Search

Human activity recognition(HAR) is an important research focus in ubiquitous computing. It has been widely applied in various domains, such as smart homes, healthcare assistance, and sports training. Accurate human activity recognition directly impacts the performance of downstream tasks. Existing methods for human activity recognition primarily rely on extracting temporal or spatial features from the data. These features are used for the task of human activity recognition. The existing methods mainly use CNN or RNN models to model the Euclidean correlations among spatially adjacent sensors or channels. However, non-Euclidean pairwise correlations among all sensors or channels are even critical for accurate classification, which has been ignored by the existing methods. In this paper, we incorporate fine-grained spatial structural data into the model to overcome these limitations. A novel deep learning model for human activity recognition is proposed, which is called the fine-grained data-oriented Spatial-Temporal Graph Transformer network (STGT). The introduction of the STGT model can eliminate the limitations of existing spatial feature extraction methods by leveraging a novel data organization approach proposed in this study. This model enhances the spatial features within the time feature extraction module for effective spatio-temporal feature extraction. We conducted experiments on four large-scale real-world HAR datasets to evaluate its performance. The experimental results demonstrate the superiority of our method over state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

In recent years, deep neural networks (DNNs) have provided high performances for various tasks, such as human activity recognition (HAR), in view of their end-to-end training process between the input data and output labels. However, the performances of the DNNs are highly dependent on the availability of large-scale data for their training processes. In this paper, we propose a novel dataset for the task of HAR , in which the labels are specified for the working environments (WE). Our proposed dataset, namely HARWE , considers multiple signal modalities, including visual signal, audio signal, inertial sensor signals, and biological signals, that are acquired using four different electronic devices. Furthermore, our HARWE dataset is acquired from a large number of participants while considering the realistic disturbances that can occur in the wild. Our HARWE data is context-driven, which means there exist a number of labels in it that even though they are correlated with each other, they have contextual differences. A deep conventional multi-modal neural network provides an accuracy of 99.06% and 68.60%, for the cases of the easy and difficult settings of our dataset, respectively, which indicates its applicability for the task of human activity recognition. • We have proposed a novel dataset for the task of human activity recognition. • Our human activity recognition dataset is specified for the smart workplaces. • The proposed human activity recognition is multi-modal and large-scale. • The labeling process of the proposed dataset is performed in a context-aware manner. [ABSTRACT FROM AUTHOR]

Falling is a significant threat for old people and disabled persons. Therefore, it is necessary to develop an automated fall detection system to make them more independent and to mitigate the physical and psychological consequences of falls. Such a system can lower the death rate among dependent people who live alone. This paper presents a review of vision-based techniques, algorithms and databases used in detecting falls. Recently proposed systems are analyzed and compared. The literature review underscores the promising potential of camera-based fall detection systems and highlights areas for further research and development. [ABSTRACT FROM AUTHOR]

Smart and connected health (SCH) accelerates the development and integration of information science and engineering approaches to support the digital transformation of health and medicine in populated societies. Sensor data based human activity recognition (HAR) as an effective means of SCH is promising for healthcare monitoring and ambient assisted living. This paper focuses on multi-position sensor data fusion enabled HAR, and proposes a multi-channel deep learning framework. The main contributions include: (1) A multi-channel hybrid deep learning model (1DCNN-Att-BiLSTM) that merges a one-dimensional convolutional neural network, a bidirectional long short-term memory model, and an attention mechanism is proposed to significantly improve the recognition performance of HAR models by extracting and selecting local and global behavioral features in the spatial and temporal domains; (2) Publicly accessible datasets Shoaib AR, Shoaib SA, and HAPT are collected and processed to build multi-position sensor data pool for model's evaluation; (3) Thoroughly evaluating the classification performance of seven sensor data fusion patterns from different body positions within the parallel network structure of the multi-channel framework; (4) Comparing the recognition performance metrics of traditional machine learning models, deep learning models, and hybrid models with our proposed model. Extensive experiments demonstrate that our approach achieves competitive performance. [ABSTRACT FROM AUTHOR]

Nowadays, for controlling crime, surveillance cameras are typically installed in all public places to ensure urban safety and security. However, automating Human Activity Recognition (HAR) using computer vision techniques faces several challenges such as lowlighting, complex spatiotemporal features, clutter backgrounds, and inefficient utilization of surveillance system resources. Existing attempts in HAR designed straightforward networks by analyzing either spatial or motion patterns resulting in limited performance while the dual streams methods are entirely based on Convolutional Neural Networks (CNN) that are inadequate to learning the long-range temporal information for HAR. To overcome the above-mentioned challenges, this paper proposes an optimized dual stream framework for HAR which mainly consists of three steps. First, a shots segmentation module is introduced in the proposed framework to efficiently utilize the surveillance system resources by enhancing the lowlight video stream and then it detects salient video frames that consist of human. This module is trained on our own challenging Lowlight Human Surveillance Dataset (LHSD) which consists of both normal and different levels of lowlighting data to recognize humans in complex uncertain environments. Next, to learn HAR from both contextual and motion information, a dual stream approach is used in the feature extraction. In the first stream, it freezes the learned weights of the backbone Vision Transformer (ViT) B-16 model to select the discriminative contextual information. In the second stream, ViT features are then fused with the intermediate encoder layers of FlowNet2 model for optical flow to extract a robust motion feature vector. Finally, a two stream Parallel Bidirectional Long Short-Term Memory (PBiLSTM) is proposed for sequence learning to capture the global semantics of activities, followed by Dual Stream Multi-Head Attention (DSMHA) with a late fusion strategy to optimize the huge features vector for accurate HAR. To assess the strength of the proposed framework, extensive empirical results are conducted on real-world surveillance scenarios and various benchmark HAR datasets that achieve 78.6285%, 96.0151%, and 98.875% accuracies on HMDB51, UCF101, and YouTube Action, respectively. Our results show that the proposed strategy outperforms State-of-the-Art (SOTA) methods. The proposed framework gives superior performance in HAR, providing accurate and reliable recognition of human activities in surveillance systems. [ABSTRACT FROM AUTHOR]

Human activity recognition plays a vital role in various applications like healthcare, sports, and smart environments. ... With the increase in the use of smart wearables and sensors, the concern for privacy has increased drastically in recognizing the activities of individuals. The legacy techniques send the data from millions of users to a central server where the processing and model generation happens. This increases the load on the server. Federated Learning is a promising method to handle sensitive and personal data in an efficient manner as well as use distributed techniques to reduce challenges that are faced in legacy approaches. Edge devices like smartphones and embedded devices extract the raw data, preprocess it, and generate a personalized model. These models are then aggregated at a central server using federated learning methods and their aggregate model is sent back to clients. Long Short-Term Memory (LSTM) is used for model generation as it can effectively capture temporal dependencies in sequential data. In order to show a real-world scenario, an AWS EC2 instance is used as a server, and laptops and smartphones are used as clients. Our results demonstrate that such methods can be adopted and can be used on a much larger scale. The performance of the global and local models is comparable to those attained by conventional methods. The utilization of federated learning shows a significant step forward in the development of human activity recognition. This approach allows for collaborative learning across multiple devices while also ensuring the privacy of users is maintained. [ABSTRACT FROM AUTHOR]

