Search

Enhancing construction worker performance, safety, and project management through automated activity classification is a promising endeavor. By extracting activity-level information, this technology provides valuable insights for informed decision-making, facilitating project schedule adjustments, efficient resource management, and improved construction site control. Previous studies in this domain focused on basic activities, neglecting optimal sensor placement and no regard for worker comfort. This paper extends beyond existing research, encompassing a broader range of complex construction activities and surpassing current methods. Utilizing unobtrusive wearables like a smartwatch and smartphone, the study determines optimal sensor positions (dominant/nondominant wrist, dominant/nondominant leg pocket). Notably, it introduces a novel deep neural network structure, merging long short-term memory (LSTM) and convolutional layers, offering an innovative solution for automated activity classification tasks in the construction industry. This model extracts activity features automatically reducing the need for manual feature engineering and performs classification with few model parameters indicating efficiency in terms of computational resources and memory requirements making the model more suitable for real-time applications and deployment on resource-constrained devices. By leveraging the strengths of both convolutional layers and LSTM, this approach offers a powerful and efficient solution for activity classification tasks. An experimental study was carried out to recognize four different activities: manual excavation, rebar stirrups, cement plastering, and bar binding. These were performed by four subjects (three males and one female) for 30 s each with different positions of smartwatch and smartphone producing 24,080 data points. Results indicate the optimal positioning of wearables to be smartwatch on dominant hand and smartphone on opposite leg pocket because of a balanced and effective coverage of the relevant movements and contextual information yielding 98.18% accuracy, 98.20% precision, 98.17% recall, and F1 score of 98.17%. Practical Applications: The integration of LSTM and convolutional neural network (CNN) architecture and strategic sensor placement on wearables holds promise for revolutionizing construction project management. By optimizing sensor placement, we can improve the accuracy of activity recognition, allowing for better tracking of worker productivity and process optimization, ultimately leading to increased efficiency and productivity on construction sites. Additionally, strategically placing sensors on wearables enables real-time monitoring of worker movements, facilitating the early identification of safety hazards and ergonomic issues, thereby enhancing worker safety and reducing workplace injuries. Furthermore, integrating wearable technology with smart construction technologies enables real-time data exchange and analysis, promoting collaboration among stakeholders and enhancing progress monitoring and control. Leveraging data from strategically positioned wearables allows for comprehensive visualization of construction processes, aiding in the detection of project bottlenecks and streamlining project management strategies. Lastly, analyzing work allocation data from optimally positioned wearables improves project bidding accuracy, minimizing cost estimation errors and increasing profit margins. These findings highlight the transformative potential of LSTM-CNN architecture and optimal wearable placement in construction project management. [ABSTRACT FROM AUTHOR]

Abstract Background Accurately measuring energy expenditure during physical activity outside of the laboratory is challenging, especially on a large scale. Thigh-worn accelerometers have gained popularity due to the possibility to accurately detect physical activity types. The use of machine learning techniques for activity classification and energy expenditure prediction may improve accuracy over current methods. Here, we developed a novel composite energy expenditure estimation model by combining an activity classification model with a stride specific energy expenditure model for walking, running, and cycling. Methods We first trained a supervised deep learning activity classification model using pooled data from available adult accelerometer datasets. The composite energy expenditure model was then developed and validated using additional data based on a sample of 69 healthy adult participants (49% female; age = 25.2 ± 5.8 years) who completed a standardised activity protocol with indirect calorimetry as the reference measure. Results The activity classification model showed an overall accuracy of 99.7% across all five activity types during validation. The composite model for estimating energy expenditure achieved a mean absolute percentage error of 10.9%. For running, walking, and cycling, the composite model achieved a mean absolute percentage error of 6.6%, 7.9% and 16.1%, respectively. Conclusions The integration of thigh-worn accelerometers with machine learning models provides a highly accurate method for classifying physical activity types and estimating energy expenditure. Our novel composite model approach improves the accuracy of energy expenditure measurements and supports better monitoring and assessment methods in non-laboratory settings.

Abstract Background The more accurate we can assess human physical behaviour in free-living conditions the better we can understand its relationship with health and wellbeing. Thigh-worn accelerometry can be used to identify basic activity types as well as different postures with high accuracy. User-friendly software without the need for specialized programming may support the adoption of this method. This study aims to evaluate the classification accuracy of two novel no-code classification methods, namely SENS motion and ActiPASS. Methods A sample of 38 healthy adults (30.8 ± 9.6 years; 53% female) wore the SENS motion accelerometer (12.5 Hz; ±4 g) on their thigh during various physical activities. Participants completed standardized activities with varying intensities in the laboratory. Activities included walking, running, cycling, sitting, standing, and lying down. Subsequently, participants performed unrestricted free-living activities outside of the laboratory while being video-recorded with a chest-mounted camera. Videos were annotated using a predefined labelling scheme and annotations served as a reference for the free-living condition. Classification output from the SENS motion software and ActiPASS software was compared to reference labels. Results A total of 63.6 h of activity data were analysed. We observed a high level of agreement between the two classification algorithms and their respective references in both conditions. In the free-living condition, Cohen’s kappa coefficients were 0.86 for SENS and 0.92 for ActiPASS. The mean balanced accuracy ranged from 0.81 (cycling) to 0.99 (running) for SENS and from 0.92 (walking) to 0.99 (sedentary) for ActiPASS across all activity types. Conclusions The study shows that two available no-code classification methods can be used to accurately identify basic physical activity types and postures. Our results highlight the accuracy of both methods based on relatively low sampling frequency data. The classification methods showed differences in performance, with lower sensitivity observed in free-living cycling (SENS) and slow treadmill walking (ActiPASS). Both methods use different sets of activity classes with varying definitions, which may explain the observed differences. Our results support the use of the SENS motion system and both no-code classification methods.

