Search

Background: The ActiPASS software was developed from the open-source Acti4 activity classification algorithm for thigh-worn accelerometry. However, the original algorithm has not been validated in children or compared with a child-specific set of algorithm thresholds. This study aims to evaluate the accuracy of ActiPASS in classifying activity types in children using 2 published sets of Acti4 thresholds. Methods: Laboratory and free-living data from 2 previous studies were used. The laboratory condition included 41 school-aged children (11.0 [4.8] y; 46.5% male), and the free-living condition included 15 children (10.0 [2.6] y; 66.6% male). Participants wore a single accelerometer on the dominant thigh, and annotated video recordings were used as a reference. Postures and activity types were classified with ActiPASS using the original adult thresholds and a child-specific set of thresholds. Results: Using the original adult thresholds, the mean balanced accuracy (95% CI) for the laboratory condition ranged from 0.62 (0.56–0.67) for lying to 0.97 (0.94–0.99) for running. For the free-living condition, accuracy ranged from 0.61 (0.48–0.75) for lying to 0.96 (0.92–0.99) for cycling. Mean balanced accuracy for overall sedentary behavior (sitting and lying) was ≥0.97 (0.95–0.99) across all thresholds and conditions. No meaningful differences were found between the 2 sets of thresholds, except for superior balanced accuracy of the adult thresholds for walking under laboratory conditions. Conclusions: The results indicate that ActiPASS can accurately classify different basic types of physical activity and sedentary behavior in children using thigh-worn accelerometer data. [ABSTRACT FROM AUTHOR]

Traditional systems for indoor pressure sensing and human activity recognition (HAR) rely on costly, high-resolution mats and computationally intensive neural network-based (NN-based) models that are prone to noise. In contrast, we design a cost-effective and noise-resilient pressure mat system for HAR, leveraging Velostat for intelligent pressure sensing and a novel hyperdimensional computing (HDC) classifier that is lightweight and highly noise resilient. To measure the performance of our system, we collected two datasets, capturing the static and continuous nature of human movements. Our HDC-based classification algorithm shows an accuracy of 93.19%, improving the accuracy by 9.47% over state-of-the-art CNNs, along with an 85% reduction in energy consumption. We propose a new HDC noise-resilient algorithm and analyze the performance of our proposed method in the presence of three different kinds of noise, including memory and communication, input, and sensor noise. Our system is more resilient across all three noise types. Specifically, in the presence of Gaussian noise, we achieve an accuracy of 92.15% (97.51% for static data), representing a 13.19% (8.77%) improvement compared to state-of-the-art CNNs.

Human Activity Recognition (HAR) is pivotal in ubiquitous computing, offering benefits to human-centric services such as health monitoring, smart homes, and eldercare systems. HAR leverages smartphones, smartwatches, and other wearable devices to collect sensory data annotated with activity labels, which are then used to train machine learning or deep learning models for automatic activity recognition. Effective HAR systems must integrate information from multiple modalities to accurately assist users. This paper introduces the XAttentionHAR model, an innovative cross-modality attention-based ensemble model for HAR. Our approach utilizes a self-attention module to extract features within each modality and an inter-domain cross-attention module to capture and integrate long-term dependencies across domains. The cross-modality attention mechanism enhances the fusion of diverse modalities, enriching the semantic information. We conducted extensive experiments on the WISDM public dataset, which includes accelerometer and gyroscope data from smartwatches and smartphones. Our results demonstrate that XAttentionHAR outperforms other state-of-the-art methods in activity recognition, achieving 98.48% accuracy for smartphone-based HAR and 98.73% accuracy for smartwatch-based HAR, paving the way for improved human-centric services. [ABSTRACT FROM AUTHOR]