This study investigates the application of Machine Learning (ML) algorithms to classify human physical activities and detect different types of falls using smartphone sensor data. Smartphones equipped with various sensors offer new opportunities for healthcare and human activity monitoring, which are crucial for health and well-being, aligning with the United Nations' sustainable development goals (SDG). The study used the "Sensor Data Collector" Android application to collect smartphone sensor data. The data collection experiments involve ten healthy participants engaged in diverse physical activities and different fall events, utilizing accelerometers and gyroscope sensors. Different types of fall detection include forward, backward, left lateral, and right lateral fall, which are important from a seriousness of health point of view. Participants placed their smartphones in the front pocket of their trousers to ensure a fixed position during the data collection process. The collected datasets are preprocessed using a moving average filter to denoise and smoothen the data. Different statistical features and extracted and ML algorithms, namely Support Vector Machine (SVM), Random Forest (RF), Naive Bayes (NB), and K-Nearest Neighbors (KNN), are employed to create robust classifiers. SVM and RF performed exceptionally well in classifying normal physical activities, achieving a 99.09% accuracy. SVM excelled in daily walking activity classification with a 99.59% accuracy, while the K-NN model performed best for crucial different types of fall classification, achieving an accuracy of 85.71%. This research emphasizes the potential of ML models for accurate physical activity and various types of fall classification, expanding the horizons of smartphone-based healthcare applications and aligning with sustainable development goals. [ABSTRACT FROM AUTHOR]

The rapid expansion of information resources on the internet has made it increasingly challenging for users to get valuable and relevant information. Recommendation systems (RSs) play a vital role in addressing this issue by providing personalized suggestions based on user preferences. However, the open nature of RSs allows for the injection of fake profiles by malicious users, leading to biased ratings and manipulation of the overall system results. These fake profiles have been intended to promote or degrade specific items, which creates bias in the system and compromises the user experience. The motivation behind these fake profiles is to promote or degrade a particular item, which affects the system and makes it more biased toward certain items. This research paper aims to detect and eliminate such fake profiles in RSs, ensuring users receive genuine and reliable results. The proposed approach utilizes a voting ensemble (VE) method, combining the strengths of multiple classification algorithms including Quadratic Discriminant Analysis (QDA), K-Nearest Neighbor (KNN), and Naive Bayes classifier (NBC). By utilizing these classifiers, the system enhances its robustness against malicious activities. The proposed VE method effectively identifies the fake profiles with a minimum number of parameters, and it shows high detection accuracy on Movielens datasets. It has been demonstrated that the proposed method gives 99.4%, 99.5%, 99.4% and 99.6% accuracy using NBC, QDA, KNN, and VE, respectively and it outperforms all other competing methods in terms of accuracy, precision, recall and MCC. [ABSTRACT FROM AUTHOR]

The recognition problem often rises in the classification of multidimensional data and is an integral part of artificial intelligence. Currently, there are many well-known methods and algorithms for solving the recognition problem. However, these methods are mainly aimed at solving such problems where objects are described by a set of independent features. A large number of features describe recognition problems in various fields of human activity. In such problems, the probability of violating the assumption of feature independence is quite high and this determines the relevance of this study. The purpose of this study is to develop a model of recognition algorithms based on the method of making statistical decisions, taking into account the high dimensionality of feature space. A statistical approach to solving the problem of recognition under the conditions of the high dimensionality of feature space is proposed. On the basis of this approach, a model of statistical recognition algorithms based on the construction of two-dimensional threshold rule is proposed. The main advantage of the proposed model is the reduction in the volume of computational operations when recognizing objects in the test set. At that, the reduction of the volume of computational operations is ensured using a number of procedures: selection of representative features; building a two-dimensional threshold rule; determination of a set of reference two-dimensional threshold rules; determination of basic two-dimensional threshold rules and calculation of a probabilistic assessment of the object belonging. This characteristic is very important for real-time recognition systems. The operability of the proposed model of recognition algorithms was investigated in the course of experiments in solving a model problem and the problem of recognizing a person's identity from a signature image. The results of the computational experiment confirm the operability of the proposed approach in solving the recognition problem under conditions when the assumption of feature independence is not satisfied. The results of experimental studies show that the proposed statistical algorithms increase the accuracy of recognition and significantly reduce the volume of computational operations by recognizing an unknown object specified in the space of interrelated features. These algorithms can be used in the compilation of various software systems focused on solving applied recognition problems in the conditions of the high dimensionality of feature space. [ABSTRACT FROM AUTHOR]

Recognition of human actions using machine learning requires extensive datasets to develop robust models. Nevertheless, obtaining real-world data presents challenges due to the costly and time-consuming process involved. Additionally, existing datasets mostly contain indoor videos due to the challenges of capturing pose data outdoors. Synthetic data have been used to overcome these difficulties, yet the currently available synthetic datasets for human action recognition lack photorealism and diversity in their features. Addressing these shortcomings, we develop the NOVAction engine to generate highly diversified and photorealistic synthetic human action sequences. We use NOVAction to create the NOVAction23 dataset comprising 25,415 human action sequences with corresponding poses and labels (available at https://github.com/celikcan-cglab/NOVAction23). In NOVAction23, the performed motions and viewpoints are varied on the fly through procedural generation, to ensure that, for a given action class, each generated sequence features a distinct motion performed by one of the 1,105 synthetic humans captured from a unique viewpoint. Moreover, each synthetic human is unique in terms of body shape (height and weight), skin tone, gender, hair, facial hair, clothing, shoes and accessories. To further increase data diversity, the motion sequences are rendered under various weather conditions and at different times of day, across three outdoor and two indoor settings. We evaluate NOVAction23 by training three state-of-the-art recognizers on it, in addition to the NTU 120 dataset, and corroborating using real-world videos from YouTube. Our results confirm that the NOVAction23 dataset can improve the performance of state-of-the-art human action recognition. [Display omitted] • NOVAction engine automatically generates massively diverse human action data. • NOVAction23, created by NOVAction, novel dataset of human action data. • Features 25,415 unique synthetic human action sequences with poses and labels. • Performed by 1,105 synthetic humans, each captured from a unique viewpoint. • Improved human action recognition performance using state-of-the-art classifiers. [ABSTRACT FROM AUTHOR]

Biometrics is a field that has been given importance in recent years and has been extensively studied. Biometrics can use physical and behavioural differences that are unique to individuals to recognize and identify them. Today, biometric information is used in many areas such as computer vision systems, entrance systems, security and recognition. In this study, a new biometrics database containing silhouette, thermal face and skeletal data based on the distance between the joints was created to be used in behavioural and physical biometrics studies. The fact that many cameras were used in previous studies increases both the processing intensity and the material cost. This study aimed to both increase the recognition performance and reduce material costs by adding thermal face data in addition to soft and behavioural biometrics with the optimum camera. The presented data set was created in accordance with both motion recognition and person identification. Various data loss scenarios and multi-biometrics approaches based on data fusion have been tried on the created data sets and the results have been given comparatively. In addition, the correlation coefficient of the motion frames method to obtain energy images from silhouette data was tested on this dataset and yielded high-accuracy results for both motion and person recognition. [ABSTRACT FROM AUTHOR]

Autonomous systems play a vital role in creating an ambient environment, but for comprehensive human-robot interaction, the system must interpret and identify specific human behaviours. The demand for recognizing human activity from videos has significantly increased across diverse domains, such as medicine, intelligent video surveillance, human-robot interaction, and ambient-assisted environments. Deep learning has emerged as a pioneering technique for human activity recognition from videos due to its ability to learn features from input videos automatically. This research proposes a novel approach for recognizing activities in videos by leveraging the combined power of the VGG-16 convolutional neural network (CNN) and the Long Short-Term Memory (LSTM). To assess the efficiency of our suggested methodology, experiments are conducted on the widely used benchmark dataset UCF-50. The experimental results demonstrate that our VGG-16 and LSTM-based model outperforms state-of-the-art methods for activity recognition, achieving higher accuracy and robustness across various activity categories. [ABSTRACT FROM AUTHOR]