Background: Accurately measuring energy expenditure during physical activity outside of the laboratory is challenging, especially on a large scale. Thigh-worn accelerometers have gained popularity due to the possibility to accurately detect physical activity types. The use of machine learning techniques for activity classification and energy expenditure prediction may improve accuracy over current methods. Here, we developed a novel composite energy expenditure estimation model by combining an activity classification model with a stride specific energy expenditure model for walking, running, and cycling. Methods: We first trained a supervised deep learning activity classification model using pooled data from available adult accelerometer datasets. The composite energy expenditure model was then developed and validated using additional data based on a sample of 69 healthy adult participants (49% female; age = 25.2 ± 5.8 years) who completed a standardised activity protocol with indirect calorimetry as the reference measure. Results: The activity classification model showed an overall accuracy of 99.7% across all five activity types during validation. The composite model for estimating energy expenditure achieved a mean absolute percentage error of 10.9%. For running, walking, and cycling, the composite model achieved a mean absolute percentage error of 6.6%, 7.9% and 16.1%, respectively. Conclusions: The integration of thigh-worn accelerometers with machine learning models provides a highly accurate method for classifying physical activity types and estimating energy expenditure. Our novel composite model approach improves the accuracy of energy expenditure measurements and supports better monitoring and assessment methods in non-laboratory settings. [ABSTRACT FROM AUTHOR]

Background: The more accurate we can assess human physical behaviour in free-living conditions the better we can understand its relationship with health and wellbeing. Thigh-worn accelerometry can be used to identify basic activity types as well as different postures with high accuracy. User-friendly software without the need for specialized programming may support the adoption of this method. This study aims to evaluate the classification accuracy of two novel no-code classification methods, namely SENS motion and ActiPASS. Methods: A sample of 38 healthy adults (30.8 ± 9.6 years; 53% female) wore the SENS motion accelerometer (12.5 Hz; ±4 g) on their thigh during various physical activities. Participants completed standardized activities with varying intensities in the laboratory. Activities included walking, running, cycling, sitting, standing, and lying down. Subsequently, participants performed unrestricted free-living activities outside of the laboratory while being video-recorded with a chest-mounted camera. Videos were annotated using a predefined labelling scheme and annotations served as a reference for the free-living condition. Classification output from the SENS motion software and ActiPASS software was compared to reference labels. Results: A total of 63.6 h of activity data were analysed. We observed a high level of agreement between the two classification algorithms and their respective references in both conditions. In the free-living condition, Cohen's kappa coefficients were 0.86 for SENS and 0.92 for ActiPASS. The mean balanced accuracy ranged from 0.81 (cycling) to 0.99 (running) for SENS and from 0.92 (walking) to 0.99 (sedentary) for ActiPASS across all activity types. Conclusions: The study shows that two available no-code classification methods can be used to accurately identify basic physical activity types and postures. Our results highlight the accuracy of both methods based on relatively low sampling frequency data. The classification methods showed differences in performance, with lower sensitivity observed in free-living cycling (SENS) and slow treadmill walking (ActiPASS). Both methods use different sets of activity classes with varying definitions, which may explain the observed differences. Our results support the use of the SENS motion system and both no-code classification methods. [ABSTRACT FROM AUTHOR]

Assembly activity recognition and prediction help to improve productivity, quality control, and safety measures in smart factories. This study aims to sense, recognize, and predict a worker's continuous fine-grained assembly activities in a manufacturing platform. We propose a two-stage network for workers' fine-grained activity classification by leveraging scene-level and temporal-level activity features. The first stage is a feature awareness block that extracts scene-level features from multi-visual modalities, including red–green–blue (RGB) and hand skeleton frames. We use the transfer learning method in the first stage and compare three different pre-trained feature extraction models. Then, we transmit the feature information from the first stage to the second stage to learn the temporal-level features of activities. The second stage consists of the Recurrent Neural Network (RNN) layers and a final classifier. We compare the performance of two different RNNs in the second stage, including the Long Short-Term Memory (LSTM) and the Gated Recurrent Unit (GRU). The partial video observation method is used in the prediction of fine-grained activities. In the experiments using the trimmed activity videos, our model achieves an accuracy of > 99% on our dataset and > 98% on the public dataset UCF 101, outperforming the state-of-the-art models. The prediction model achieves an accuracy of > 97% in predicting activity labels using 50% of the onset activity video information. In the experiments using an untrimmed video with continuous assembly activities, we combine our recognition and prediction models and achieve an accuracy of > 91% in real time, surpassing the state-of-the-art models for the recognition of continuous assembly activities. [ABSTRACT FROM AUTHOR]

