While the sensor-based recognition of Activities of Daily Living (ADLs) is a well-established research area, few high-quality labeled datasets are available to compare the results of different approaches. This is especially true for multi-inhabitant settings, where multiple residents live in the same home performing both individual and collaborative ADLs. The reference multi-inhabitant datasets consider only environmental sensors data and two residents in the same home. In this paper, we present MARBLE: a novel multi-inhabitant ADLs dataset that combines both smart-watch and environmental sensors data. MARBLE includes sixteen hours of ADLs considering scripted but realistic scenarios where up to four subjects live in the same home environment. Twelve volunteers participated in data collection. We describe MARBLE also providing details on the design of data collection and tools. We also present initial benchmarks of ADLs recognition on MARBLE, obtained by applying state-of-the-art deep learning methods. Our goal is to share the result of a complex and time consuming data acquisition and annotation task, hoping that the challenge of improving the current baselines on MARBLE will contribute to the progress of the research in multi-inhabitant ADLs recognition.
CITATION STYLE
Arrotta, L., Bettini, C., & Civitarese, G. (2022). The MARBLE Dataset: Multi-inhabitant Activities of Daily Living Combining Wearable and Environmental Sensors Data. In Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST (Vol. 419 LNICST, pp. 451–468). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-94822-1_25
Mendeley helps you to discover research relevant for your work.