Brain tumors arise from the emergence of abnormal cells in brain tissue and are considered one of the most perilous conditions affecting individuals of all ages, including both children and adults. The disease advances swiftly, and the likelihood of survival diminishes significantly without prompt and adequate treatment. Hence, accurate diagnosis and meticulous treatment planning play a pivotal role in improving the patient's life expectancy. Neurologists and radiologists play a crucial role in the early detection of brain tumors. However, manually identifying and segmenting brain tumors from Magnetic Resonance Imaging (MRI) data poses significant challenges and is susceptible to inaccuracies. The need for an automated brain tumor detection method becomes imperative to achieve early detection of brain tumors. Objective: The objective of the paper is to measure the capability of You Only Look Once version 5 (YOLOv5x) in brain tumor detection in the early stage so that patients can be treated accordingly. Methods: YOLOv5x is examined; Brats (Brain Tumor Segmentation) image and roboflow dataset has been used. Model performance is evaluated using precision, recall rate, F1 score, and Mean Average Precision (mAP). Results: YOLOv5x exhibited precision (98.7), recall (95.6), F1-score (93), mAP at a learning rate of 0.5 (98.4), and the total time taken for implementation of work is 193.20 minutes. Conclusion: YOLOv5x showed improved performance for the detection of brain tumors on dynamic contrast-enhanced MRI when compared with state of art existing work. It is also the fastest and accurate method indicating a greater potential for clinical application. [ABSTRACT FROM AUTHOR]

Human pose estimation and action recognition have important applications in the areas of security, medicine and sports. In this work, we propose an improved YOLOv7-Pose algorithm to solve the problem of human posture and action recognition for fitness movements in various scenarios. For the algorithm based on YOLOv7-Pose, we add the function of classification to the original network. And we introduce Coordinate Attention to the network to improve the model's ability to identify important features of human skeletal joints and action classifications. The improved ConvNeXt network structure is introduced to replace the CBS convolution kernel of the original model to improve the accuracy of human key point detection and action classification of the model. We optimise the spatial pyramid pool structure, which can reduce the loss function and accelerate the convergence rate of the model. We adopt EIOU as the regression function of the target detection frame to improve the accuracy of the coordinate regression. Experiments show that the improved YOLOv7-Pose has an mAP of 95.9% on a homemade test set of fitness actions, which is 5.4% higher than HRNet, with a 4.2% improvement over the original YOLOv7 algorithm, suggesting that the accuracy of action recognition and key point estimation has significantly improved. [Display omitted] [ABSTRACT FROM AUTHOR]

With the maturity of intelligent technology such as human–computer interaction, human action recognition (HAR) technology has been widely used in virtual reality, video surveillance, and other fields. However, the current video-based HAR methods still cannot fully extract abstract action features, and there is still a lack of action collection and recognition for special personnel such as prisoners and elderly people living alone. To solve the above problems, this paper proposes a multidimensional feature fusion network, called P-MTSC3D, a parallel network based on context modeling and temporal adaptive attention module. It consists of three branches. The first branch serves as the basic network branch, which extracts basic feature information. The second branch consists of a feature pre-extraction layer and two multiscale-convolution-based global context modeling combined squeeze and excitation (MGSE) modules, which can extract spatial and channel features. The third branch consists of two temporal adaptive attention units based on convolution (TAAC) to extract temporal dimension features. In order to verify the validity of the proposed network, this paper conducts experiments on the University of Central Florida (UCF) 101 dataset and the human motion database (HMDB) 51 dataset. The recognition accuracy of the proposed P-MTSC3D network is 97.92% on the UCF101 dataset and 75.59% on the HMDB51 dataset, respectively. The FLOPs of the P-MTSC3D network is 30.85G, and the test time is 2.83 s/16 samples on the UCF101 dataset. The experimental results demonstrate that the P-MTSC3D network has better overall performance than the state-of-the-art networks. In addition, a prison action (PA) dataset is constructed in this paper to verify the application effect of the proposed network in actual scenarios. [ABSTRACT FROM AUTHOR]

Systems for recognizing human activities are being developed as part of a larger framework that will allow for continuous monitoring of human behavior in intelligent home environments for elderly care, ambient assisted living, rehabilitation, sports injury detection, surveillance, and entertainment. The most challenging component of the sensor-based human activity detection pipeline for mobile and wearable devices is extracting relevant features. Using mobile and wearable sensors, the most challenging step in the pipeline for recognizing the behaviors of humans is extracting relevant features. The algorithm's performance is affected by feature extraction, which also reduces computation complexity and time. However, existing methods of recognizing human activity rely on hand-crafted characteristics that cannot handle complicated behaviors. There are several known systems for recognizing individual activities, including deep learning, but more must be developed to recognize transitions between actions. To solve these challenges, a new deep learning-based system is proposed for recognizing human actions in this research. Initially, a fuzzy logic-based genetic algorithm (FGA) is used for extracting features from sensor data. The introduced feature selection approach, called PSOACO, aims to enhance the particle swarm optimizer (PSO) algorithm's performance through the ant colony optimizer (ACO) operators. Then, the deep learning approach, named Deep Convolutional Neural Network and Long Short Term Memory (DCNN-LSTM), is used to classify the activities. F1 Score, Recall, Precision, and Accuracy are used to analyze the implementation's effectiveness. The overall recognition accuracy of the proposed model for UCI-HAR, PAMAP2, and WISDM datasets is 99.92%, 99.86%, and 99.94 %. According to the implementation, the HAR with privacy performs better than existing methods done on the Python platform. [ABSTRACT FROM AUTHOR]

The Meta Video Dataset (MetaVD) provides annotated relations between action classes in major datasets for human action recognition in videos. Although these annotated relations enable dataset augmentation, it is only applicable to those covered by MetaVD. For an external dataset to enjoy the same benefit, the relations between its action classes and those in MetaVD need to be determined. To address this issue, we consider two new machine learning tasks: action class relation detection and classification. We propose a unified model to predict relations between action classes, using language and visual information associated with classes. Experimental results show that (i) pre-trained recent neural network models for texts and videos contribute to high predictive performance, (ii) the relation prediction based on action label texts is more accurate than based on videos, and (iii) a blending approach that combines predictions by both modalities can further improve the predictive performance in some cases. • Two proposed tasks aim to predict relations between action classes. • The ground-truth relations are provided from Meta Video Dataset (MetaVD). • Recent pre-trained models in NLP and CV are useful in the tasks. • Action label texts contribute to higher predictive performance than videos. • Using both action label texts and videos can improve the performance. [ABSTRACT FROM AUTHOR]

The recognition of human actions in videos is one of the most active research fields in computer vision. The canonical approach consists in a more or less complex preprocessing stages of the raw video data, followed by a relatively simple classification algorithm. Here we address recognition of human actions using the reservoir computing algorithm, which allows us to focus on the classifier stage. We introduce a new training method for the reservoir computer, based on "Timesteps Of Interest", which combines in a simple way short and long time scales. We study the performance of this algorithm using both numerical simulations and a photonic implementation based on a single non-linear node and a delay line on the well known KTH dataset. We solve the task with high accuracy and speed, to the point of allowing for processing multiple video streams in real time. The present work is thus an important step towards developing efficient dedicated hardware for video processing. [ABSTRACT FROM AUTHOR]