Introduction: Water polo upper limb external load monitoring cannot be currently measured accurately because of technological and methodological challenges. This is problematic as large fluctuations in overhead movement volume and intensity may affect performance and alter injury risk. Inertial measurement units (IMU) and machine learning techniques have been shown to accurately classify overhead movements in other sports. We investigated the model accuracy and class precision, sensitivity, and specificity of IMU and machine learning techniques to classify standard overhead drill movements in elite women's water polo. Methods: Ten women's water polo players performed standard drills of swimming, blocking, low-intensity throwing and high-intensity throwing under training conditions. Athletes wore two IMU: one on the upper back and the other on the distal forearm. Each movement was videoed and coded to a standard overhead drill movement. IMU and coded video data were merged to verify the IMU-detected activity classification of each movement to that of the video. Data were partitioned into a training and a test set and used to form a decision tree algorithm. Model accuracy and class precision, sensitivity, and specificity were assessed. Results: IMU resultant acceleration and angular velocity values displayed drill-specific values. A total of 194 activities were identified by the model in the test set, with 8 activities being incorrectly classified. Model accuracy was 95.88%. Percentage class precision, sensitivity, and specificity were as follows: blocking (96.15, 86.21, 99.39), high-intensity throwing (100, 100, 100), low-intensity throwing (93.48,93.48,97.97), and swimming (94.81,98.65,96.67). Conclusions: IMU and machine learning techniques can accurately classify standard overhead drill movements in elite women's water polo. [ABSTRACT FROM AUTHOR]

Resultant hip joint forces can currently only be recorded in situ in a laboratory setting using instrumented total hip replacements (THRs) equipped with strain gauges. However, permanent recording is important for monitoring the structural condition of the implant, for therapeutic purposes, for self-reflection, and for research into managing the predicted increasing number of THRs worldwide. Therefore, this study aims to investigate whether a recently proposed THR with an integrated piezoelectric element represents a new possibility for the permanent recording of hip joint forces and the physical activities of the patient. Hip joint forces from nine different daily activities were obtained from the OrthoLoad database and applied to a total hip stem equipped with a piezoelectric element using a uniaxial testing machine. The forces acting on the piezoelectric element were calculated from the generated voltages. The correlation between the calculated forces on the piezoelectric element and the applied forces was investigated, and the regression equations were determined. In addition, the voltage outputs were used to predict the activity with a random forest classifier. The coefficient of determination between the applied maximum forces on the implant and the calculated maximum forces on the piezoelectric element was R2 = 0.97 (p < 0.01). The maximum forces on the THR could be determined via activity-independent determinations with a deviation of 2.49 ± 13.16% and activity-dependent calculation with 0.87 ± 7.28% deviation. The activities could be correctly predicted using the classification model with 95% accuracy. Hence, piezoelectric elements integrated into a total hip stem represent a promising sensor option for the energy-autonomous detection of joint forces and physical activities. [ABSTRACT FROM AUTHOR]

BACKGROUND: Hip-worn accelerometers are commonly used, but data processed using the 100 counts per minute cut point do not accurately measure sitting patterns. We developed and validated a model to accurately classify sitting and sitting patterns using hip-worn accelerometer data from a wide age range of older adults. METHODS: Deep learning models were trained with 30-Hz triaxial hip-worn accelerometer data as inputs and activPAL sitting/nonsitting events as ground truth. Data from 981 adults aged 35-99 years from cohorts in two continents were used to train the model, which we call CHAP-Adult (Convolutional Neural Network Hip Accelerometer Posture-Adult). Validation was conducted among 419 randomly selected adults not included in model training. RESULTS: Mean errors (activPAL - CHAP-Adult) and 95% limits of agreement were: sedentary time -10.5 (-63.0, 42.0) min/day, breaks in sedentary time 1.9 (-9.2, 12.9) breaks/day, mean bout duration -0.6 (-4.0, 2.7) min, usual bout duration -1.4 (-8.3, 5.4) min, alpha .00 (-.04, .04), and time in ≥30-min bouts -15.1 (-84.3, 54.1) min/day. Respective mean (and absolute) percent errors were: -2.0% (4.0%), -4.7% (12.2%), 4.1% (11.6%), -4.4% (9.6%), 0.0% (1.4%), and 5.4% (9.6%). Pearsons correlations were: .96, .92, .86, .92, .78, and .96. Error was generally consistent across age, gender, and body mass index groups with the largest deviations observed for those with body mass index ≥30 kg/m2. CONCLUSIONS: Overall, these strong validation results indicate CHAP-Adult represents a significant advancement in the ambulatory measurement of sitting and sitting patterns using hip-worn accelerometers. Pending external validation, it could be widely applied to data from around the world to extend understanding of the epidemiology and health consequences of sitting.

Populations ranging from elite athletes to the elderly can benefit from unobtrusive activity monitoring. Inertial motion capture, which utilizes inertial measurement units (IMUs), is a relatively inexpensive and common technology that is uniquely advantageous when employed to achieve this goal. This study considers how varying IMU configurations in human activity recognition influences performance metrics. Ensemble learning was used for classification with support vector machine and k-nearest neighbor machine learning models utilized as inputs for a shallow neural network. The accuracy and F1 score for the neural network were compared between the four sensor configurations (full set, necessary only, person only, and tool only). The full set and necessary only performed the best with accuracy and F1 score above 97% and 0.95 respectively. The improved performance of these sets over the person only and tool only showed the benefits of a mixed composition of sensors.

An active learning-type class is a class in which students take the initiative. In order to improve active learning-type classes, attempts have been made to review the content after the class using video taken by the lecturer, but this is burdensome because it takes a long time to review the video. Although methods have been proposed for estimating the classroom process at each time, there is still room for improvement. In this paper, we propose a method for estimating the classroom process at each time with higher accuracy than conventional methods. The proposed method uses deep learning to improve the accuracy of the conventional method, which only uses classical SVM. We also used an ablation study to find an appropriate combination of input modalities. Furthermore, we introduced ensemble LSTM to handle data with different modalities. The proposed method achieved the highest accuracy of 98.8% and the lowest accuracy of 64.4% in estimating classroom activities, with an average classification accuracy of 80.1%.

