Open Accelerometer Datasets
Labelled accelerometer data (i.e. accelerometer data paired with some kind of ground truth, such as annotated activities or energy expenditure) is a crucial resource for researchers in the field of wearable technology and human activity recognition. Such data allows researchers to develop and validate algorithms and machine learning models for activity recognition, walking speed estimation, energy expenditure estimation, and other tasks.
Collecting high-quality labelled accelerometer data can be a challenge. It often requires specialised equipment, careful planning, and a considerable amount of time and resources (e.g. when annotating videos for activities performed). This can be a barrier for researchers who want to explore new ideas or validate their models but do not have the means to collect their own data. Therefore, access to openly available accelerometer datasets can be a valuable resource for the research community.
We found that there is no comprehensive list of openly available accelerometer datasets, which makes it difficult for researchers to find and utilize these resources for their work. As part of the Wearable Landscape, we have therefore started compiling a list of such datasets. These datasets can be used for your research, eliminating the need to collect data yourself. We have gathered important metadata to help you quickly identify any datasets that match your requirements.
On that note, there are two very recent efforts to create new labelled datasets, which we want to highlight:
- WEALTH (https://wealth-stamify.com)
WEALTH collected a large-scale dataset with more than 600 participants across four countries. The data include labelled laboratory and free-living activities using accelerometers on the hip, thigh and wrist. Data collection has finished and data may be shared upon request.
- OxWears (https://doi.org/10.2196/78779)
OxWears aims to collect data from up to 160 healthy adults aged 40 years and older. Notably, recruitment is stratified by age, sex, and BMI groups. Data is collected over 3 days and 4 nights using accelerometers on the wrists, chest, hip, thigh, and ankle. Ground-truth measures include ECG, PSG; annotated video for activity type as well as step count. Data collection is ongoing and expected to conclude in 2026.
⚠️ This list is currently under active development and not exhaustive. We are currently focussing on research-grade accelerometers placed on the wrist, hip, thigh and lower back. The list is maintained by Claas Lendt (claas.lendt@ur.de). If you know of any additional dataset that should be listed here or want to contribute otherwise, please reach out.
🔍 We plan to advance the current list by conducting a comprehensive, systematic search for open datasets. Feel free to get in touch, if you want to be part of this work!
| # | Dataset | Maintainer | Source | Device (sampling frequency and placement) | Setting (protocol) | Outcome (reference measure) |
|---|---|---|---|---|---|---|
| 1 | HARTH | Aleksej Logacjov & Kerstin Bach | GitHub | Axivity AX3 (50Hz; lower back and thigh) | Free-living (non-standardised) | Activity type (annotated video) |
| 2 | HAR70+ | Aleksej Logacjov & Kerstin Bach | GitHub | Axivity AX3 (50Hz; lower back and thigh) | Free-living (semi-standardised) | Activity type (annotated video) |
| 3 | HARChildren | Aleksej Logacjov & Kerstin Bach | GitHub | Axivity AX3 (50Hz; lower back and thigh) | Free-living (semi-standardised and non-standardised) | Activity type (annotated video) |
| 4 | IUWDS | Marta Karas et al. | PhysioNet | ActiGraph GT3X+ (100Hz; wrist and hip) | Free-living (standardised) | Activity type (manually tracked); Steps |
| 5 | CAPTURE-24 | Shing Chan et al. | Oxford University Research Archive | Axivity AX3 (100Hz; wrist) | Free-living (non-standardised) | Activity type (annotated video); Energy expenditure (activity-matched MET) |
| 6 | SENS AccType | Claas Lendt | Zenodo | SENS motion (12.5Hz; thigh) | Laboratory (standardised) + Free-living (non-standardised) | Activity type (annotated video) |
| 7 | Walking Speed | Aleksej Logacjov & Kerstin Bach | GitHub | Axivity AX3 (50Hz; lower back and thigh) | Free-living | Walking speed |
| 8 | TWAGA | Claas Lendt | Zenodo | Axivity AX6 (100 Hz; thigh) | Laboratory (standardised) | Steps (GRF-based); Walking speed (treadmill) |
| 9 | OxWalk | Scott Small et al. | Oxford University Research Archive | Axivity AX3 (25Hz and 100Hz; wrist and hip) | Free-living (non-standardised) | Steps (annotated video) |
| 10 | CATSE3 | Claas Lendt | Zenodo | Axivity AX6 (100Hz; thigh) | Laboratory (standardised) | Energy expenditure (indirect calorimetry) |
| 11 | WSE3 | Patrick Slade | SimTK | NaN (100Hz; thigh) | Laboratory (standardised) | Energy expenditure (indirect calorimetry) |
| 12 | OpenMetabolics | Haedo Cho & Patrick Slade | GitHub | NaN (100Hz; thigh) | Free-living (semi-standardised) | Energy expenditure (indirect calorimetry) |
| 13 | WearGait-PD | Kimberly Kontson et al. | Synapse | Movella/Xsens MTw Awinda (100Hz, thigh, wrist and lower back) | Laboratory (standardised) | Activity type; Steps; Walking speed |
| 14 | EPIC Biomech | Jonathan Camargo | Dropbox | Yost (200Hz; thigh) | Laboratory (standardised) | Activity type; Steps; Walking speed |
| 15 | Mobilise-D TVS | Silvia Del Din et al. | Zenodo | McRoberts DynaPort MM+ IMU (100Hz; lower back); Undefined custom IMU (100Hz; wrist) | Laboratory (standardised) + Free-living (non-standardised) | Steps; Walking Speed |
| 16 | WEAR | Marius Bock et al. | GitHub Pages | Bangle JS (50Hz; wrist) | Free-living (semi-standardised) | Activity type |
| 17 | MHEALTH | Oresti Banos et al. | UCI Machine Learning Repository | Shimmer2 (50Hz; wrist) | Out-of-Laboratory (standardised) | Activity type |
| 18 | DualSleep | Aleksej Logacjov & Kerstin Bach | GitHub | Axivity AX3 (50Hz; lower back and thigh) | Laboratory + Free-living | Sleep stage |
(last update 2026-04-14)
Other datasets
In addition to openly available datasets, there are also other datasets that are not openly available but can be accessed upon request.
- DaLiAc - Daily Life Activities (https://www.mad.tf.fau.de/research/activitynet/daliac-daily-life-activities)