Human Activity Recognition (HAR) plays a crucial role in communication and the Internet of Things (IoT), by enabling vision sensors to understand and respond to human behavior more intelligently and efficiently. Existing deep learning models are complex to deal with the low illumination, diverse viewpoints, and cluttered backgrounds, which require substantial computing resources and are not appropriate for edge devices. Furthermore, without an effective video analysis technique it processes entire frames, resulting inadequate performance. To address these key challenges, a cloud-assisted IoT computing framework is proposed for HAR in uncertain low-lighting environments, which is mainly composed of two tiers: edge and cloud computing. Initially, a lightweight Convolutional Neural Network (CNN) model is developed which is responsible to enhance the low-light frames, followed by the human detection algorithm to process the selective frames, thus enabling efficient resource utilization. Next, these refined frames are then transmitted to the cloud for accurate HAR, where dual stream CNN and transformer fusion network extract both short-and long-range spatiotemporal discriminative features followed by proposed Optimized Parallel Sequential Temporal Network (OPSTN) with squeeze and excitation attention to efficiently learn HAR in complex scenarios. Finally, extensive experiments are conducted over three challenging HAR datasets to deeply examine the proposed framework from various perspectives such as complex activity recognition, lowlighting, etc., where the results are outperformed compared with the state-of-art methods. [ABSTRACT FROM AUTHOR]

With sensors built into smartphones and wearable devices, Human Activity Recognition (HAR) makes it possible to identify regular activities. One of the challenges is to access the motion information from activity data and extract the features from this information. A general model trained without motion features may perform poorly while recognizing the contextual patterns in human activities. In this paper, we proposed the mHAR (motion HAR) model for recognizing static, gradual, and dynamic motion activity to address this challenge. We examine the impact of the proposed model against the general non-convolutional and convolutional recurrent networks in terms of accuracy rates and F1 score. As an experimental dataset, the publicly accessible WISDM was employed. Additionally, the findings of the experiments demonstrate that, in contrast to the unbalanced class problem, motion features play a significant role in activity recognition by achieving accuracy rates of 92.29% and an average F1 score of 83.33%. [ABSTRACT FROM AUTHOR]

Sleep posture is linked to several health conditions such as nocturnal cramps and more serious musculoskeletal issues. However, in-clinic sleep assessments are often limited to vital signs (e.g. brain waves). Wearable sensors with embedded inertial measurement units have been used for sleep posture classification; nonetheless, previous works consider only few (commonly four) postures, which are inadequate for advanced clinical assessments. Moreover, posture learning algorithms typically require longitudinal data collection to function reliably, and often operate on raw inertial sensor readings unfamiliar to clinicians. This paper proposes a new framework for sleep posture classification based on a minimal set of joint angle measurements. The proposed framework is validated on a rich set of twelve postures in two experimental pipelines: computer animation to obtain synthetic postural data, and human participant pilot study using custom-made miniature wearable sensors. Through fusing raw geo-inertial sensor measurements to compute a filtered estimate of relative segment orientations across the wrist and ankle joints, the body posture was characterised in a way comprehensible to medical experts. The proposed sleep posture learning framework offers plug-and-play posture classification by capitalising on a novel kinematic data augmentation method that requires only one training example per posture. Additionally, a new metric together with data visualisations were employed to extract meaningful insights from the postures dataset, demonstrate the added value of the data augmentation method, and explain the classification performance. The proposed framework attained promising overall accuracy as high as 100% on synthetic data and 92.7% on real data, on par with state of the art data-hungry algorithms available in the literature. • Used computer graphics tools to simulate a virtual sleep experiment. • Classification of sleep postures both through in silico and in vivo experiments. • Learning of twelve human sleep postures with one observation per posture. • Inertial sensor fusion enabled posture classification using quadruple joint angles. • Metrics and visualisations enabled quantitative and qualitative posture analysis. [ABSTRACT FROM AUTHOR]

One important aspect of human behavior understanding is the recognition and monitoring of daily activities. An accurate activity recognition system can improve the quality of life in many key areas. The multi-metric feature extraction and DeepForest classifier designed in this paper effectively had solved the problems of incomprehensive feature extraction and insufficient classifier accuracy. First, this paper extracts a total of 450 dimensional feature vectors using mean, variance, maximum, skewness, minimum, kurtosis, regression of independent variables and sample entropy as feature indicators, so that the features of the original vectors can be presented comprehensively. The dimensionality of the feature set is too large, so PCA is used to reduce the dimensionality of the high-dimensional feature vector. After studying the relationship between the dimensionality reduction and the cumulative contribution rate, it is concluded that the cumulative contribution rate reaches 90% when the dimensionality is 15, which can retain the original features better. Then DeepForest was used as the classifier, and the reduced-dimensional feature set was used as the sample set to divide the training set and test set by 3:1. The model was tested on the test set after training, and the classification accuracy of the test set was 98.202%. Six classifiers, GaussianNB, SVM, K-NN, XGBoost, RandomForest and DecisionTree, were selected as the control experimental group, of which only the RandomForest model reached 97%, while the rest of the control models did not achieve 95% effect, indicating that DeepForest was more accurate in classifying human activities. Next, the generalisation ability of the DeepForest classifier was assessed using Monte Carlo Cross Validation (MCCV), K-fold and its confusion matrix. The MCCV validation used 50% of the data as the training set and 30% of the data as the test set, and set the number of splits to 10, resulting in a mean accuracy of 98.227%. The size of K in the K-fold validation was determined to be 8 based on the number of people conducting the human activity experiment, and the final mean accuracy value obtained was 97.790%. The combined confusion matrix from the K-fold validation (which aggregates the confusion matrix for each classification result) was calculated, and the results showed that the highest accuracy reached 100% for A9 and A14 classifications, where A5 and A10, A15 and A10, A12 and A5 were more likely to be confused, with the highest error rate for A5 classification, which was 4.167%, and the rest of the activity classifications were better. [ABSTRACT FROM AUTHOR]

Online monitoring of natural gas in urban underground space has proven to be an effective method for resolving minor leakage and explosion accidents in urban low-pressure natural gas pipelines. However, the harsh environment of urban underground space is often accompanied by the generation of biogas, whose components are more similar to those of natural gas, causing great disturbance to early warning of natural gas leakage. This study correlates the biogas generation principle and its transport law with human activities and environmental changes, analyzes and extracts the characteristics of the natural-gas-biogas CH 4 concentration using big data analysis technology, and establishes a natural-gas-biogas sample database. The imbalance between positive and negative samples in the database is addressed by two oversampling techniques. The models trained by two different machine learning algorithms were then evaluated. The results of the study are summarized as follows: (1) There are daily and annual cycles in the CH 4 concentration in biogas. The daily trend of biogas in the same manhole is similar, and the biogas in different manholes shows various changes periodically with human activities. Biogas early warnings often occur during the high temperature season (April-September). (2) Features such as period, temperature of alarm, and average concentration over 24 h are positive for improving model accuracy. (3) The combined model of XGBoost and the borderlineSMOTE algorithm has an f-score of 72.7 %, an accuracy of 71.2 %, and a recall of 73.4 %. Compared with the traditional manual classification method, the model proposed in this study can identify natural gas and biogas in a more real-time and accurate manner, reduce the workload of on-site confirmation, and effectively shorten the emergency response time. [ABSTRACT FROM AUTHOR]