Summary: Recently, researchers have placed a great deal of emphasis on modeling activity patterns to better understand human behavior. Several approaches have been researched so far to develop automatic human activity recognition systems that infer detailed semantics from visual images, aiming to understand real human behavior patterns. However, there is still a need for a cost effective solution to distinguish human actions in the real‐world environment. With this encouragement, a novel approach is proposed to recognize shoplifting acts by examining the posture evidence of the human being. This approach begins by obtaining the two‐dimensional pose reflecting human's body joints as a skeleton from the recorded frames. Subsequently, a preprocessing step is used to preprocess skeleton data, which can handle the occlusion too. Postural feature generation is then used to extract pertinent features from such preprocessed skeletons. Finally, feature deduction is performed to downsize the derived features to a smaller dimension, and activity classification is performed on such reduced features to identify shoplifting behaviors in real time. A synthetic shoplifting dataset and real store recorded videos are used to conduct the experiments, the findings of which appear more promising than those obtained using other cutting‐edge methods, with an accuracy of 97.36% and 91.66% for synthesized and real store recorded inputs. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Textiles & Engineers / Tekstil ve Mühendis is the property of Union of Chambers of Turkish Engineers & Architects, Chamber of Textile Engineers and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

According to (he Intergovernmental Panel on Climate Change (IPCC) 2023 Report, Climate change is assumed to be currently an undisputed phenomenon of anthropogenic nature. Globally, the need to undertake and carry out a systematic policy to limit their negative consequences prevails during the last decades, and the decisions contained in the Paris Agreement, as well as in the policies of the European Union and, in recent months, of the United States are an illustration. Global uncertainty has continued growing enormously because of the COVID-19 pandemic and wars in Ukraine and in the Middle East. Before these military events, a sluggish growth has characterized different geographical areas, such as USA and Europe. Consequently, these armed conflicts and sanitary crisis have intensified the global recession and raising the question of the need to modify the agenda on which to implement climate policies. These issues are global and the banking system in Bulgaria is one of the main actors in Bulgaria to give an impulsion for a sustainable growth. This paper aims at providing an inventory of the banks efforts to green their assets and their climate change exposure within the country. It also outlines the main obstacles and future recommendations. [ABSTRACT FROM AUTHOR]

Effectively monitoring and analyzing on-site module installation for modular integrated construction (MiC) is essential to properly coordinating the MiC process. In this study, the authors propose an automated productivity monitoring framework for on-site module installation operations consisting of three modules: object detection, activity classification, and productivity analysis. The object detection module detects mobile cranes and modules interacting with mobile cranes, and the activity classification module classifies module installation activities into five different activities by considering the spatiotemporal relationship between the detected objects. Finally, the productivity analysis module analyzes the productivity of the module installation process by utilizing the accumulated activity classification results over image frames. The proposed model achieves an average accuracy of 89% (hooking: 85.71%, lifting: 84.44%, positioning: 94.90%, returning: 83.09%, and idling: 96.87%) in classifying the five activities. The developed framework enables practitioners to measure the productivity of the on-site module installation process automatically. In addition, productivity data stored from diverse construction sites contribute to identifying progress-impeding factors and improving the productivity of the entire MiC process.

Football (soccer) stands as the world's most popular sport, enthralling millions of fans globally. Within the dynamic sphere of modern football, gaining comprehensive insights into the nuanced technical and physical demands placed on players is pivotal for performance optimization. This paper introduces an innovative deep learning-powered image classification algorithm for recognizing football player activities directly from videos and images. Our pioneering approach harnesses the synergistic capabilities of convolutional neural networks (CNNs) and graph convolutional networks (GCNs) to decipher intricate spatial–temporal patterns in player poses and motions. The methodology employs a hybrid CNN and GCN architecture. The CNN leverages consecutive convolutional and pooling layers to automatically extract discriminative visual features from input frames capturing player poses. The GCN models' skeletal joints as graph nodes with bones as edges, performing graph convolutions to aggregate spatial and temporal information from neighboring body parts. This enables capturing localized pose dynamics. The complementary CNN and GCN outputs are fused through fully connected layers to classify player activities based on both visual appearances and pose configurations. The model is trained end to end on richly annotated football video data using a multi-class cross-entropy loss. Data augmentation and regularization techniques enhance robustness. Extensive experiments validate the proposed architecture's effectiveness, achieving 97.4% accuracy, 96% precision, 95.5% recall, 95.4% F1-score, 0.90 Matthews correlation, and 93% specificity in classifying 17 complex football activities significantly higher than previous benchmarks. Detailed ablation studies confirm the contributions of the CNN, GCN, and fused model components. The work represents a major advance in leveraging deep neural networks for accurate and granular analysis of sports performances. [ABSTRACT FROM AUTHOR]

BackgroundMachine learning has been used for classification of physical behavior bouts from hip-worn accelerometers; however, this research has been limited due to the challenges of directly observing and coding human behavior "in the wild." Deep learning algorithms, such as convolutional neural networks (CNNs), may offer better representation of data than other machine learning algorithms without the need for engineered features and may be better suited to dealing with free-living data. The purpose of this study was to develop a modeling pipeline for evaluation of a CNN model on a free-living data set and compare CNN inputs and results with the commonly used machine learning random forest and logistic regression algorithms.MethodTwenty-eight free-living women wore an ActiGraph GT3X+accelerometer on their right hip for 7 days. A concurrently worn thigh-mounted activPAL device captured ground truth activity labels. The authors evaluated logistic regression, random forest, and CNN models for classifying sitting, standing, and stepping bouts. The authors also assessed the benefit of performing feature engineering for this task.ResultsThe CNN classifier performed best (average balanced accuracy for bout classification of sitting, standing, and stepping was 84%) compared with the other methods (56% for logistic regression and 76% for random forest), even without performing any feature engineering.ConclusionUsing the recent advancements in deep neural networks, the authors showed that a CNN model can outperform other methods even without feature engineering. This has important implications for both the model's ability to deal with the complexity of free-living data and its potential transferability to new populations.

