Activity recognition for elderly care by evaluating proximity to objects and human skeleton data

Abstract

Recently, researchers have shown an increased interest in the detection of activities of daily living (ADLs) for ambient assisted living (AAL) applications. In this study, we present an algorithm that detects activities related to personal hygiene. The approach is based on the evaluation of pose information and a person’s proximity to objects belonging to the typical equipment of bathrooms, such as sink, toilet and shower. In addition to this high-level reasoning, we developed a skeleton-based algorithm that recognises actions using a supervised learning model. Therefore, we analysed several feature vectors, especially with regard to the representation of joint trajectories in the frequency domain. The results gave evidence that this high-level reasoning algorithm can reliably recognise hygiene-related activities. An evaluation of the skeleton-based algorithm shows that the defined actions were successfully classified with a rate of 96.66%.