Human Activity Recognition (HAR) has become a crucial element for smart healthcare applications due to the fast adoption of wearable sensors and mobile technologies. Most of the existing human activity recognition frameworks deal with a single modality of data that degrades the reliability and recognition accuracy of the system for heterogeneous data sources. In this article, we propose a multi-level feature fusion technique for multimodal human activity recognition using multi-head Convolutional Neural Network (CNN) with Convolution Block Attention Module (CBAM) to process the visual data and Convolutional Long Short Term Memory (ConvLSTM) for dealing with the time-sensitive multi-source sensor information. The architecture is developed to be able to analyze and retrieve channel and spatial dimension features through the use of three branches of CNN along with CBAM for visual information. The ConvLSTM network is designed to capture temporal features from the multiple sensors' time-series data for efficient activity recognition. An open-access multimodal HAR dataset named UP-Fall detection dataset is utilized in experiments and evaluations to measure the performance of the developed fusion architecture. Finally, we deployed an Internet of Things (IoT) system to test the proposed fusion network in real-world smart healthcare application scenarios. The findings from the experimental results reveal that the developed multimodal HAR framework surpasses the existing state-of-the-art methods in terms of multiple performance metrics. • Multimodal data fusion for HAR through multi-head architecture, and ConvLSTM. • Retrieving the channel and spatial dimension features via CBAM. • Model deployment in IoT networks for gauging performance in IoHT. • Tests on the UP-Fall detection dataset demonstrate state-of-the-art results. [ABSTRACT FROM AUTHOR]

Educational processes have undergone significant evolution over recent years. The evolution of educational processes is a source of consideration in numerous academic and professional publications. This has been influenced on the one hand, by information and communication technologies, the development of which has affected all areas of human activity; on the other hand, expectations regarding the content of educational processes as well as the emphasis on the development of soft skills have changed significantly. At the interface of these two predictors influencing the evolution of educational processes, the question arose: how often is the use of creative problem-solving methods in educational processes addressed in the literature, and is this trend maintained in the case of processes using advanced ICT? The above determinants influenced the choice of the aim of the research conducted by the authors of this article, which is to identify the range of publications and trends in scientific studies devoted to techniques, tools, and approaches to the implementation of educational processes. [ABSTRACT FROM AUTHOR]

Human intent prediction (HIP) and human activity recognition (HAR) are important for human–robot interactions. However, human–robot interface signals are user-dependent. A classifier trained on labeled source subjects performs poorly on unlabeled target subjects. Besides, previous methods used a single learner, which may only learn a subset of features and degrade their performance on target subjects. Last, HIP and HAR require real-time computing on edge devices whose computational capabilities limit the model size. To address these issues, this paper designs an ensemble diverse hypotheses (EDH) and knowledge distillation (EDHKD) method. EDH mitigates the cross-subject divergence by training feature generators to minimize the upper bound of the classification discrepancy among multiple classifiers. EDH also maximizes the discrepancy among multiple feature generators to learn diverse and complete features. After training EDH, a lightweight student network (EDHKD) distills the knowledge from EDH to a single feature generator and classifier to significantly decrease the model size but remain accurate. The performance of EDHKD is theoretically demonstrated and experimentally validated. Results show that EDH can learn diverse features and adapt well to unknown target subjects. With only soft labels provided by EDH, the student network (EDHKD) can inherit the knowledge learned by EDH and classify unlabeled target data of a 2D moon dataset and two human locomotion datasets with the accuracy at 96.9%, 94.4%, and 97.4%, respectively, in no longer than 1 millisecond. Compared to the benchmark method, EDHKD lifts the target-domain classification accuracy by 1.3% and 7.1% in the two human locomotion datasets. EDHKD also stabilizes learning curves. Therefore, EDHKD significantly increases the generalization ability and efficiency of the HIP and HAR. [Display omitted] • A novel, accurate, and efficient unsupervised cross-subject adaptation method. • Ensemble diverse hypotheses can learn diverse features. • Knowledge distillation can increase both accuracy and efficiency. • Theoretic explanation of the ensemble learners and the knowledge distillation. • Comprehensive cross-subject evaluations on wearable sensor datasets. [ABSTRACT FROM AUTHOR]

• We focus on the data dilemma encountered in the field of human activity understanding, solve practical application problems from a new perspective, and use the proposed method to reduce the model's dependence on data. • We construct a human physical activity dataset containing 10 categories Human SA-10 for use in human activity understanding research. • A Core-Weight Entropy data information evaluation method that can be applied to human behavior recognition tasks is proposed. On Human SA-10, our method can reduce data usage by 50%. Compared to other methods, this method achieved state-of-the-art performance. • In addition, we propose a new unlabeled data redundancy information removal module, which effectively avoids introducing similar data into the training set. With the development of computer vision, the research on human activity understanding has been greatly promoted. The recognition algorithm based on vision transformer has made some achievements in a large number of computer vision tasks, but it still needs to be driven by a large amount of data. How to get rid of the constraints of large amounts of data is crucial for human behavior recognition based on vision transformer. This paper focuses on solving the dilemma of big data, and tries to achieve a high-performance model through a small amount of high information human activity data. The advantage of our work is that by studying feature distribution, we proposed a core weight entropy data information evaluation method for obtaining high information data, and through redundant information elimination strategy, we can avoid introducing similar data. A large number of experimental results show the effectiveness of the proposed method. Compared with existing methods, our method reduces the data consumption by 5% to 30%, and can achieve the performance of using only 50% of 100% data. More importantly, the data our method selected has no redundancy, which is not available in other methods. In addition, we carried out a large number of ablation experiments to prove the rationality of the method. The work of this paper solves the challenge of relying on a large amount of data when using the visual converter to recognize human behavior, which is of practical significance for realizing efficient human activity understanding research with low data. [ABSTRACT FROM AUTHOR]

A new tumor treatment strategy called " Triboelectric Immunotherapy " has been proposed. The therapy is driven by a fabric triboelectric nanogenerator which can convert bio-mechanical energy into pulsed direct current. The pulsed electrical stimulation can directly kill tumor cells, and effectively activate the body's T cell-mediated adaptive immune system, which further inhibits tumor growth, and prolongs the survival of tumor-bearing mice. [Display omitted] Electrical stimulation is a promising technique for drug-free treatment of complex malignant tumors because it can directly kill tumor cells and induce immunotherapy, avoiding the risk of drug-related side effects, toxicity, adverse reactions, drug dependence, drug resistance, etc. Herein, a drug-free tumor therapy strategy, defined as "Triboelectric Immunotherapy", is invented to directly damage tumor cells and recruit immune cells by applying pulsed direct-current (DC) generated from a small size fabric DC triboelectric nanogenerator (DC-TENG). The pulsed DC is generated by coupling triboelectrification effect and electrostatic breakdown effect. The electrical stimulation can promote immunogenic cell death (ICD) of 4T1 cells to release damage-associated molecular patterns that attract dendritic cells to aggregate in tumors and present exposed abundant antigens to T cells, activating T cells-mediated adaptive immunity response to further inhibit tumor growth. Importantly, tissue structures such as blood vessels were preserved during electrical stimulation, allowing cytokines to accumulate to the tumor site through blood supply, significantly inhibiting the growth of 4T1 solid tumors and effectively prolonging survival. This work provides a highly efficient, cost-effective and safe solution for miniaturized wearable tumor treatment system. [ABSTRACT FROM AUTHOR]

• This study examined the relations between parental media behaviors and child's executive function (EF). • Parental problematic smartphone use was positively related to children's EF difficulties. • Technoference mediated the relationship between parental problematic smartphone use and EF. • These relationships were varied between preschooler and elementary school child. Smartphones can affect the relationship between parents and their children, and this technoference of technology has been associated with children's socio-emotional development. Nonetheless, few studies have examined the association between the problematic use of smartphones by parents in relation to children's executive function (EF). The current study thus explored whether parental problematic smartphone use and technoference were associated with children's EF. A total of 472 parents of children aged 3–9 years answered questionnaires that measured their problematic smartphone use, technoference, and children's EF. Results indicated that parental problematic smartphone use was significantly correlated with children's EF difficulties. Furthermore, the positive relationship between parental problematic smartphone use and children's EF problems was partly due to the interference of technology with parent–child activities. Furthermore, the relationships between parental problematic smartphone use, technoference, and children's working memory problems were moderated by the children's age group. These findings emphasize the important influence of parental media use on children's EF. [ABSTRACT FROM AUTHOR]

