PAMAP2 Physical Activity Monitoring — 9 Subjects, 18 Activities [3 IMUs + HR]
Abstract
"Wearable IoT dataset with 18 physical activities from 9 subjects wearing 3 IMUs and a heart rate monitor. 54 columns including temperature, acceleration, and gyroscope data. CSV format. Used for HAR, activity classification, and intensity estimation."
Description
Overview
The PAMAP2 Physical Activity Monitoring dataset is a standard benchmark for human activity recognition (HAR) using wearable IoT sensors. Nine subjects performed 18 different activities — from walking and cycling to soccer and rope jumping — while wearing three inertial measurement units (IMUs) placed on the hand, chest, and ankle, plus a heart rate monitor.
The dataset was designed to support algorithm development for activity recognition and intensity estimation, including data processing, segmentation, feature extraction, and classification. Its combination of IMU data with heart rate makes it particularly suitable for comprehensive physiological activity modeling.
Released in 2012 through the UCI Machine Learning Repository, PAMAP2 remains one of the most widely used public datasets in the HAR and IoT healthcare research communities.
Column Schema
| Column | Description |
|---|---|
| timestamp | Unix timestamp in seconds. |
| activityID | Numeric code for the physical activity performed. |
| heart_rate | Heart rate in beats per minute. |
| IMU_hand_temp | Temperature from hand IMU in degrees Celsius. |
| IMU_hand_acc_x/y/z | 3D acceleration from hand IMU (±16g, 13-bit). |
| IMU_chest_acc_x/y/z | 3D acceleration from chest IMU. |
| IMU_ankle_acc_x/y/z | 3D acceleration from ankle IMU. |
| IMU_gyro_x/y/z | Gyroscope readings from each IMU placement. |
| IMU_magnetometer_x/y/z | Magnetometer readings from each IMU placement. |
Key Statistics
- Total Subjects: 9
- Activities: 18 (walking, cycling, rope jumping, soccer, etc.)
- Features: 54 columns per record
- File Format: DAT (space-delimited, CSV-compatible)
- Sampling Rate: 100 Hz (IMU), 1 Hz interpolated (HR)
- Time Period: Donated 2012
Use Cases
- Human activity recognition and classification across 18 activity types
- Physical activity intensity estimation using wearable sensors
- Multi-sensor fusion and feature engineering for IoT health devices
- Transfer learning benchmarking for wearable HAR models
Source & Attribution
PAMAP2 was created by Attila Reiss and Didier Stricker at the German Research Center for Artificial Intelligence (DFKI) and donated to the UCI ML Repository. It has been cited in hundreds of HAR and wearable computing publications.
View Data Structure
To explore column names, data types, and sample rows, visit the official dataset page on UCI.
Preview on UCICite This Dataset
Reiss, Attila, & Stricker, Didier (2012). PAMAP2 Physical Activity Monitoring — 9 Subjects, 18 Activities [3 IMUs + HR]. [Dataset]. UCI Machine Learning Repository. https://archive.ics.uci.edu/dataset/231/pamap2+physical+activity+monitoring
Source: UCI Machine Learning Repository (2012)
Indexed by IoTDataset.com on Apr 10, 2026
Ready to Start Your Research?
Download this dataset directly from the official repository and start building your next breakthrough project.