"EJUST-ADL-2"
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
"The EJUST-ADL-2 dataset (HAD-AW) is a benchmark dataset for human activity recognition (HAR) using wearable inertial sensors. It contains multivariate time-series data collected from a smartwatch worn on the right wrist of 16 participants performing 31 activities of daily living (ADL) in realistic, unsupervised settings. The dataset includes four tri-axial sensor modalities: acceleration, angular velocity, rotation (orientation), and gravity, sampled at 50 Hz, resulting in 12 signal channels. Each activity is segmented into 10-second windows (500 samples), with approximately 160 samples per activity.The dataset covers a wide range of daily behaviors, including eating, working, cooking, sports, hygiene, and household tasks, supporting both coarse- and fine-grained recognition. Additionally, it enables efficient feature representations such as the AMED descriptor, which improves accuracy and reduces computational cost. EJUST-ADL-2 provides a valuable resource for developing and evaluating machine learning and deep learning models for wearable sensing applications."
Fields of Study & Subjects
Context-Aware Activity Recognition Systems
Time Series Analysis and Forecasting
Gait Recognition and Analysis
Publication details
- Journal / source
- IEEE DataPort
- Publication date
- 28 March 2026
- Language
- EN
- Document type
- Dataset