Resultant hip joint forces can currently only be recorded in situ in a laboratory setting using instrumented total hip replacements (THRs) equipped with strain gauges. However, permanent recording is important for monitoring the structural condition of the implant, for therapeutic purposes, for self-reflection, and for research into managing the predicted increasing number of THRs worldwide. Therefore, this study aims to investigate whether a recently proposed THR with an integrated piezoelectric element represents a new possibility for the permanent recording of hip joint forces and the physical activities of the patient. Hip joint forces from nine different daily activities were obtained from the OrthoLoad database and applied to a total hip stem equipped with a piezoelectric element using a uniaxial testing machine. The forces acting on the piezoelectric element were calculated from the generated voltages. The correlation between the calculated forces on the piezoelectric element and the applied forces was investigated, and the regression equations were determined. In addition, the voltage outputs were used to predict the activity with a random forest classifier. The coefficient of determination between the applied maximum forces on the implant and the calculated maximum forces on the piezoelectric element was R2 = 0.97 (p < 0.01). The maximum forces on the THR could be determined via activity-independent determinations with a deviation of 2.49 ± 13.16% and activity-dependent calculation with 0.87 ± 7.28% deviation. The activities could be correctly predicted using the classification model with 95% accuracy. Hence, piezoelectric elements integrated into a total hip stem represent a promising sensor option for the energy-autonomous detection of joint forces and physical activities.

Surface electromyogram (SEMG) shows the vital information of human motor activity. This information can identify the motor activity and controlling assistive and rehabilitative devices. However, its efficient extraction and interpretation plays a very critical role. The present study is focused on lower limb activity classification by selecting the efficient feature vector of the lower limb SEMG signal. Here, the SEMG signal-dependent continuous locomotion mode classification method has been proposed along with a dual-stage Feature Selection Algorithm (FSA). To evaluate the proposed method, SEMG signals of fifteen subjects were recorded from two lower limb muscles (Fibularis longus and Biceps Femoris) for five lower limb activities. The performance of six different classifiers was compared for intuitive feature vectors and FSA. The study analysed the performance of the proposed model for a single muscle approach and a dual muscle approach of activity classification. The time domain features were found more effective over other features, whereas Biceps Femoris muscle came out as a dominant muscle. The FSA enhanced the performance of classification model with fewer features as compared to intuitive feature subsets. Moreover, the dual muscle approach outperformed (p-value<0.05) over the single muscle approach. The best performance, for the dual muscle approach, was achieved 97.73 ± 2.47%, 98.99 ± 1.26% and 96.33 ± 2.70% for Neural Network, Linear Discriminant Analysis and Support Vector Machine classifiers, respectively (p-value > 0.05). [ABSTRACT FROM AUTHOR]

Human Activity Recognition (HAR) from video data collections is the core application in vision tasks and has a variety of utilizations including object detection applications, video-based behavior monitoring, video classification, and indexing, patient monitoring, robotics, and behavior analysis. Although many techniques are available for HAR in video analysis tasks, most of them are not focusing on behavioral analysis. Hence, a new HAR system analysis the behavioral activity of a person based on the deep learning approach proposed in this work. The most essential aim of this work is to recognize the complex activities that are useful in many tasks that are based on object detection, modelling of individual frame characteristics, and communication among them. Moreover, this work focuses on finding out the human actions from various video resolutions, invariant human poses, and nearness of multi objects. First, we identify the key and essential frames of each activity using histogram differences. Secondly, Discrete Wavelet Transform (DWT) is used in this system to extract coefficients from the sequence of key-frames where the activity is localized in space. Finally, an Adaptive Weighted Flow Net (AWFN) algorithm is proposed in this work for effective video activity recognition. Moreover, the proposed algorithm has been evaluated by comparing it with the existing Visual Geometry Group (VGG-16) convolution neural networks for making performance comparisons. This work focuses on competent deep learning-based feature extraction to discriminate the activities for performing the classification accuracy. The proposed model has been evaluated with VGG-16 using a combination of regular UCF-101 activity datasets and also in very challenging Low-quality videos such as HMDB51. From these investigations, it is proved that the proposed AWFN approach gives higher detection accuracy of 96%. It is approximately 0.3% to 7.88% of higher accuracy than state-of-art methods. [ABSTRACT FROM AUTHOR]

From recent past, shoplifting has become a serious concern for business in both small/big shops and stores. It customarily involves the buyer concealing store items inside clothes/bags and then leaving the store without payment. Unfortunately, no cost-effective solution is available to overcome this problem. We, therefore intend to build an expert monitoring system to automatically recognize shoplifting events in megastores/shops by recognizing object-stealing actions of humans. The method proposed utilizes a deep convolutional-based InceptionV3 architecture to mine the prominent features from video clips. These features are used to custom Long Short Term Memory (LSTM) network to discriminate human stealing actions in video sequences. Optimizing recurrent learning classifier using different modeling parameters such as sequence length and batch size is a genuine contribution of this work. The experiments demonstrate that the system proposed has achieved an accuracy of 89.36% on the synthesized dataset, which comparatively outperforms other existing methods.

