The daily monitoring of ageing population is a current issue which can be effectively tackled by applying daily activity monitoring via smart sensing technology. The purpose of the monitoring is mostly aimed at collecting health conditional related activity awareness and emergency events detection. This is a pilot study that uses low pixel resolution infrared sensors for nonintrusive human activity detection and recognition without body attachments and taking of individual image. In this work, we design and implement a multiple IR sensors system and a serial experiment to verify the availability of applying low-resolution IR data for human activity recognition for both single and multiple target scenarios in the healthcare context. In the experimental setup, the sensor system achieves 82.44% accuracy in general and reaches 100% accuracy rate for some particular activities. The work proves that the low-resolution IR information is an effective metric for human activity monitoring in healthcare applications. Infrared sensors. Human activity detection. Classification methods. Elderly care homes.

Schools in many parts of the world use robots as social peers in order to interact with children and young students for a rich experience. Such use has shown significant enhancement of children's learning. This project uses the humanoid robot NAO which provides object recognition of colours, shapes, typed words, and handwritten digits and operators. The recognition of typed words provides performance of the corresponding movements in the sign language. Five classifiers including neural networks are used for the handwritten recognition of digits and operators. The accuracy of the object recognition algorithms are within the range of 82%-92% when tested on images captured by the robot including the movements which represent words in the sign language. The five classifiers for handwritten recognition produce highly accurate results which are within the range of 87%-98%. This project will serve as a promising provision for an affective touch for children and young students.

Passive radio frequency (RF) sensing and monitoring of human daily activities in elderly care homes is an emerging topic. Micro-Doppler radars are an appealing solution considering their nonintrusiveness, deep penetration, and high-distance range. Unsupervised activity recognition using Doppler radar data has not received attention, in spite of its importance in case of unlabelled or poorly labelled activities in real scenarios. This study proposes two unsupervised feature extraction methods for the purpose of human activity monitoring using Doppler-streams. These include a local Discrete Cosine Transform (DCT)-based feature extraction method and a local entropy-based feature extraction method. In addition, a novel application of Convolutional Variational Autoencoder (CVAE) feature extraction is employed for the first time for Doppler radar data. The three feature extraction architectures are compared with the previously used Convolutional Autoencoder (CAE) and linear feature extraction based on Principal Component Analysis (PCA) and 2DPCA. Unsupervised clustering is performed using K-Means and K-Medoids. The results show the superiority of DCT-based method, entropy-based method, and CVAE features compared to CAE, PCA, and 2DPCA, with more than 5%-20% average accuracy. In regards to computation time, the two proposed methods are noticeably much faster than the existing CVAE. Furthermore, for high-dimensional data visualisation, three manifold learning techniques are considered. The methods are compared for the projection of raw data as well as the encoded CVAE features. All three methods show an improved visualisation ability when applied to the encoded CVAE features.

No abstract

This paper explores the feasibility of using low-resolution infrared (LRIR) image streams for human activity recognition (HAR) with potential application in e-healthcare. Two datasets based on synchronized multichannel LRIR sensors systems are considered for a comprehensive study about optimal data acquisition. A novel noise reduction technique is proposed for alleviating the effects of horizontal and vertical periodic noise in the 2D spatiotemporal activity profiles created by vectorizing and concatenating the LRIR frames. Two main analysis strategies are explored for HAR, including (1) manual feature extraction using texture-based and orthogonal-transformation-based techniques, followed by classification using support vector machine (SVM), random forest (RF), k-nearest neighbor (k-NN), and logistic regression (LR), and (2) deep neural network (DNN) strategy based on a convolutional long short-term memory (LSTM). The proposed periodic noise reduction technique showcases an increase of up to 14.15% using different models. In addition, for the first time, the optimum number of sensors, sensor layout, and distance to subjects are studied, indicating the optimum results based on a single side sensor at a close distance. Reasonable accuracies are achieved in the case of sensor displacement and robustness in detection of multiple subjects. Furthermore, the models show suitability for data collected in different environments.