Search

The current 3D-ConvNet-based action recognition algorithms generally use GAP to compress feature information. However, it leads to issues of information loss, redundancy, and network overfitting. To address these issues and enhance the retention of high-level semantic information extracted by the convolutional layer, this paper proposed an action recognition al- gorithm based on GFDP. Firstly, DCT shows that GAP is a special case of feature decomposition in the frequency domain. Therefore, the algorithm introduced more frequency components to increase the specificity between feature channels and reduce the information redundancy after information compression. Secondly, to better suppress the overfitting problem, the algorithm introduced the batch normalization strategy to the convolutional layer and extended it to the fully connected layer of the action recognition model with ERB-Res3D as the skeleton to optimize the data distribution. Finally, this paper verified the proposed method on the UCF101 dataset. The results reveals that the model's computational load is 3.5 GFlops, with 7.4 million para- meters. The final recognition accuracy improved by 3.9% based on the ERB-Res3D model and 17.4% based on the original Res3D model. This improvement effectively achieves more accurate behavior recognition results. [ABSTRACT FROM AUTHOR]

To achieve Wi-Fi cross-domain human activity perception that was not dependent on target domain data, a domain-generalization human activity recognition model based on CSI instance normalization called INDG-Fi was proposed. The instance normalization standardization was utilized to remove domain information from the representation of CSI features by INDG-Fi. Then action classifiers and domain classifiers were constructed for shared feature extraction. By employing activity bias learning and adversarial domain learning, the model biased the features extracted from the encoding layer towards signal variations caused by human actions while moving away from domain signals. To enhance the model’s focus on subcarrier signals that were more significantly influenced by human actions, a subcarrier attention module was incorporated into the encoding layer. The implemented results demonstrate that the proposed INDG-Fi achieves perceptual accuracies of 97.99% and 92.73% for unseen users and locations, respectively, thus enabling robust cross-domain perception.

The existing methods for human activity recognition using wearable sensors could not capture the structural information between the sampling points of time series effectively and might ignore the potential connections between samples. To address this issue, a graph convolutional neural network model with hybrid time-frequency and structural characteristics was proposed for human activity recognition. Firstly, the time-frequency characteristics of the original signal were obtained by wavelet-packet transform, and the spatio-temporal graph was further constructed to extract the structural characteristics to identify the dynamic characteristics between the sampling points. The distance constraint was added to the structural characteristics to weaken the influence of long-distance neighbors on the central node on the spatio-temporal graph. Considering that the extraction of structural characteristics was greatly affected by the topological relationship of the spatio-temporal graph. The time-frequency characteristics of the samples to construct the input topology of the graph convolutional neural network were selected, and the timefrequency and structural characteristics were combined as the input features of the network. Finally, the input features propagated along the input topology to obtain the final classification result. To evaluate the performance of the proposed model, experiments were conducted on the WHARF and DataEgo datasets. Results in terms of F1 scores indicated that the proposed model outperformed existing convolutional neural network-based methods, achieving a maximum improvement of 19. 58 percentage points on the WHARF dataset and 26. 44 percentage points on the DataEgo dataset. It demonstrated that the proposed model could effectively enhance the capability of activity recognition by exploiting dynamic characteristics. [ABSTRACT FROM AUTHOR]

Copyright of Telecommunication Engineering is the property of Telecommunication Engineering 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.)

Copyright of Journal of Data Acquisition & Processing / Shu Ju Cai Ji Yu Chu Li is the property of Editorial Department of Journal of Nanjing University of Aeronautics & Astronautics 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.)