Abstract Activity classification and biometric authentication have become synonymous with wearable technologies such as smartwatches and trackers. Although great efforts have been made to develop electrocardiogram (ECG)‐based biometric verification and identification modalities using data from these devices, in this paper, we explore the use of adaptive techniques based on prior activity classification in an attempt to enhance biometric performance. In doing so, we also compare two waveform similarity distances to provide features for classification. Two public datasets which were collected from medical and wearable devices provide a cross‐device comparison. Our results show that our method is able to be used for both wearable and medical devices in activity classification and biometric verification cases. This study is the first study which uses only ECG signals for both activity classification and biometric verification purposes.

Whether the shield tunnel can be used in the ground fissure site is an issue in the metro construction and the structural adaptability and countermeasures are key points in Xi'an, China. Taking Xi'an Metro ring line(Metro Line 8)as the engineering background, based on field investigation, data collection, InSAR monitoring and finite element numerical simulation, the active trend of ground fissures in Xi'an and the classification of each ground fissure along Metro Line 8 were discussed, as well as the adaptability of shield tunnels with different assembly modes and section sizes in the ground fissure site was analyzed; finally, the countermeasures such as secondary lining and concealed beam, steel segment, grouting reinforcement, and local reinforcement of handholes were proposed, and the effectiveness was also validated. The results show that the shield tunnels can be used in Class Ⅳ sites with weak activity of ground fissure, and the staggered-joint assembly mode is better than the straight joint assembly mode for shield tunnel crossing the ground fissure zones, and the optimal diameter(D)of shield tunnel is 6.2 m; the longitudinal fortification length of sheild tunnel is 0.75D to 1.50D distance from ground fissure, which is related to the intersection angle between the tunnel and the ground fissure zones. The research can provide a scientific guidance for the structural design and disaster control of metro shield tunnel crossing ground fissure zones. [ABSTRACT FROM AUTHOR]

Activity classification and biometric authentication have become synonymous with wearable technologies such as smartwatches and trackers. Although great efforts have been made to develop electrocardiogram (ECG)‐based biometric verification and identification modalities using data from these devices, in this paper, we explore the use of adaptive techniques based on prior activity classification in an attempt to enhance biometric performance. In doing so, we also compare two waveform similarity distances to provide features for classification. Two public datasets which were collected from medical and wearable devices provide a cross‐device comparison. Our results show that our method is able to be used for both wearable and medical devices in activity classification and biometric verification cases. This study is the first study which uses only ECG signals for both activity classification and biometric verification purposes. [ABSTRACT FROM AUTHOR]

The employment of machine learning algorithms to the data provided by wearable movement sensors is one of the most common methods to detect pets' behaviors and monitor their well-being. However, defining features that lead to highly accurate behavior classification is quite challenging. To address this problem, in this study we aim to classify six main dog activities (standing, walking, running, sitting, lying down, and resting) using high-dimensional sensor raw data. Data were received from the accelerometer and gyroscope sensors that are designed to be attached to the dog's smart costume. Once data are received, the module computes a quaternion value for each data point that provides handful features for classification. Next, to perform the classification, we used several supervised machine learning algorithms, such as the Gaussian naïve Bayes (GNB), Decision Tree (DT), K-nearest neighbor (KNN), and support vector machine (SVM). In order to evaluate the performance, we finally compared the proposed approach's F-score accuracies with the accuracy of classic approach performance, where sensors' data are collected without computing the quaternion value and directly utilized by the model. Overall, 18 dogs equipped with harnesses participated in the experiment. The results of the experiment show a significantly enhanced classification with the proposed approach. Among all the classifiers, the GNB classification model achieved the highest accuracy for dog behavior. The behaviors are classified with F-score accuracies of 0.94, 0.86, 0.94, 0.89, 0.95, and 1, respectively. Moreover, it has been observed that the GNB classifier achieved 93% accuracy on average with the dataset consisting of quaternion values. In contrast, it was only 88% when the model used the dataset from sensors' data. [ABSTRACT FROM AUTHOR]

Purpose: Semantic segmentation and activity classification are key components to create intelligent surgical systems able to understand and assist clinical workflow. In the operating room, semantic segmentation is at the core of creating robots aware of clinical surroundings, whereas activity classification aims at understanding OR workflow at a higher level. State-of-the-art semantic segmentation and activity recognition approaches are fully supervised, which is not scalable. Self-supervision can decrease the amount of annotated data needed. Methods: We propose a new 3D self-supervised task for OR scene understanding utilizing OR scene images captured with ToF cameras. Contrary to other self-supervised approaches, where handcrafted pretext tasks are focused on 2D image features, our proposed task consists of predicting relative 3D distance of image patches by exploiting the depth maps. By learning 3D spatial context, it generates discriminative features for our downstream tasks. Results: Our approach is evaluated on two tasks and datasets containing multiview data captured from clinical scenarios. We demonstrate a noteworthy improvement in performance on both tasks, specifically on low-regime data where utility of self-supervised learning is the highest. Conclusion: We propose a novel privacy-preserving self-supervised approach utilizing depth maps. Our proposed method shows performance on par with other self-supervised approaches and could be an interesting way to alleviate the burden of full supervision. [ABSTRACT FROM AUTHOR]

