Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CAPTURE-24: A large dataset of wrist-worn activity tracker data collected in the wild for human activity recognition

Shing Chan, Hang Yuan, Catherine Tong, Aidan Acquah, Abram Schonfeldt, Jonathan Gershuny, Aiden Doherty

TL;DR

CAPTURE-24 addresses the generalization gap of laboratory HAR datasets by releasing a large-scale, in-the-wild wrist-worn accelerometer dataset with wearable-camera-derived ground truth and sleep diaries from 151 participants (3883 hours total; 2562 hours annotated). Data are processed into fixed 10-second windows and annotated using the Compendium of Physical Activities (CPA) with multiple mapping schemas, enabling both activity-level and intensity-level analyses. Benchmarking across RF, XGBoost, CNN, RNN, and HMM-based approaches shows temporal modelling substantially improves HAR performance, with CNN+HMM often providing the best overall results, and highlights the critical impact of dataset size on deep learning performance. The dataset, released under CC BY 4.0 with privacy protections (excluding image data), aims to enable data-hungry methods, improve generalization to real-world activity patterns, and guide future research directions in multimodal and open-set HAR.

Abstract

Existing activity tracker datasets for human activity recognition are typically obtained by having participants perform predefined activities in an enclosed environment under supervision. This results in small datasets with a limited number of activities and heterogeneity, lacking the mixed and nuanced movements normally found in free-living scenarios. As such, models trained on laboratory-style datasets may not generalise out of sample. To address this problem, we introduce a new dataset involving wrist-worn accelerometers, wearable cameras, and sleep diaries, enabling data collection for over 24 hours in a free-living setting. The result is CAPTURE-24, a large activity tracker dataset collected in the wild from 151 participants, amounting to 3883 hours of accelerometer data, of which 2562 hours are annotated. CAPTURE-24 is two to three orders of magnitude larger than existing publicly available datasets, which is critical to developing accurate human activity recognition models.

CAPTURE-24: A large dataset of wrist-worn activity tracker data collected in the wild for human activity recognition

TL;DR

CAPTURE-24 addresses the generalization gap of laboratory HAR datasets by releasing a large-scale, in-the-wild wrist-worn accelerometer dataset with wearable-camera-derived ground truth and sleep diaries from 151 participants (3883 hours total; 2562 hours annotated). Data are processed into fixed 10-second windows and annotated using the Compendium of Physical Activities (CPA) with multiple mapping schemas, enabling both activity-level and intensity-level analyses. Benchmarking across RF, XGBoost, CNN, RNN, and HMM-based approaches shows temporal modelling substantially improves HAR performance, with CNN+HMM often providing the best overall results, and highlights the critical impact of dataset size on deep learning performance. The dataset, released under CC BY 4.0 with privacy protections (excluding image data), aims to enable data-hungry methods, improve generalization to real-world activity patterns, and guide future research directions in multimodal and open-set HAR.

Abstract

Existing activity tracker datasets for human activity recognition are typically obtained by having participants perform predefined activities in an enclosed environment under supervision. This results in small datasets with a limited number of activities and heterogeneity, lacking the mixed and nuanced movements normally found in free-living scenarios. As such, models trained on laboratory-style datasets may not generalise out of sample. To address this problem, we introduce a new dataset involving wrist-worn accelerometers, wearable cameras, and sleep diaries, enabling data collection for over 24 hours in a free-living setting. The result is CAPTURE-24, a large activity tracker dataset collected in the wild from 151 participants, amounting to 3883 hours of accelerometer data, of which 2562 hours are annotated. CAPTURE-24 is two to three orders of magnitude larger than existing publicly available datasets, which is critical to developing accurate human activity recognition models.
Paper Structure (18 sections, 5 figures, 8 tables)

This paper contains 18 sections, 5 figures, 8 tables.

Table of Contents

  1. Accelerometer
  2. Wearable Camera
  3. Data Annotation
  4. Data Extraction
  5. De-identification
  6. Challenges in Activity Recognition in the Wild
  7. Performance against Dataset Size
  8. Limitations.
  9. Research directions.
  10. List of hand-crafted features
  11. Hyperparameter tuning details
  12. Baseline architecture
  13. Data augmentation
  14. Optimization
  15. Computational resources
  16. ...and 3 more sections

Figures (5)

  • Figure 1: Overview of the creation of the CAPTURE-24 Dataset. Recruited subjects wore an activity tracker for roughly 24 hours. They also wore a camera during daytime and used a diary to register their sleep times during nighttime. The collected data was processed and harmonised to obtain acceleration time-series data annotated with the activities performed. CPA: Compendium of Physical Activities; MET: metabolic equivalent. Note that the camera images are not part of the dataset release.
  • Figure 2: Confusion matrix for random forest + Hidden Markov Model
  • Figure 3: F1-score as a function of dataset size for physical activity classification
  • Figure 4: Top 10 most frequent Compendium of Physical Activities code annotations found in Capture-24
  • Figure 5: Distribution of activities different labelling schema