In recent years, railway construction has been plagued by frequent safety accidents, with workers’ safety during the construction process remaining a major concern. To mitigate this issue, intelligent monitoring of railway workers’ activity has been proposed as a means of improving the safety coefficient of construction. Human activity recognition (HAR) based on wearable devices holds significant application value in areas such as health monitoring, motion analysis, and intelligent assistance. Recently, convolutional neural networks (CNNs) have gained extensive adoption and demonstrated outstanding performance in HAR. However, current HAR research still faces some challenges, including problems with establishing spatial–temporal dependencies and addressing the demand for lightweight models. To address the above issues, we propose a lightweight dual-stream convolution model (LDSC) based on deformable convolution and hierarchical segmentation. The model adaptively captures significant variations in sensor readings over time from portable cards of railway personnel through a temporal stream and learns the interactive information among sensor channels over a spatial stream. LDSC consists of three lightweight convolutional modules that combine deep convolution and point convolution to reduce model parameters, thus meeting the demand for a lightweight model. Experiments and ablation studies are conducted on three available datasets (UCI-HAR, UniMiB-SHAR, and WISDM) to evaluate the proposed model. The experimental results indicate that our model outperforms existing state-of-the-art methods in terms of recognition accuracy, validating the effectiveness and feasibility of LDSC. In addition, theoretical analysis and ablation experiments demonstrate that the proposed LDSC embodies lightweight characteristics. [ABSTRACT FROM AUTHOR]

In order to solve the problem of low accuracy and efficiency in video human behavior recognition algorithm, a deep learning video human behavior recognition algorithm is proposed, which is based on an improved time division network. This method innovates on the classical two-stream convolutional neural network framework, and the core is to enhance the performance of the time division network by implementing the sliding window sampling technique with multiple time scales. This sampling strategy not only effectively integrates the full time-series information of the video, but also accurately captures the long-term dependencies hidden in human behavior, which further improves the accuracy and efficiency of behavior recognition. Experimental results show that the method proposed in this paper has achieved good advantages in multiple data sets. On HMDB51, our method achieves 84% recognition accuracy, while on the more complex Kinetics and UCF101 datasets, it also achieves 94% and significant recognition results, respectively. In the face of complex scenes and changeable human body structure, the proposed algorithm shows excellent robustness and stability. In terms of real-time, it can meet the high requirements of real-time video processing. Through the validation of experimental data, our method has made significant progress in extracting spatiotemporal features, capturing long-term dependencies, and focusing on key information. [ABSTRACT FROM AUTHOR]

AbstractThe ability to synthesize realistic data in a parametrizable way is valuable for a number of reasons, including privacy, missing data imputation, and evaluating the performance of statistical and computational methods. When the underlying data generating process is complex, data synthesis requires approaches that balance realism and simplicity. In this paper, we address the problem of synthesizing sequential categorical data of the type that is increasingly available from mobile applications and sensors that record participant status continuously over the course of multiple days and weeks. We propose the paired Markov Chain (paired-MC) method, a flexible framework that produces sequences that closely mimic real data while providing a straightforward mechanism for modifying characteristics of the synthesized sequences. We demonstrate the paired-MC method on two datasets, one reflecting daily human activity (time use) patterns collected via a smartphone application, and one encoding the intensities of physical activity measured by wearable accelerometers. In both settings, sequences synthesized by paired-MC better capture key characteristics of the real data than alternative approaches. Supplemental materials for this article are available online. [ABSTRACT FROM AUTHOR]

Real-time lower limb movement resistance monitoring is critical for various applications in clinical and sports settings, such as rehabilitation and athletic training. Current methods often face limitations in accuracy, computational efficiency, and generalizability, which hinder their practical implementation. To address these challenges, we propose a novel Mobile Multi-Task Learning Network (MMTL-Net) that integrates MobileNetV3 for efficient feature extraction and employs multi-task learning to simultaneously predict resistance levels and recognize activities. The advantages of MMTL-Net include enhanced accuracy, reduced latency, and improved computational efficiency, making it highly suitable for real-time applications. Experimental results demonstrate that MMTL-Net significantly outperforms existing models on the UCI Human Activity Recognition and Wireless Sensor Data Mining Activity Prediction datasets, achieving a lower Force Error Rate (FER) of 6.8% and a higher Resistance Prediction Accuracy (RPA) of 91.2%. Additionally, the model shows a Real-time Responsiveness (RTR) of 12 ms and a Throughput (TP) of 33 frames per second. These findings underscore the model's robustness and effectiveness in diverse real-world scenarios. The proposed framework not only advances the state-of-the-art in resistance monitoring but also paves the way for more efficient and accurate systems in clinical and sports applications. In real-world settings, the practical implications of MMTL-Net include its potential to enhance patient outcomes in rehabilitation and improve athletic performance through precise, real-time monitoring and feedback. [ABSTRACT FROM AUTHOR]

