EmoPhone
three-wave (2020–2022) in-the-wild smartphone + Fitbit affect-sensing dataset with dense ESM labels (KAIST, NeurIPS 2026 E&D)
KAIST three-wave in-the-wild affect dataset: smartphone sensing, Fitbit Inspire HR (1 Hz HR) and dense ESM labels from 282 students, with a three-setting benchmark (temporal, cross-user, cross-wave). Accepted to the NeurIPS 2026 E&D track; data under a credentialed DUA.
- License
CC BY 4.0 (documentation); data files under credentialed access with a data use agreement; Apache-2.0 (benchmark code)
- Type
- Dataset
- Host
- Not reported
- DOI
- Not reported
- Licence
- CC BY 4.0 (documentation); data files under credentialed access with a data use agreement; Apache-2.0 (benchmark code)
- Code & repos
- Access
- Gated / on request
- Subjects
- Not reported
- Signals
- HRActivitysmartphone-sensing
- Devices
- PPG sites
- None
- Setting
- In the wild, Lab
- Wrist
- Not reported
- Raw waveforms
- Not reported
- Updated
- Nov 16, 2026
Original description
This repository contains the documentation, metadata, preprocessing notes, and benchmark scaffolding for EmoPhone (D1, D2, and D3), three consecutive annual waves of an in-the-wild affective sensing study. The project website at emophone.iclab.dev hosts the paper, data-access instructions, and documentation. Each wave pairs passive sensor data from participants' own Android smartphones and Fitbit wearables with dense in-situ affective state labels collected via the Experience Sampling Method (ESM).
Together, the three waves provide 53,139 ESM responses from 297 participants (328 recruited; 9.8% / 11.6% / 7.0% excluded by QC in D1 / D2 / D3), collected between 2020-02-07 and 2022-01-11 in South Korea.
Verbatim fromGitHub· Description from the repository README; licence: NOASSERTION

Key numbers
How it works
No model in the hub reports results on this dataset yet.
- Reference labels
- self report
Devices· 2

Full photo credits and licences are on each device page. An icon means we have no openly licensed or manufacturer-cleared photo of that exact device yet.
Cite
No citation metadata yet.
Cite the ChatPPG record
Cite the dated snapshot 2026-10 for a fixed, reproducible version. Please also cite the original publications listed above.
APA
ChatPPG. (2026). EmoPhone: dataset record (Version 2026-10) [Data set]. ChatPPG Research Hub. https://chatppg.com/research/datasets/emophone
BibTeX
@misc{chatppg_dataset_emophone,
author = {{ChatPPG}},
title = {{EmoPhone: dataset record}},
year = {2026},
month = nov,
version = {2026-10},
howpublished = {\url{https://chatppg.com/research/datasets/emophone}},
note = {ChatPPG Research Hub, Snapshot 2026-10, https://chatppg.com/research/snapshots/2026-10}
}CITATION.cff
cff-version: 1.2.0 message: "If you use this resource, please cite it as below." type: dataset title: "EmoPhone: dataset record" authors: - name: "ChatPPG" url: "https://chatppg.com/research/datasets/emophone" repository-artifact: "https://chatppg.com/research/snapshots/2026-10" version: "2026-10" date-released: "2026-11-16"
History
Record updated
Papers
A cross-subject stress recognition method integrating bidirectional cross-modal multi-path fusion and adaptive adversarial learning (WESAD; in Chinese)Same task: StressSheng Wu Yi Xue Gong Cheng Xue Za Zhi · 2026
FIRMED: A Peak-Centered Multimodal Dataset with Fine-Grained Annotation for Emotion Recognition (Tang et al., ACM MM 2026)Same task: StressACM Multimedia 2026 · 2026
PPG-based affect recognition with long-range deep models: a measurement-driven comparison of CNN, Transformer and Mamba architecturesSame task: StressarXiv 2604.26078 · 2026
Datasets
- BEATSSame task: StressBiomarkers, Experience and Affect from Therapeutic Soundscapes (30 participants × 21 days, Empatica EmbracePlus PPG/EDA/IMU/temp + music listening)PPG · IMU · EDA
EmognitionSame task: Stresswearable emotion film-clip dataset (E4 + Galaxy Watch + Muse 2)43 subjects · PPG · IMU · EDA
LAUREATESame task: Stresslongitudinal university student wellbeing (E4)54 subjects · PPG · IMU · EDA