Through-the-wall radar can penetrate walls and realize indoor human target detection. Deep learning is commonly used to extract the micro-Doppler signature of a target, which can be used to effectively identify human activities behind obstacles. However, the test accuracy of the deep-learning-based recognition methods is low with poor generalization ability when different testers are invited to generate the training set and test set. Therefore, this study proposes a method for recognition of anomalous human gait termination based on micro-Doppler corner features and Non-Local mechanism. In this method, Harris and Moravec detectors are utilized to extract the corner features of the radar image, and the corner feature dataset is established in this manner. Thereafter, multilink parallel convolutions and the Non-Local mechanism are utilized to construct the global contextual information extraction network to learn the global distribution characteristics of the image pixels. The semantic feature maps are generated by repeating four times the global contextual information extraction network. Finally, the probabilities of human activities are predicted using a multilayer perceptron. The numerical simulation and experimental results demonstrate that the proposed method can effectively identify such abnormal gait termination activities as sitting, lying down, and falling, among others, which occur in the process of indoor human walking, and successfully control the generalization accuracy error to be no more than \begin{document}$ 6.4\% $\end{document} under the premise of increasing the recognition accuracy and robustness.

Human activity recognition widely exists in applications such as sports management and activity classification.The current human activity recognition applications are mainly divided into three types: camera-based, wearable device-based, and Wi-Fi awareness-based.Among them, the camera-based human activity recognition application has the risk of privacy leakage, and the wearable device-based human activity recognition application has problems such as short battery life and poor accuracy.Human activity recognition based on Wi-Fi sensing generally uses Wi-Fi network cards or software-defined radio devices to identify the rules of channel state information changes, so as to infer user activity.It does not have the problems of privacy leakage and short battery life.But Wi-Fi network cards need to rely on computers and software-defined radio platforms are expensive, which greatly limit the application scenarios of Wi-Fi sensing.Aiming at the above problems, a human activity recognition system based on the low-cost IoT chip ESP32 was proposed.Specifically, the Hampel filter and Gaussian filter were used to preprocess the channel state information obtained by ESP32.Then, the principal component analysis and discrete wavelet transform were utilized to reduce the dimension of the data.Finally, the K-nearest neighbor (KNN) algorithm was applied to classify data.The experimental results show that the system can achieve a recognition accuracy which close to the current mainstream Wi-Fi perception system (Intel 5300 network card) when only two ESP32 nodes are deployed, and the average accuracy rate for the six activities is 98.6%.

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. 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.)

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. 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.)

Rapid extraction and precise identification of freshwater aquaculture ponds can play an important role in the management and decision-making of the aquaculture industry. Satellite remote sensing can be expected to provide the critical extraction of freshwater aquaculture ponds, due to its rapid, timely and effective way of Earth observation. However, it is still elusive to delineate the surface area and the variation in the aquaculture ponds from the satellite remote sensing images. In this study, an improved classification was introduced to delineate the changes in aquaculture ponds from Landsat remote sensing images using the coordinate attention with the U-Net neural network model. Firstly, the dataset was collected from the freshwater aquaculture ponds in the Gaochun district of Nanjing City, Jiangsu Province, China. Landsat remote sensing images were also captured from 1985 to 2021, and supplemented by GF-1, GF-2 satellite data and field survey data. Secondly, a coordinate attention model with U-Net as the backbone was created to fully extract the spatial information of features, where the information remained on the fragmented aquaculture ponds. Finally, the long-term Landsat images were selected to explore the temporal and spatial changes of freshwater aquaculture ponds. Six state-of-the-art models were also utilized to verify the improved model. The experimental results show that the performance of the improved model outperformed the rest, in terms of extraction accuracies. Furthermore, there was a dramatic variation in the surface area and distribution of freshwater aquaculture ponds in the study period. Specifically, the surface areas of aquaculture ponds increased from 0.48 to 36.92 km² in the early stages from 1985 to 2000, with an annual increase of 2.43km². There was rapid growth in the second stage from 2000 to 2017. Among them, the surface areas of aquaculture ponds increased from 36.92 to 234.47 km², with an annual increase of 11.62 km². A gradual shrinkage was found in the surface areas of freshwater aquaculture ponds from 234.47 to 209.58 km² in the third stage from 2017 to 2021, with an annual decrease of 6.22 km². Moreover, the social and economic factors can be attributed to the main driving factors for the dramatic variation in the surface area of the aquaculture areas. In summary, the high overall accuracies of the improved model were achieved to rapidly extract the aquaculture ponds. Additionally, the food demand and economic value can be the significant driving factors for the rapid changes in the aquaculture ponds, while the policies and human activities can be the key factors for the area changes of the aquaculture ponds. The finding can also provide technical support for the scientific management and decision-making of the aquaculture industry. [ABSTRACT FROM AUTHOR]

Copyright of Geography & Geographic Information Science is the property of Geography & Geo-Information Science Editorial Office 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.)