In this work, a novel method is presented for non-contact non-invasive physical activity monitoring, which utilizes a multi-axial inertial measurement unit (IMU) to measure activity-induced structural vibrations in multiple axes. The method is demonstrated in monitoring the activity of a mouse in a husbandry cage, where activity is classified as resting, stationary activity and locomotion. In this setup, the IMU is mounted in the center of the underside of the cage floor where vibrations are measured as accelerations and angular rates in the X-, Y- and Z-axis. The ground truth of activity is provided by a camera mounted in the cage lid. This setup is used to record 27.67 h of IMU data and ground truth activity labels. A classification model is trained with 16.17 h of data which amounts to 3880 data points. Each data point contains eleven features, calculated from the X-, Y- and Z-axis accelerometer data. The method achieves over 90% accuracy in classifying activity versus non-activity. Activity is monitored continuously over more than a day and clearly depicts the nocturnal behavior of the inhabitant. The impact of this work is a powerful method to assess activity which enables automatic health evaluation and optimization of workflows for improved animal wellbeing. [ABSTRACT FROM AUTHOR]

Fall is a challenging task that poses a great danger to the elderly person’s health as they carry out their daily routines and activities and could lead to serious injuries, long hospitalization, or even death. One of the key solutions to this important problem is the prompt and automatic detection of their fall motion, among other activities, so that immediate help can be rendered to avoid further complications. Cameras and other wearable sensors are the conventional means employed to monitor the daily activities performed by elderly people. However, recently radar sensing technology has been favored largely due to its short-range and velocity estimation. And most importantly, their ability to remotely capture the human activity motion electromagnetic wave signal without infringing on the user’s privacy. To this end, we present an elderly people fall detection system that integrates the radar signal micro-Doppler features extracted with the convolutional neural networks (CNNs). The features were extracted using Alex-Net, VGG-16-Net, and VGG-19-Net pre-trained models. We adapted the canonical correlation analysis (CCA) algorithm by proposing a channel attention network to fuse the extracted features effectively. The channel attention module employed a series of convolutional filters to determine the most discriminative features to focus on and discard the redundant ones. The fused features are classified using an SVM classifier. Our proposed method achieved the best performance against the state-of-the-art approaches. Specifically, our approach attained 99.77% test accuracy, which is about a 2% to 4 % increase compared to the recent state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

Activity Recognition and Classification (ARC) have enabled growth of many automated applications like recommender services, old age assistance, health monitoring, security and surveillance etc. This becomes possible due to advancement of technology in wearable devices and smartphones. The small size, easy availability of various relevant sensors, ever decreasing cost, ability to monitor continuously and handy to use features have made them prominent devices to use in ARC. In this work, we provide a comprehensive survey on ARC using smartphones and wearable sensors. The work begins with the understanding of the ARC process followed by description of inertial sensors present in the smartphones and wearables. It covers the various feature extraction methods and the models used in traditional methods and the trending deep learning based methods. It is observed that, performance of any ARC method largely depends on number of sensors, classification technique, kind of device and placement and orientation of the device among many other parameters considered in the work. In our study, we present a detailed comparison of work done in this area considering ten such important parameters, which, to the best of our knowledge, is the first of its kind of surveys. Finally, we present ten challenges in this area and provide prospective dimensions that can be explored in future research. [ABSTRACT FROM AUTHOR]

Nowadays, the world is witnessing a significant rise in the cases of both reported and unnoticed violations. As an answer to this rising menace, video surveillance can fill the gap of covering untapped actions which lead to violence, while also ensuring a secure life. In our everyday life, surveillance can be accomplished efficiently by activity classification from drone videos. The prominent fields that have employed this technology are police work, video categorization, biometrics, and human–computer interaction. So far, no public dataset is available for violent activity classification using drone surveillance. Hence, this work aims to look into the domain of machine-driven recognition and classification of human actions from drone videos. In this study, the dataset is created using drones from different heights for an unconstrained environment. The study begins by performing key-point extraction and generate 2D skeletons for the persons in the frame. These extracted key points are given as features in the classification module to recognize the actions. The classification models used in the proposed method are SVM (support vector machine) and Random Forest. Experimental results show that the SVM model with RBF (radial basis function) kernel for activity classification is more efficient when compared to the prior proposed approaches and other experimented models. The research work has also analyzed the run time performance of the proposed system and achieve its real-time performance. [ABSTRACT FROM AUTHOR]

The employment of machine learning algorithms to the data provided by wearable movement sensors is one of the most common methods to detect pets’ behaviors and monitor their well-being. However, defining features that lead to highly accurate behavior classification is quite challenging. To address this problem, in this study we aim to classify six main dog activities (standing, walking, running, sitting, lying down, and resting) using high-dimensional sensor raw data. Data were received from the accelerometer and gyroscope sensors that are designed to be attached to the dog’s smart costume. Once data are received, the module computes a quaternion value for each data point that provides handful features for classification. Next, to perform the classification, we used several supervised machine learning algorithms, such as the Gaussian naïve Bayes (GNB), Decision Tree (DT), K-nearest neighbor (KNN), and support vector machine (SVM). In order to evaluate the performance, we finally compared the proposed approach’s F-score accuracies with the accuracy of classic approach performance, where sensors’ data are collected without computing the quaternion value and directly utilized by the model. Overall, 18 dogs equipped with harnesses participated in the experiment. The results of the experiment show a significantly enhanced classification with the proposed approach. Among all the classifiers, the GNB classification model achieved the highest accuracy for dog behavior. The behaviors are classified with F-score accuracies of 0.94, 0.86, 0.94, 0.89, 0.95, and 1, respectively. Moreover, it has been observed that the GNB classifier achieved 93% accuracy on average with the dataset consisting of quaternion values. In contrast, it was only 88% when the model used the dataset from sensors’ data.