Smart wearable devices detection and recording of people's everyday activities is critical for health monitoring, helping persons with disabilities, and providing care for the elderly. Most of the research that is being conducted uses a machine learning-based methodology; however, these approaches frequently have issues with high computing resource consumption, burdensome training data gathering, and restricted scalability across many contexts. This research suggests a behaviour detection technology based on multi-source sensing and logical reasoning to address these problems. In order to realize the natural fusion of signal processing and logical reasoning in behavior recognition research, this work designs a lightweight behavior recognition solution using the pertinent theories of ontology reasoning in classical artificial intelligence. Machine learning technology is also employed for behavior recognition using the same data set. Once the best model has been chosen, the cross-person recognition results after testing and modification of parameters are 90.8% and 92.1%, respectively. This technology was used to create a behaviour recognition system, and several tests were run to assess how well it worked. The findings demonstrate that the suggested strategy achieves over 90% recognition accuracy for 11 different daily activities, including jogging, walking, and stair climbing. Additionally, the suggested strategy dramatically minimises the quantity of user-provided training data needed in comparison to machine learning-based behaviour identification techniques. [ABSTRACT FROM AUTHOR]

Existing datasets used for human activity recognition (HAR) usually suffer from limitations in terms of size variation and distribution of activity classes, which can impair the generalizability and robustness of the trained model, especially in the case of activity classes with minority data. This paper proposes an architecture utilizing bi-directional generative adversarial networks (Bi-GANs) beside fast Fourier transform, which stacks the 1-D accelerometer signals as m frequency bins × frames × 3 orientations and produces an RGB-based features pattern. The extracted patterns allow the 2D-CNN-based Bi-GAN architecture to learn the accelerometer signals' cross-axis relationships, which served as input for training a deep learning model for activity recognition equipped with a fuzzy inference dense layer. Also, we have used Hidden Markov Models (HMMs) for post-processing the classifier's output, which integrates the window-level decision in more extended periods, obtaining a significant performance improvement. The proposed method examined and conducted on MobiAct, Up-Fall, Opportunity, and WISDM datasets with different rates of augmentation, reaching to the 99.7%, 99.96%, 86.8%, and 99.12% rate of accuracy, respectively. [ABSTRACT FROM AUTHOR]

Violence recognition is challenging because it can be presented in very different forms. For example, it can be present in an image by a person hitting another person or present in audio by a person being rude to another. Thus, audio and video are essential features to be analysed. In the audio approach, speech processing, music, and ambient sound are some of the main points of this problem since finding similarities and differences between these domains is necessary. Human activity can be classified into four different categories in the video approach, depending on the complexity and the number of body parts involved in the action. Examples of Human activity categories are considered: gestures, actions, interactions and activities. Recognizing human actions in the video becomes a challenge with this varied set of human activities. Furthermore, in the last years, the growth of deep learning techniques applied to this area has been enormous, and the reason is that their results surpass traditional signal processing on a large scale. This article is based on audio and video signals inside a vehicle to detect violence. Furthermore, the architecture used was ResNet model with Mel‐spectrogram methodology for audio signals. The proposed method for video signal representation was RGB, which applied four different models: C2D, I3D, X3D, and Flow‐Gated. Finally, multimodal fusion was applied at the end of the process. [ABSTRACT FROM AUTHOR]

