Elderly Care: Using Deep Learning for Multi-Domain Activity Classification

Abstract

Nowadays, health monitoring issues are increasing as the worldwide population is aging. In this paper, the radar modality is used to classify with radar signature automatically. The classic approach is to extract features from micro-Doppler signatures for classification. This data representation domain has its limitations for activities presenting similar accelerations like a frontal fall and picking up an object from the floor that lead to wrongly labeled activities. In this work, we propose to combine multiple radar data domains with deep learning. Features are extracted from four domains, namely, Range-Time, Range-Doppler, Doppler-Time, and Cadence Velocity Diagram. The extracted features are set as the input of a Convolutional Neural Network, yielding 91% accuracy with 10-fold cross-validation based on the University of Glasgow "Radar signatures of human activities" open dataset.

Continuous Human Activity Recognition With Distributed Radar Sensor Networks and CNN–RNN Architectures

“…In this work, spectrograms are used to capture the various velocity patterns of different body parts in human movements. However, recent works [17], [82], [83] have shown that combining multidomain information for HAR can also be advantageous, e.g., combining the RD, DT, CVD, and RT information.…”

Detecting Elderly Behaviors Based on Deep Learning for Healthcare: Recent Advances, Methods, Real-World Applications and Challenges

“…[46] proposed CNN based angel assistance system (AAS) to improve the fall detection accuracy performance of the AAS for the elderly by minimizing the false positive alerts. The CNN model achieved 98% accuracy with more or less of 17% reduction in the false positive alerts compared to the conventional AAS [51] proposed CNN for multi-domain activity classification in elderly healthcare. The CNN extracts pattern features and classifies based on six different possible activities.…”

Elderly Care - Human Activity Recognition Using Radar with an Open Dataset and Hybrid Maps

“…Up to now, the most common methods of human activity detection are visionbased detection like using cameras and sensor-based detection such as using wearable sensors, radar and smartphone sensors [3,4]. Among all the methods, radar technology outperforms the other for the following aspects [5][6][7][8]. Environmental insensitivity: radar detection is not influenced by harsh light; Contactless Sensing: users do not need to wear or connect with any devices, which provides a high capability of comfort and convenience; Privacy Protection: radar technology collects human activity data without showing their actual images, ensuring the privacy of individuals.…”

“…As machine learning developed, many researchers began to directly consider the grayscale of RGB images of spectrograms as features. Then, convolutional neural networks derived from vision-based classification were applied to those images for classification [3][4][5][6][7]10,22]. Compared with the conventional hand-crafted feature extraction approaches, the use of deep learning technology can increase the accuracy in classification.…”