The population of elderly is growing and is projected to outnumber the youth in the future. Many researches on elderly assisted living technology were carried out. One of the focus areas is activity monitoring of the elderly. AReM dataset is a time series activity recognition dataset for seven different types of activities, which are bending 1, bending 2, cycling, lying, sitting, standing and walking. In the original paper, the author used a many-to-many recurrent neural network for activity recognition. Here, we introduced a time series classification method where Gated Recurrent Units with many-to-one architecture were used for activity classification. The experimental results obtained showed an excellent accuracy of 97.14%. [ABSTRACT FROM AUTHOR]

Motor activity data allows for analysis of complex behavioral patterns, including the diagnosis of mental disorders, such as depression or schizophrenia. However, the classification of actigraphy signals remains a challenge. The main reasons are small datasets and the need for sophisticated feature engineering. The recent development of AutoML approaches allows for automating feature extraction and selection. In this work, we compare automatic and manual feature engineering for applications in mental health. We also analyze classifier evaluation methods for small datasets. The automated approach results in better classification, as measured with several metrics, and in a shorter, cleaner code, providing software engineering advantages. [ABSTRACT FROM AUTHOR]

Even though cats inhabit a large number of our homes, their activity patterns, biological rhythm and behavior are still under-described and, therefore, poorly understood. This lack of knowledge is partly explained by the limited number of tools available for observing cats with limited effort and minimal disturbances to their routines. To this day, we still lack a faithful description of what constitutes a "normal behavior" of a domestic cat. Bio-loggers have been proven to help researchers decipher the individual and group behaviors of wild animals, and are starting to emerge as a tool for monitoring pets' health as well. These tools have been used not only to analyze cats' activity patterns but also examine their detailed activities, such as eating and walking. Bio-logging coupled with data analysis and Artificial Intelligence tools give hopeful perspectives on more complex activities such as grooming, jumping, social interaction, etc. In this review, our aim is to offer an overview of the contribution of bio-logging devices to the current knowledge and understanding regarding the cat's behavior, and their detailed activities. We specifically focus on the knowledge brought to the understanding of cats' biological rhythm and behavior by accelerometry. This review will provide insight into the major advances facilitated by this tool, its limitations, as well as the possible future development in this field of study. • Accelerometry is a valuable tool to objectively describe cat's biological rhythm. • Cats can adapt to seasonal changes, temperature, human, prey and predator presence. • Advances in Machine Learning could provide a detailed report of cats' activities. • Questions about the cat's daily life evolution could be answered with precision. [ABSTRACT FROM AUTHOR]

Rollators are widely used by people with mobility problems, but previous studies have been limited to self-report approaches when evaluating their real-world effectiveness. To support studies based on more robust datasets, a method to estimate mobility parameters, such as gait speed and distance traveled, in the real world is needed. Body-worn sensors offer one approach to the problem, but rollator-mounted sensors have some practical advantages providing direct insight into patterns of walking device used, an under-researched area. We present a novel method to estimate speed and distance traveled from a single rollator-mounted IMU. The method was developed using data collected from ten rollator users performing a series of walking tasks including obstacle negotiation. The IMU data is first pre-processed to account for noise, orientation offset, and rotation-induced accelerations. The method then uses a two-stage approach. First, activity classification is used to separate the rollator data into one of three classes (movement, turning, or other). Subsequently, the speed of movement and distance traveled is estimated, using a separate estimation model for each of the three classes. The results showed high classification accuracy (precision, recall, and F1 statistics all >0.9). Speed estimation showed mean absolute errors below 0.2 m/s. Estimates for distance traveled showed errors which ranged from 5% (straight line walking) to over 70%. The results showed some promise but further work with a larger data set is needed to confirm the performance of our approach.

The modern sedentary lifestyle is negatively influencing human health, and the current guidelines recommend at least 150 min of moderate activity per week. However, the challenge is how to measure human activity in a practical way. While accelerometers are the most common tools to measure activity, current activity classification methods require calibration studies or labelled datasets—requirements that slow the research progress. Therefore, there is a pressing need to classify and quantify human activity efficiently. In this work, we propose an unsupervised approach to classify activities from accelerometer data using hidden semi-Markov models. We tune and infer the model parameters on accelerometer data from the UK Biobank and select the optimal model based on features used and informativeness of the prior. The best model achieves an average correlation of 0.4 between the inferred activities and the reference ones, with the overall physical activity obtaining a correlation of 0.8. Additionally, to prove the clinical significance of the method, we validate it by performing a linear regression between the inferred activities and anthropometric measures such as BMI and waist circumference. We show that for a sedentary behaviour and total physical activity, the proposed method achieves comparable regression coefficients to the reference labelled dataset. Moreover, the proposed method achieves a good agreement with a labelled dataset for daily time spent in a sedentary behaviour and total physical activity. The unsupervised nature of the method allows for a data-driven classification that does not require calibration studies or labelled datasets and can thus facilitate both clinical research as well as lifestyle recommendations.