Copyright of Telecommunication Engineering is the property of Telecommunication Engineering 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.)

In order to solve the problems of poor anti-interference and invasion of user privacy in traditional methods of using cameras for human activity recognition, this paper proposed a human activity recognition network based on spatiotemporal attention of millimeter wave radar 3D point cloud data to achieve accurate perception of intelligent application context. The network firstly used a secondary sliding time window to accumulate and separate the point cloud data generated by human activities as the input of the classifier, then used the PointLSTM unit to aggregate point features and states according to the point cloud coordinate relationship to extract the time sequence features of human activities, and then spliced temporal-spatial features, reduced the overall network computation and enhanced the network’s aggregation ability for local featured through sampling grouping modules, and finally used a stacked attention module to deeply fuse global and local features in temporal-spatial point cloud data to complete the accurate classification of human activities. This paper used millimeter wave radar to collect point cloud datasets of various human activities, the experimental results show that the average accuracy of the proposed spatiotemporal attention network can reach 98.64%,which can effectively identify complex and small-difference human activities, and meet the requirements of the human activity recognition system. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Chongqing University of Posts & Telecommunications (Natural Science Edition) is the property of Chongqing University of Posts & Telecommunications 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.)

Copyright of Railway Standard Design is the property of Railway Standard Design Editorial Office 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.)

Copyright of Telecommunication Engineering is the property of Telecommunication Engineering 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.)

In recent years, various human action recognition algorithms have achieved excellent recognition performance based on a large number of labeled samples. However, in practical applications, acquiring training samples and their corresponding labels is time-consuming and laborious, which limits the actual implementation of the algorithm. This paper summarized the deep learning algorithms for action recognition under weak supervision and few-shot learning. Firstly, in the case of weak supervision, it classified and summarized the semi-supervised action recognition methods and unsupervised domain adaptation video action recognition methods separately. Then, it reviewed video action recognition algorithms based on few-shot learning in detail. Further, it summarized the relevant human behavior datasets and analyzed and compared the performance of various relevant video action recognition algorithms on these datasets. Finally, the paper summarized the full text and discusses the future development trend of human action recognition. [ABSTRACT FROM AUTHOR]

The emergence of geographic big data provides a new data source for the study of urban spatial structures. Identifying the polycentric urban structure based on geographic big data is currently a hot research topic in academic communities. This study proposed a method for identifying the polycentric urban structure based on multi - source geographic big data. First, the spatial units in the study area were determined using a region segmentation algorithm based on drainage divides. Then, the urban centers and subcenters were identified using the two -stage algorithm for urban center identification. Finally, the identification results were compared and verified. The results of this study are as follows: (1) The region segmentation algorithm based on drainage divides can effectively identify the spatial features of nighttime light data, and the basic spatial units acquired using this algorithm can be used to identify urban spatial structures; (2) The urban centers identified based on the Weibo (MicroBlog) check-in data, which can effectively reflect urban human activities, and the two - stage algorithm for urban center identification are roughly consistent with those set in the urban planning. Therefore, the method proposed in this study is of great significance for expanding the application scope of geographic big data and enriching the existing research methods for urban spatial structures. [ABSTRACT FROM AUTHOR]

With the rapid socio-economic development and the increasing demand for natural resources in China, the protection of natural reserves is facing increasing difficulties. The remote sensing - based research on monitoring the disturbance and the restoration of mangrove forests through time series analysis is still in its initial stage. Moreover, time series algorithms are highly complex. Based on the LandTrendr time segmentation algorithm of Google Earth Engine (GEE) and the Landsat image time-series data, this study investigated the disturbance to mangrove forests in the Dongzhaigang Mangrove Nature Reserve during 1990-2020. The results are as follows: (1) A total of 42.39 hm of mangrove forests were disturbed during 1990-2020, among which the largest disturbance area of 12.78 hm occurred in 2014; (2) During 1990-2020, minor, moderate, and severe disturbances accounted for 65.39%, 30.78%, and 3.83%, respectively; (3) The overall identification accuracy of the pixels of mangrove forests subject to changes was 89.50%, and the overall detection accuracy of years witnessing disturbance was 88%, with a Kappa coefficient of 0.79. This study analyzed the years and areas of the disturbance to mangrove forests in the Dongzhaigang Mangrove Nature Reserve over 30 years based on LandTrendr. Moreover, this study analyzed the disturbance factors according to the actual situation and concluded that human activities are the main disturbance factor, followed by natural factors, such as diseases, pests, and extreme weather events. This study will provide a scientific basis and a decision reference for the management of the mangrove forest reserve. [ABSTRACT FROM AUTHOR]

