time to account for its original position.

Access modesMNE → braindecode → PyTorch → ML

.rawMNE Raw object — standard tools (filter, epoch, ICA, plot_psd).mne

BaseConcatDatasetEach record is a lazy BaseDataset from braindecode — windowed via create_windows_from_events.braindecode

DataLoaderWraps the windowed dataset into a PyTorch DataLoader; supports parallel workers and on-the-fly augmentations.pytorch

Zarr cacheOptional braindecode Zarr mirror for fast resume; persisted to cache_dir.zarr

Hugging FaceNo per-dataset mirror published yet — browse the EEGDash org listing for sibling datasets. See the datasets loader API.huggingface

Croissant 1.0Machine-readable JSON-LD descriptor — NM000105.croissant.json (MLCommons schema, ingestible by PyTorch / TensorFlow / JAX).mlcommons

Examples using EEGDashcurated · start here

Find datasets with the EEGDash APIQuery the catalogue, filter by task or modality, list candidates.

Load one EEG recordingResolve a single record to an MNE Raw with channels and events.

EEG recording to PyTorch DataLoaderWrap braindecode windows in a DataLoader for model training.

Preprocess EEG and create windowsFilter, resample, epoch — and persist the windowed dataset.

Save and reload prepared dataCache a windowed dataset to disk and reattach it without recompute.

Download a dataset locallyPrefetch BIDS files to a local cache and validate the layout.

Swap any load_dataset(...) call for nm000105 to reproduce the tutorial on this dataset.

Citation

Patrick Kaifosh, Thomas R. Reardon, CTRL-labs at Reality Labs (2019). FRL Discrete Gestures: Hand Gesture Recognition from Surface Electromyography. 10.82901/nemar.nm000105

Provenance

¹Contributed to nemar in BIDS format.

²Curated & ingested by the EEGDash catalog; see CITATION.cff for canonical reference.

³Persistent identifier: 10.82901/nemar.nm000105.

Related & sibling datasets

NM000107EMG · 100 subjSame authors NM000106EMG · 100 subjSame authors NM000159EMG · 16 subj NM000104EMG · 108 subj NM000155EMG · 6 subj

+ 1 more — see See Also below →

BIDS

BIDS 1.11.0

Sidecars

events · channels · electrodes

Provenance

CC-BY-NC 4.0 · 10.82901/nemar.nm000105

Machine-readable

schema.org/Dataset · Croissant

Mirrors

OpenNeuro · NEMAR · HF org

Dataset ID	`NM000105`
Title	FRL Discrete Gestures: Hand Gesture Recognition from Surface Electromyography
Author (year)	`Kaifosh2025`
Canonical	—
Importable as	`NM000105`, `Kaifosh2025`
Year	2019
Authors	Patrick Kaifosh, Thomas R. Reardon, CTRL-labs at Reality Labs
License	CC-BY-NC 4.0
Citation / DOI	10.82901/nemar.nm000105
Source links	OpenNeuro \| NeMAR \| Source URL

NM000105: emg dataset, 100 subjects#

Quickstart#

About This Dataset#

Cohort#

Dataset Statistics#

Signal · Electrodes & live trace#

NEMAR Processing Statistics#

Manifest#

File Explorer#

API Reference#

Citation

Provenance

Related & sibling datasets

See Also#