Human action recognition (HAR) in untrimmed videos can make insightful predictions of human behaviour. Previous work on HAR‐included models trained on spatial and temporal annotations and could classify limited actions from trimmed videos. These methods reported limitations such as (1) performance degradation due to the lack of precision temporal regions proposal and (2) poor adaptability of the models in the clinical domain because of unrelated actions of interest. We propose an innovative method that could analyse untrimmed behavioural videos to recommend actions of interest leading to diagnostic and functional assessments for children with Autism Spectrum Disorder (ASD). Our method entails end‐to‐end behaviour action recognition (BAR) pipeline, including child detection, temporal action localization, and actions of interest identification and classification. The model trained on the data of 400 ASD children and 125 with other developmental delays (ODD) accurately identified ASD, ODD, and Neurotypical children with 79.7%, 77.2%, and 80.8% accuracy, respectively. The model's performance on an independent benchmark Self‐Stimulatory Behaviour Dataset (SSBD) reported top‐1 accuracy of 78.57% for combined localization with action recognition, significantly higher than the earlier reported outcomes. [ABSTRACT FROM AUTHOR]

In recent years, the use of mobile sensors for human activity recognition has become an intriguing research area due to the proliferation of wearable and mobile sensors. In Muslim life, the prayer (Salah) is an activity that believers are obligated to perform five times a day. In this study, a new dataset, including Salah, is presented for use in human activity recognition. Named HAR-P (Human Activity Recognition for Praying), the dataset comprises linear acceleration, acceleration, magnetic field, and gyroscope sensor data for eight activities: walking, running, typing, downstairs, upstairs, sitting, standing, and praying. Data were collected from 50 male volunteers aged 15-60 using a smartwatch for the HAR-P dataset. The classification performance of LSTM, ConvLSTM, and CNN-LSTM models was compared for the HAR-P dataset. The highest average classification accuracy of 91% was achieved with the LSTM method using linear acceleration sensor data and the ConvLSTM model using acceleration sensor data, while the lowest average accuracy of 83.6% was attained with the gyroscope sensor data and the ConvLSTM method. [ABSTRACT FROM AUTHOR]

The article introduces a special issue on Realistic Synthetic Data, focusing on the generation, learning, and evaluation of synthetic data for various modalities. It discusses the importance of synthetic data in Machine Learning and Deep Learning, highlighting the challenges of annotating real data and the potential of generative models like GANs. The special issue features articles on topics such as object detection, fake news detection, facial expression generation, and underwater image synthesis, showcasing innovative approaches and solutions in the field of synthetic data generation. [Extracted from the article]

Training machine learning models to identify human behavior is a difficult yet essential task to develop autonomous and adaptive systems such as smart homes. These models require large and diversified amounts of labeled data to be trained effectively. Due to the high variety of home environments and occupant behaviors, collecting datasets that are representative of all possible homes is a major challenge. In addition, privacy and cost are major hurdles to collect real home data. To avoid these difficulties, one solution consists of training these models using purely synthetic data, which can be generated through the simulation of home and their occupants. Two challenges arise from this approach: designing a methodology with a simulation able to generate credible simulated data and evaluating this credibility. In this article, we explain the methodology used to generate diversified synthetic data of daily activities, through the combination of an agent model to simulate an occupant and a simulated 3D house enriched with sensors and effectors to produce such data. We demonstrate the credibility of the generated synthetic data by comparing their efficacy for training human context understanding models against the efficacy generated by real data. To achieve this, we replicate a real dataset collection setting with our smart home simulator. The occupant is replaced by an autonomous agent following the same experimental protocol used for the real dataset collection. This agent is a BDI-based model enhanced with a scheduler designed to offer a balance between control and autonomy. This balance is useful in synthetic data generation since strong constraints can be imposed on the agent to simulate desired situations while allowing autonomous behaviors outside these constraints to generate diversified data. In our case, the constraints are those imposed during the real dataset collection that we want to replicate. The simulated sensors and effectors were configured to react to the agent's behaviors similarly to the real ones. We experimentally show that data generated from this simulation are valuable for two human context understanding tasks: current human activity recognition and future human activity prediction. In particular, we show that models trained solely with simulated data can give reasonable predictions about real situations occurring in the original dataset. We also report experimental results regarding statistical analysis and C2ST to assess the credibility of generated data. We discuss the generality of our approach for evaluating the credibility of simulated data from their use as training data. [ABSTRACT FROM AUTHOR]