Copyright of Journal of South China Normal University (Natural Science Edition) / Huanan Shifan Daxue Xuebao (Ziran Kexue Ban) is the property of Journal of South China Normal University (Natural Science Edition) Editorial Office 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.)

Copyright of Arid Zone Research / Ganhanqu Yanjiu is the property of Arid Zone Research Editorial Office 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.)

Microdroplets have many unique features derived from the large surface-volume ratio and the ultrahigh electric field at the air-water interface, therefore they can remarkably accelerate various slow reactions and trigger reactions that won’t occur in bulk solutions. The size of atmospheric aerosols is similar to that of the microdroplets, so reactions performed in microdroplets and ambient conditions can act as good simulations of important atmospheric processes, which will provide a better understanding of the role of aerosols. Aldehydes are important oxidized species in the atmosphere derived from releases of human activities, such as vehicle emissions and biomass burning, which can generate volatile organic compounds (VOCs) and are highly related to salt-rich aerosol formation through further oxidation. Studying of aldehydes’ hydration and oxidation reactions is very helpful for understanding the complex conversion process from VOCs to secondary organic aerosols (SOA). In this work, several aromatic aldehydes with different electron donating or withdrawing groups were studied by spraying their solutions into microdroplets, then directly characterizing them by online mass spectrometry. It showed that these aromatic aldehydes could be oxidized into carboxylic acids under the space-restricting condition of microdroplets, and the hydration reactions of aromatic aldehydes could be induced by carboxylate under the extreme pH environment and abundant reactive oxygen species provided by the air-water interface of microdroplets. Together, the products of oxidation and hydration reactions would combine to yield relatively unstable “carboxylate-aldehyde hydrate” complex ions. The lifetime of the ions was roughly estimated by conducting a collision-induced disassociation (CID) experiment and comparing it to the instrument running procedure, and the structure was checked using CID and a cross-forming reaction between two different aldehydes. This work revealed that the complex ions were potential intermediates of the atmospheric aldehyde oxidation reaction, providing a previously unknown pathway of atmospheric carboxylic acids and SOA. This result supported the view that atmospheric oxidation might take place in aerosols, such as fog and clouds, which was the missing part of SOA sources and matched the observations. However, unlike other liquid phase or gaseous simulation research of the oxidation process of aldehydes, the oxidation reaction would occur even without the participation of UVs or hydrogen peroxide addition in microdroplets, because the air-water interface could generate a large number of reactive oxygen species (ROS) like hydroxyl radicals and hydrogen peroxide, implicating that the air-water interface of aerosols might be an important place to perform many atmospheric reactions. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Frontiers of Computer Science & Technology is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. 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.)

Copyright of Journal of Beijing University of Technology is the property of Journal of Beijing University of Technology, Editorial Department 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.)