• A cross-modal and cross-scale fusion module is proposed to perform multimodal feature interaction. • The proposed fusion network can handle different multimodal input combinations and obtain significant performance improvement. • Visualization of multimodal features shows the complementary information learned by the fusion network. • Comparisons with state-of-the-art methods on public benchmarks show the superiority of the proposed method. Human action recognition can benefit from multimodal information to address the classification problem under complex situations. However, existing works either use score fusion or perform simple feature integration methods to combine multiple heterogeneous modalities which failed to effectively utilize multimodal complementary information. In this paper, we proposed a Cross-Scale Cascade Multimodal Fusion Transformer (CSCMFT) to perform interaction and fusion among modalities of multi-scale features, thus obtaining a multimodal complementary representation for RGB-D-based human action recognition. Cross-Modal Cross-Scale Mixer (CCM) is the basic component in CSCMFT, which captures cross-modal relations and propagates the fused information across scales. Furthermore, our CSCMFT can still achieve significant improvements when applied to different multimodal combinations, indicating its generality and scalability. Experimental results show that CSCMFT fully exploits complementary semantic information between RGB and depth maps and outperforms state-of-the-art RGB-D-based methods on NTU RGB+D 60 & 120 and PKU-MMD datasets. [ABSTRACT FROM AUTHOR]

This paper investigates the potential of synthetic data in enhancing the performance of Machine Learning classifiers. Our focus lies primarily on the Human Activity Recognition task, Specifically through wearable device sensors. We analyze the performance of three Generative Adversarial Networks (GANs) and a Diffuse model, considering fidelity, diversity, and generalization metrics. In addition, we assess the relationship between the addition of synthetic samples to the training data and the impact of imbalanced classes on the generative model in the production of synthetic samples. We also introduce a novel GAN designed to generate synthetic samples of time series data from wearable devices. After conducting nearly 400 experiments, our results suggest that synthetic samples can significantly improve the performance of Machine Learning models when real data are scarce. More importantly, our findings underline that data quality precedes the quantity of synthetic data added to training samples. [ABSTRACT FROM AUTHOR]

• Domain adaptation was embedded in BSN fusion framework to overcome multidomain deviation. • Domain adaptation contains evaluation efforts from distribution to domain. • A solution for quantitatively evaluating the domain weights was proposed. • Transfer fusion framework has promising effects and universality for cross-domain fusion in BSNs. Information fusion is a scenario-driven, long-term, challenging task in body sensor networks (BSNs). Recent strategies, such as supervised or semi-supervised learning, rely heavily on fusion-decision models that are supported by prior knowledge or experience. Such techniques typically need a large amount of labeled physiological data and require the target data to have the same distribution as the prior data, which is challenging to achieve in practice. Domain adaptation (DA) is considered a viable solution for the cross-domain problem in BSNs. However, the correlation between the fusion rules and DA was rarely considered in existing studies. The independence of the domain deviation for each sensing source was ignored. This study aims to provide a universal solution for cross-domain information fusion scenarios in BSNs. Firstly, a transfer fusion framework (TF) was proposed to simultaneously solve the problems of the information fusion and multidomain deviation in BSNs by adjusting the correlation between the DA and fusion. Secondly, a DA algorithm based on dynamic domain evaluation (DDE) was proposed for the DA stage of the TF framework. In addition, we provided a feasible solution for quantitatively evaluating the domain weights. Extensive experiments on emotion recognition, fall detection, daily activity recognition, and hand movement recognition have verified the adaptability and promising performance of TF and DDE for cross-domain information fusion in BSNs. The average classification accuracy of the TF is 8.39% and 6.71% higher than that of the conventional fusion transfer framework and the baseline method, respectively. In all transfer tasks, DDE shows the best performance, which is 4.66% higher than the classification accuracy of the optimal comparison method. This study can provide technical support for BSN applications based on cross-domain information fusion. [ABSTRACT FROM AUTHOR]

• Sensor data contribution analysis method based on status frequency-inverse frequency. • Spatial distance matrix for context-awareness of cross-activities. • Novel wide time-domain convolutional neural network (WCNN, powerful parallel feature extraction). • The proposed method can save time and attain comparatively good performance. With the development of artificial intelligence and the broad application of sensors, human activity recognition (HAR) technologies based on noninvasive environmental sensors have received extensive attention and have shown great application value. Owing to the initiative of human activities and machine learning-based methods relying on domain knowledge, obtaining a uniform model to understand the daily behaviors of different residents is difficult. From the perspective of data feature constraints to recognition methods, we constructed a methodology for single user's daily behavior recognition that can adaptively constrain the sensor noise during human activities in multitenant smart home scenarios. We propose a sensor data contribution significance analysis (CSA) method based on the sensor status frequency-inverse type frequency for HAR. This method is employed to measure the contribution of a particular type of sensor to a certain type of behavior recognition. We then build a spatial distance matrix based on the layout of environmental sensors for context-awareness and reducing data noise. Finally, we propose a HAR algorithm based on wide time-domain convolutional neural network and multienvironment sensor data (HAR_WCNN) for daily behavior recognition. Comparative experiment results on the CASAS dataset show that the proposed HAR_WCNN outperforms the compared state-of-the-art methods in terms of HAR accuracy and time consumption. [ABSTRACT FROM AUTHOR]

The primary challenge that smart home users faced is the high energy consumption bills they received every month and they do not know how to optimize the energy consumption of their smart home devices. This increases the country's energy demand and accelerates the greenhouse effect worldwide. Additionally, when users run non-shiftable appliances unevenly, their energy consumption may surpass the maximum permissible power consumption limit. As a result, a short blackout happened in smart homes. To address such challenges, this study builds a mobile application that allows smart home users to control the energy consumption of their smart home appliances effectively. A scheduling task based on particle swarm optimization (PSO) is used to monitor and optimize energy consumption based on human activities in the home. Two experiments were conducted: with and without the implementation of the PSO algorithm. The result showed that the PSO algorithm performs better in optimizing energy consumption. [ABSTRACT FROM AUTHOR]

With the emerging use of machine learning (ML) techniques, there has been particular interest in using wearable data for health economics and outcomes research (HEOR). We aimed to understand the emerging patterns of how ML has been applied to wearable data in HEOR. We identified studies published in PubMed between January 2016 and March 2021. Studies that included at least 1 HEOR-related Medical Subject Headings term, applied an ML, and used wearable data were eligible for inclusion. Two reviewers abstracted information including ML application types and data on which ML was applied and analyzed them using descriptive analyses. A total of 148 studies were identified from PubMed, among which 32 studies met the inclusion criteria. There has been an increase over time in the number of ML studies using wearable data. ML has been more frequently used for monitoring events in real time (78%) than to predict future events (22%). There has been a wide range of outcomes examined, ranging from general physical or mental health (24%) to more disease-specific outcomes (eg, disease incidence [19%] and progression [13%]) and treatment-related outcomes (eg, treatment adherence [9%] and outcomes [9%]). Data for ML models were more often derived from wearable devices with specific medical purposes (60%) than those without (40%). There has been a wide range of applications of ML to wearable data. Both medical and nonmedical wearable devices have been used as a data source, showing the potential for providing rich data for ML studies in HEOR. • The widespread use of wearable devices resulted in a constant flow of individual data related to one's daily activities and certain biometrics. Machine learning (ML) has gained increasing attention in managing large amounts of complex wearable data. Health data collected from wearable devices, benefited by the advanced analyzing technologies such as ML, will likely generate meaningful knowledge in health economics and outcomes research (HEOR). • Our study describes how ML has been applied to data collected from wearable devices in HEOR, which has not been studied yet. ML has not only been applied to monitor general health status but also to monitor or forecast outcomes specific to certain types of disease or treatment. Most studies using devices that were not necessarily designed for medical purposes (eg, smartwatches) have started to be published in relatively recent years. • Our findings suggest a potential for the application of wearable data, coupled with ML techniques, to be expanded to disease- or treatment-specific research in HEOR. The detailed description of the emerging patterns of ML applications with wearable data can address uncertainties in how and when to use ML with wearable data among HEOR researchers, potentially generating additional real-world evidence that can inform treatment and reimbursement decisions. [ABSTRACT FROM AUTHOR]