Human action recognition (HAR) identifies and classifies human activities by analyzing activity patterns, with feature extraction and classification accuracy being key challenges. In the field of surveillance vision-based, it requires the ability to accurately detect suspicious human activities to provide public safety. In the classifier model for action recognition, the steps start from feature extraction to classification. The classification step uses recurrent neural network architectures such as LSTM to handle sequential data. However, this approach struggles to process spatial information in video data, necessitating the need for a model to learn spatiotemporal patterns from feature data. To address these issues, this study proposes a novel method for classifying activities based on the pattern of 2D skeleton joint swing and angle features for each activity. Additionally, it introduces a novel modified AlexNet architecture with two LSTM layers, called AlexNet-2LSTM, to improve the accuracy of human activity classification. In the performance experiments, the proposed method was evaluated on the KTH and Weizmann datasets, both of which include videos of several people performing different actions. Moreover, to demonstrate the accuracy of the proposed classifier model, it was compared against other state-of-the-art (SOTA) deep learning classifiers, namely Optimized-LSTM, Triple Parallel LSTM, Hybrid CNN-LSTM, LCSWnet, and CC-LSTM-CNN, which the AlexNet-2LSTM achieved precision of 0.95, recall of 0.95, F1-score of 0.94, and accuracy of 0.96on the KTH dataset. Besides that, on the Weizmann dataset 0.95, 0.94, 0.94, and 0.93, respectively. These achievements highlight the proposed model contribution to improving feature extraction and classification accuracy in vision-based HAR systems. [ABSTRACT FROM AUTHOR]

Sensor-based Human Activity Recognition has advanced rapidly, incorporating techniques such as sliding windows, feature extraction, and machine-learning parameter optimization for activity classification. Sliding windows and feature extraction are vital, as they occur early and significantly influence the classification outcomes. The proposed novel approach involves using feature ratios from three sub-windows and a static window combined with feature selection via Analysis of Variance (ANOVA) and machine learning algorithms such as Artificial Neural Networks (ANN) and Extreme Gradient Boosting (XGBoost). The primary objective of the proposed approach is to enhance the efficacy of machine-learning algorithms for recognizing human activities. The effectiveness of the proposed approach is evaluated using three datasets: FORTH-TRACE, SBHARPT, and WISDM. The experimental results indicate that the highest accuracy, precision, recall, and F1 score were achieved on the WISDM dataset, with values of 97.64%, 97.64%, 97.64%, and 97.64%, respectively, using 45 features and an Artificial Neural Network (ANN) classifier. The experiments demonstrated that an overlapping window of 25% enhanced the performance of the machine-learning model. [ABSTRACT FROM AUTHOR]

Federated learning (FL) is an emerging approach to distributed learning from decentralized data, designed with privacy concerns in mind. FL has been successfully applied in several fields, such as the internet of things (IoT), human activity recognition (HAR), and natural language processing (NLP), showing remarkable results. However, the development of FL in real-world applications still faces several challenges. Recent optimizations of FL have been made to address these issues and enhance the FL settings. In this paper, we categorize the optimization of FL into five main challenges: communication efficiency, heterogeneity, privacy and security, scalability, and convergence rate. We provide an overview of various optimization frameworks for FL proposed in previous research, illustrated with concrete examples and applications based on these five optimization goals. Additionally, we propose two optional integrated conceptual frameworks (CFs) for optimizing FL by combining several optimization methods to achieve the best implementation of FL that addresses the five challenges. [ABSTRACT FROM AUTHOR]