In the past few decades, smartphone-based human activity recognition research has played an important role in many fields, including smart buildings, healthcare, and the military. However, the CPU and storage space of smartphones are very limited, so developing a lightweight human activity recognition learning model has become a research focus and hot spot in this field. To address the abovementioned problems, this paper proposed a lightweight human activity recognition learning model based on the nearest neighbor component analysis (NCA), L2 regularization, and stochastic configuration networks (SCNs). In the proposed model, aiming first at the problem of high dimension and poor separability exhibited by the human activity data, NCA was used to select a subset of highly relevant data from the dataset to improve the lightness of calculation using the learning algorithm in the modeling process and recognition accuracy of the established model. Second, to prevent the occurrence of overfitting when there are too many hidden layer nodes in SCNs, the L2 regularization method was adopted to enhance the generalization ability of SCNs. At the same time, the method of using the supervision mechanism to restrict the generation of hidden layer parameters greatly improved the lightness of the SCNs model. Finally, the proposed learning model and other learning models were verified experimentally on the UCI human activity recognition dataset. Experimental results show that compared with SCNs, the proposed L2−SCNs model reduces the lightness of the number of parameters by 20% and helps improve the accuracy of the model. The introduction of the NCA method has greatly facilitated the recognition accuracy and lightness (modeling time) of the L2−SCNs model, increasing by 3.41% and 70.24%, respectively. Moreover, compared with other state-of-the-art models, such as the support vector machine and long short-term memory network, the proposed model achieves the best recognition accuracy of 97.48% in the shortest time. To sum up, the model proposed herein is a lightweight human activity recognition model with exceptional recognition accuracy and a fast modeling speed.

Copyright of Systems Engineering & Electronics is the property of Journal of Systems Engineering & Electronics Editorial Department 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.)

Copyright of Bulletin of Soil & Water Conservation is the property of Bulletin of Soil & Water Conservation Editorial Office 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.)

Copyright of Experimental Technology & Management is the property of Experimental Technology & Management Editorial Office 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.)

Copyright of Arid Zone Research / Ganhanqu Yanjiu is the property of Arid Zone Research Editorial Office 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.)

Copyright of Journal of South China Normal University (Natural Science Edition) / Huanan Shifan Daxue Xuebao (Ziran Kexue Ban) is the property of Journal of South China Normal University (Natural Science Edition) Editorial Office 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.)

Copyright of Journal of Frontiers of Computer Science & Technology is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. 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.)

Copyright of Journal of South China University of Technology (Natural Science Edition) is the property of South China University of Technology 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.)

Copyright of Journal of Test & Measurement Technology is the property of Publishing Center of North University of China 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.)

Copyright of Journal of Chongqing University of Posts & Telecommunications (Natural Science Edition) is the property of Chongqing University of Posts & Telecommunications 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.)

Copyright of China Rural Water & Hydropower is the property of China Rural Water & Hydropower Editorial Office 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.)

Copyright of Chinese Journal of Liquid Crystal & Displays is the property of Chinese Journal of Liquid Crystal & Displays 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.)

Health monitoring based on human activity recognition (HAR) is an important means to discover health abnormalities.However,in daily activity recognition,it is difficult to obtain training samples containing all possible activity categories in advance.When new categories appear in the prediction stage,the traditional support vector machine (SVM) will incorrectly classify them as known category.A robust classifier should be able to distinguish the newly added categories so that they can be processed differently from the known categories.This paper proposes a human activity recognition method based on class increment SVM,and the idea of hypersphere is introduced,which can not only identify known activity categories with high accuracy,but also detect new categories.The multiple hyperspheres obtained through training divide the entire feature space,so that the classifier has the ability to detect newly added activity categories.The experimental results show that compared with the traditional multi-class SVM method,our method can realize the detection of new categories without significantly reducing the classification effect of known categories,thereby improving the classifier's ability to recognize human activity in an open environment.

Human activity recognition system based on deep learning and Wi-Fi sensing has gradually become the mainstream research field and has been developed in recent years.However, related systems heavily rely on training with huge labeled samples to reach a high accuracy, which is labor-intensive and unrealistic for many real-world scenarios.To solve this problem, a system that combines active learning with Wi-Fi based human activity recognition—ALSensing was proposed, which was able to train a well-perform classifier with limited labeled samples.ALSensing was implemented with commercial Wi-Fi devices and evaluated in six real environments.The experimental results show that ALSensing achieves 52.83% recognition accuracy using 3.7% of total training samples, 58.97% recognition accuracy using 15% of total training samples, while the existing full-supervised system reaches 62.19% recognition accuracy.It demonstrates that ALSensing has a similar performance with baseline but requires much less labeled samples.