Human activity recognition (HAR) using drone-mounted cameras has attracted considerable interest from the computer vision research community in recent years. A robust and efficient HAR system has a pivotal role in fields like video surveillance, crowd behavior analysis, sports analysis, and human–computer interaction. What makes it challenging are the complex poses, understanding different viewpoints, and the environmental scenarios where the action is taking place. To address such complexities, in this paper, we propose a novel Sparse Weighted Temporal Attention (SWTA) module to utilize sparsely sampled video frames for obtaining global weighted temporal attention. The proposed SWTA is comprised of two parts. First, temporal segment network that sparsely samples a given set of frames. Second, weighted temporal attention, which incorporates a fusion of attention maps derived from optical flow, with raw RGB images. This is followed by a basenet network, which comprises a convolutional neural network (CNN) module along with fully connected layers that provide us with activity recognition. The SWTA network can be used as a plug-in module to the existing deep CNN architectures, for optimizing them to learn temporal information by eliminating the need for a separate temporal stream. It has been evaluated on three publicly available benchmark datasets, namely Okutama, MOD20, and Drone-Action. The proposed model has received an accuracy of 72.76%, 92.56%, and 78.86% on the respective datasets thereby surpassing the previous state-of-the-art performances by a margin of 25.26%, 18.56%, and 2.94%, respectively. [ABSTRACT FROM AUTHOR]

[Display omitted] Video datasets suffer from huge inter-frame redundancy, which prevents deep networks from learning effectively and increases computational costs. Therefore, several methods adopt random/uniform frame sampling or key-frame selection techniques. Unfortunately, most of the learnable frame selection methods are customized for specific models and lack generality, independence, and scalability. In this paper, we propose a novel two-stage video-to-video summarization method termed FastPicker , which can efficiently select the most discriminative and representative frames for better action recognition. Independently, the discriminative frames are selected in the first stage based on the inter-frame motion computation, whereas the representative frames are selected in the second stage using a novel Transformer-based model. Learnable frame embeddings are proposed to estimate each frame contribution to the final video classification certainty. Consequently, the frames with the largest contributions are the most representative. The proposed method is carefully evaluated by summarizing several action recognition datasets and using them to train various deep models with several backbones. The experimental results demonstrate a remarkable performance boost on Kinetics400, Something-Something-v2, ActivityNet-1.3, UCF-101, and HMDB51 datasets, e.g., FastPicker downsizes Kinetics400 by 78.7% of its size while improving the human activity recognition. [ABSTRACT FROM AUTHOR]

Identification of human actions from video has gathered much attention in past few years. Most of the computer vision tasks such as Health Care Activity Detection, Suspicious Activity detection, Human Computer Interactions etc. are based on the principle of activity detection. Automatic labelling of activity from videos frames is known as activity detection. With the introduction of deep networks, the process of activity detection is clustered into two groups known as hand-crafted feature based approach and automatic feature extraction approach. This paper focuses on various approaches used in recent literature based on traditional and automatic approach. Moreover, hierarchy for different approaches under them such as space based, motion based, genetic based, fuzzy based, dictionary based are discussed. With introduction of Convolutional Neural Networks and Recurrent Neural Networks, automatic learning capability from input modality makes them first choice to be implemented for activity recognition. In this paper various approaches have been analyzed according to methodology, accuracy, classifier and datasets. [ABSTRACT FROM AUTHOR]

Human activity recognition can deduce the behaviour of one or more people from a set of sensor measurements. Despite its widespread applications in monitoring activities, robotics, and visual surveillance, accurate, meticulous, precise and efficient human action recognition remains a challenging research area. As human beings are moving towards the establishment of a smarter planet, human action recognition using ambient intelligence has become an area of huge potential. This work presents a method based on Bi-Convolutional Recurrent Neural Network (Bi-CRNN) -based Feature Extraction and then Random Forest classification for achieving outcomes utilizing Ambient Intelligence that are at the cutting edge of human action recognition for Autonomous Robots. The auto fusion technique used has improved fusion for utilizing and processing data from various sensors. This paper has drawn comparisons with already existing algorithms for Human Action Recognition (HAR) and tried to propose a heuristic and constructive hybrid deep learning-based algorithm with an accuracy of 94.7%. [ABSTRACT FROM AUTHOR]

Time series analysis can facilitate the detection of complex behavioral patterns and potentially provide new opportunities to assess animal welfare. The aim was to investigate whether dairy cows exhibit daily, individual patterns in activity and in area use in the barn. We predicted that behavioral patterns will be more consistent (1) within than between cows, (2) when area categorization is more specific and, thus, allows the detection of individual preferences for areas, and (3) during the night. We conducted the study at an experimental farm with 20 lactating Brown Swiss and Swiss Fleckvieh cows. The animals were housed in cubicles, and they received feed and were milked twice daily. Activity was recorded with IceTag pedometers (IceRobotics Ltd.), and area use with the SMARTBOW sensor system (Zoetis). Data were collected for 55 consecutive days and analyzed at 1-min intervals. To investigate the behavioral time series, we performed a hierarchical clustering analysis. A clustering process calculated distances between days, which were compared within and between cows based on t -tests and analyses of variance. Dendrograms of activity and area use showed that days of individual cows could not be grouped more closely together than those of different cows. A slightly better grouping was achieved with a more specific area categorization, but not during a specific time period. However, the average distances between days were always smaller within (mean ± SD; activity: 95.62 ± 76.88, lying areas: 0.14 ± 0.03, functional areas: 0.12 ± 0.01) than between cows (activity: 109.62 ± 75.33, lying areas: 0.16 ± 0.02, functional areas 0.13 ± 0.01). Considering that the time series of individual cows were slightly but always more similar compared with those between cows, and that more consistent patterns were found when the area categorization was more specific, it can be concluded that the cows exhibited weak individual preferences in area use and also weak daily individual patterns in activity and area use. Because the visual exploratory and empirical approaches used in this study do not account for variability, they do not seem to be suitable for the detection of patterns in animals that display greater plasticity in their temporal structure of activity. Thus, although determining the temporal structure of activity and area use bears the potential to assess the behavior and, in turn, for example, the physiological state and health status of cows, it does not seem to be achievable with a cluster analysis. Therefore, time series methods that account for temporal fluctuations in behavior should be further explored. [ABSTRACT FROM AUTHOR]