A radar human behavior perception system has penetration detection ability, which gives it a wide application prospect in the fields of security, rescue, medical treatment, and so on. Although the development of deep learning technology has promoted radar sensor research in human behavior perception, it requires more prompted dataset availability. This paper provides a four-dimensional imaging dataset of human activity using ultra-wideband radar, UWB-HA4D, which uses three-dimensional ultra-wideband multiple-input multiple-output radar as the detection sensor to capture the range-azimuth-height-time four-dimensional activity data of a human target. The dataset contains the activity data of 2757 groups for 11 human targets, including 10 common activities, such as walking, waving, and boxing. It also contains penetration and nonpenetration detection experimental scenarios. The radar system parameters, data generation process, data distribution, and other information of the dataset are introduced in detail herein. Meanwhile, several deep learning algorithms that are based on the PaddlePaddle framework and are widely used in the computer version field are applied to this dataset for human activity recognition. The experimental comparison results can be used to provide references for scholars and facilitate further investigation and research on this basis.

Objectives: To explore the relationship between urban ecological environment and human activities is an important research content in the current urbanization process. With the in-depth development of the era of big data, multi-source data ubiquitous in the Internet has been fully excavated, which has played an important role in promoting the research of urban ecological environment. Methods: Based on multisource data, we propose to construct human activity indicators (residential area walkability index, street vitality index, urban function mixing index) using point of interest(POI), OpenStreetMap (OSM) and residential area data, and urban ecological environment indicator (remote sensing ecological index)using remote sensing images. Combing machine learning models such as polynomial regression(PLR), random forest regression( RFR), extreme gradient boosting regression(XGB) and support vector regression machine(SVR), it is effective to make regression analysis on urban ecological environment and human activity indicators. By comparing the performance of different models in this dataset, the relationship between the urban ecological environment and human activities is revealed.Results:We demonstrate the application of our method using a case study of Nanchang city. The results show that: (1)The three indexes of human activities all present a central high and gradually decrease to the surroundings, while the urban ecological environment indicators show an opposite trend. (2) From the analysis of the performance results of each model on the dataset, XGB has the best regression effect, followed by PLR.(3)There is a strong negative correlation between the urban ecological environment and human activities, and the street vitality index, the urban function mixing index are more relevant to the urban ecological environment, and the walkability index of the residential area is less relevant to the urban ecological environment. (4)In areas where human activities have less impact, the urban ecological environment will be disturbed by other factors, resulting in the low prediction accuracy, while the prediction accuracy in areas with strong human activities is high.Conclusions: Using multisource data and machine learning models can provide an important reference for exploring the relationship between urban ecological environment and human activities. [ABSTRACT FROM AUTHOR]

【Objective】Land use is comprehensively affected by natural conditions and socio-economic factors, and the study of time series changes in land use is conducive to providing a scientific basis for the rational use of urban land resources.【Method】Landsat satellite remote sensing images of six periods were selected at five-year intervals for land use classification from 1995 to 2020, and the land use structure, dynamics and land use transfer in Guangzhou were analyzed. The influence intensity and interaction mechanism of various factors on land use change in different periods were discussed by using geographic detectors. 【Result】（i） Under the scenario of long time series（1995-2020） urban land use change, geodetector model could analyze the factors of urban land use change in different periods and reveal its change rules.（ii） From 1995 to 2020, the construction land in Guangzhou showed a trend of ‘stable-slow increase-sharp increase’, while the cultivated land showed a trend of ‘slow decrease-stable-substantial decrease’, and the transferred area mainly came from cultivated land and forest land. The forest land and grassland showed a monotonous decreasing trend, mainly transferred out as construction land and forest land. The water area fluctuated up and down, mainly from cultivated land. Unused land was gradually reduced, mainly converted into construction land.（iii）In different periods from 1995 to 2020, the influence of driving factors of land use change in Guangzhou was different. Among them, the influence of socioeconomic factors on land use change was higher than that of physical geography.（iv） In different periods from 1995 to 2020, the q value of the interaction factor was higher than that of the single factor q value, which showed that the interaction between the factors was two-factor enhancement. 【Conclusion】The impacts of human activities on land use change are obviously different in spatial distribution. The overlapping of multiple factors will aggravate the conflict of land use. Reducing the impact of human activities on the environment is an important way to achieve sustainable development. [ABSTRACT FROM AUTHOR]

Copyright of Asian Journals of Ecotoxicology is the property of Gai Kan Bian Wei Hui 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.)