Ambient Intelligence is an emerging technology that allows for reasoning of activities that are carried out in a smart environment. Smart sensors play a vital role in activity recognition tasks. Though a great number of sensors provide more pinpointed information about where and when the interaction in the environment occurred, addition of sensors also results in higher power consumption and greater cost constraints. The number of training samples used for activity recognition tasks also increases and becomes computationally expensive. Thus, to overcome these constraints sensor optimization in smart environment is carried out. Mapping of sensors to a user's activity becomes tedious when more than one activity is carried out at the same timestamp. Unambiguous sensor-activity relationship is derived with the help of spatial and temporal context represented by ontology model that captures real time sensor data. Since there exits spatial noise during activity annotation process, it is eliminated with help of pointwise mutual information score. Spatial and temporal data of an activity helps sensor optimization in multi resident living with concurrent activity occurrence. Inactive sensors are removed using mutual information score and redundant sensors are contextually paired based on where an activity happens. By doing so, only unique set of sensors are paired every time even when concurrent activity occurs. The quality of features derived is tested by testing the accuracy of the model by activity recognition task using Random Forest algorithm. Nearly 24% of sensors were identified as inactive in the smart environment. The accuracy of Human Activity Recognition (HAR) task remained constant even after removing the inactive sensors. Addition of spatial context helped to improve HAR with increased accuracy. Contextual pairing of redundant sensors resulted in minimal drop in the model's accuracy. [ABSTRACT FROM AUTHOR]

In the face of urbanization and the widespread use of CCTV cameras, the processing of surveillance videos has gained importance. This study endeavors to create a city-wide monitoring system utilizing human action recognition that can elevate the social sustainability of citizens. The primary goal is to develop an entire framework to detect unusual events within urban environments, with a specific focus on identifying four aberrant actions: "falling," "violence," "loitering," and "intrusion.". The processing of CCTV images is vulnerable to adverse weather conditions, particularly impacting human detection and tracking when obstructions like body parts occlusion, such as during falling events. To address these challenges, the paper proposes tracking compensation techniques that boost the system's ability to detect anomalies without requiring additional training. The proposed approach demonstrates a remarkable 21.21% enhancement in detecting falling events, without compromising its handling of other event types. Overall, the system achieves an impressive average F1 score of 93% across diverse event categories. The system's effectiveness is thoroughly assessed through an extensive subway domain case study, shedding light on its robustness and adaptability for potential real-world deployment. This study also delves into transfer learning dynamics based on sample quantity and pre-training with relevant human-of-interest data. • Developed a CCTV event detection system to enhance social sustainability. • Proposed a method for tracking fallen individuals in adverse weather conditions. • Introduced an filtering method to reduce misclassification of non-human objects. • Conducted an adaptation study of the proposed system in the subway domain. [ABSTRACT FROM AUTHOR]

With the rapid advancement of wearable devices, sensor-based human activity recognition has emerged as a fundamental research area with broad applications in various domains. While significant progress has been made in this research field, energy consumption remains a critical aspect that deserves special attention. Recognizing human activities while optimizing energy consumption is essential for prolonging device battery life, reducing charging frequency, and ensuring uninterrupted monitoring and functionality. The primary objective of this survey paper is to provide a comprehensive review of energy-aware wearable human activity recognition techniques based on wearable sensors without considering vision-based systems. In particular, it aims to explore the state-of-the-art approaches and methodologies that integrate activity recognition with energy management strategies. Finally, by surveying the existing literature, this paper aims to shed light on the challenges, opportunities and potential solutions for energy-aware human activity recognition. [ABSTRACT FROM AUTHOR]

Human Activity Recognition (HAR) using wearable Internet of Things (IoT) devices represents a well investigated researched field encompassing various application domains. Many current approaches rely on cloud-based methodologies for gathering data from diverse users, resulting in the creation of extensive training datasets. Although this strategy facilitates the application of powerful Machine Learning (ML) techniques, it raises significant privacy concerns, which can become particularly severe given the sensitivity of HAR data. Moreover, the labeling process can be extremely time-consuming and even more challenging for IoT wearable devices due to the absence of efficient input systems. In this paper, we address both aforementioned challenges by designing, implementing, and validating edge-based Human Activity Recognition (HAR) systems that operate on resource-constrained IoT devices, which relies on the utilization of Self-Organizing Maps (SOM) for activity detection. We incorporate a feature selection process before training to reduce data dimensionality and, consequently, the SOM size, aligning with the resource limitations of wearable IoT devices. Additionally, we explore the application of Federated Learning (FL) techniques for HAR tasks, enabling new users to leverage SOM models trained by others on their respective datasets. Our federated Extreme Edge (EE)-aware HAR system is implemented on a wearable IoT device and rigorously tested against state-of-the-art and experimental datasets. The results demonstrate that our C++-based SOM implementation achieves a consistent reduction in model size compared to state-of-the-art approaches. Furthermore, our findings highlight the effectiveness of the FL-based approach in overcoming personalized training challenges, particularly in onboarding scenarios. [ABSTRACT FROM AUTHOR]

Preparation of a wearable ACGP hydrogel SERS sensor based on nappies and its analysis of urine collection. [Display omitted] • Ag NWs align Ag NCs, create "hot spots," boost detection sensitivity. • ACGP hydrogel: tough, antimicrobial, biocompatible. • Hydrogel's porous structure adsorbs urine, preserves freshness, enhances test accuracy. • Detect creatinine, uric acid, bilirubin in urine; limits: 0.59 μM, 69 nM, 89 nM. Wearable sensors are essential for continuous monitoring of an individual's health status to enable personalized health management. Urine as a typical non-invasive biological fluid is rich in biological information. Although diaper-based biosensors have been developed, the development of wearable sensors for multiple biomarkers in urine remains a major challenge to be explored. In this paper, a hydrogel SERS sensor consisting of two metal nanostructures of silver nanowires (Ag NWs) and silver nanocubes (Ag NCs) was developed and adhered to nappies to form a wearable nappy-based SERS sensor for the rapid and sensitive detection of bio markers (creatinine, uric acid and bilirubin) and pH in urine. The hydrogel's excellent adsorption and water-locking properties greatly simplify the collection of urine samples. At the same time, its encapsulation effect can effectively maintain the freshness of the urine sample and ensure the accuracy of the test. The introduction of silver nano dual-structure improves the antimicrobial properties of the hydrogel and reduces the risk of infection when the sensor comes into contact with the skin. Moreover, the silver nano dual-structure triggers abundant electromagnetic "hot spots" around intersecting nanojunctions and gaps, which greatly improves SERS sensitivity. For the detection of creatinine, uric acid and bilirubin, the LODs were as low as 0.59 μM, 69 nM, 89 nM. Finally, the urine pH was accurately detected using a functionalized Ag NWs-Ag NCs/CCNF/Gel/PVA hydrogel. This wearable diaper sensor is expected to be used for personal health condition monitoring and holds great promise for effective care of infants, the elderly, and those with limited mobility. [ABSTRACT FROM AUTHOR]

The development of wearable devices for sweat analysis has experienced significant growth in the last two decades, being the main focus the monitoring of athletes health during workouts. One of the main challenges of these approaches has been to attain the continuous monitoring of sweat for time periods over 1 h. This is the main challenge addressed in this work by designing an analytical platform that combines the high performance of potentiometric sensors and a fluidic structure made of a plastic fabric into a multiplexed wearable device. The platform comprises Ion-Sensitive Field-Effect Transistors (ISFETs) manufactured on silicon, a tailor-made solid-state reference electrode, and a temperature sensor integrated into a patch-like polymeric substrate, together with the component that easily collects and drives samples under continuous capillary flow to the sensor areas. ISFET sensors for measuring pH, sodium, and potassium ions were fully characterized in artificial sweat solutions, providing reproducible and stable responses. Then, the real-time and continuous monitoring of the biomarkers in sweat with the wearable platform was assessed by comparing the ISFETs responses recorded during an 85-min continuous exercise session with the concentration values measured using commercial Ion-Selective Electrodes (ISEs) in samples collected at certain times during the session. The developed sensing platform enables the continuous monitoring of biomarkers and facilitates the study of the effects of various real working conditions, such as cycling power and skin temperature, on the target biomarker concentration levels. [ABSTRACT FROM AUTHOR]