Copyright of CT Theory & Applications is the property of Editorial Department of CT Theory & Applications 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.)

Existing human pose action recognition methods ignore the role of the previous pose estimation algorithms and do not fully extract action features. This paper proposes an action recognition method that combines lightweight Openpose and attention-guided graph convolutional network. The method includes an Openpose algorithm based on shufflenet and a graph convolution algorithm based on attention of different scales of adjacency matrices. The input video is processed by lightweight Openpose to obtain 18 human body keypoint information, which is expressed as the basic spatiotemporal graph data form. The adjacency matrix corresponding to the neighbor information of different scales of the node calculates the influence through the self- attention mechanism, and weights the adjacency matrix of each scale. The input graph convolutional network is merged to extract features. The extracted discriminative features are output action categories through global average pooling and softmax classifier. Experiments on the Le2i Fall Detection dataset and the custom UR-KTH dataset show that the accuracy of action recognition is 95.52% and 95.07%, respectively, achieving the expected results. [ABSTRACT FROM AUTHOR]

Collecting wearable motion sensor signals and using transfer learning to overcome the inconsistency of data distribution to identify the daily behavior of the human body are very popular technologies. Using wearable sensors to collect signals will result in generating noise samples that affect the transfer effect. Traditional algorithms lack the processing of these samples. To solve this problem， the traditional algorithm was improved by introducing a sample screening algorithm based on Mahalanobis distance， and a transfer learning algorithm T-WMD was proposed that can be used for human activity recognition. And compared with other five algorithms on two public human activity recognition data sets， the results show that the algorithm proposed in this paper can effectively improve the effect of transfer learning. [ABSTRACT FROM AUTHOR]

Copyright of Electric Machines & Control / Dianji Yu Kongzhi Xuebao is the property of Electric Machines & Control 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.)

An ecosystem has been much more fragile against global warming in southwest China in recent years. Rapid and accurate monitoring of vegetation growth can greatly contribute to clarifying the response mechanism of vegetation growth to the driving factors. Therefore, it is very necessary to explore the spatiotemporal variation in the vegetation net primary productivity (NPP) under climate change and human activities. In this study, a systematic investigation was made to determine the spatiotemporal variation of vegetation NPP, and the effects of climate change and human activities on the regional and landform unit scales from 2000 to 2020. The moderate-resolution imaging spectroradiometer (MODIS) NPP was selected to combine with in situ meteorological data, population density, night-time light, and land use type datasets. Multiple mathematical models were also utilized, such as Theil-Sen Median trend, variance inflation factor, multivariate regression, residual and partial correlation analysis. Results showed that 1) there was an increasing fluctuation in the overall, predicted, and residual vegetation NPP over the past 21 years from 2000 to 2020, whereas, the Tibetan Plateau presented a relatively decrease trend. Among them, the vegetation NPP, predicted vegetation NPP, and residual vegetation NPP in the Sichuan Basin exhibited the most significant increasing trend, with a magnitude of 7.14, 3.72, and 3.44 g/(m2·a), respectively. 2) The impact of climate change on the vegetation NPP was higher than that of human activities during the study period. The areas with extremely significant and significant increases accounted for 45.18% and 18.55% of the study area, respectively, under the climate change and human activities. By contrast, the areas with extremely significant and significant increases in the predicted vegetation NPP and residual vegetation NPP in the Sichuan Basin accounted for the highest proportion in all landform units, with a magnitude of 69.42% and 50.75%, respectively. The vegetation NPP variation was dominated by the changes of temperature-like factors at the regional and landform unit scales. Specifically, the relative humidity and atmospheric pressure showed inhibitory effects on vegetation growth in southwest China. Correspondingly, the conversion of cultivated land to urban land can be an important reason for the decline of vegetation NPP, whereas, the conversion of non-cultivated land to the cultivated land and the non-forest land to forest land can be another important reason for the increase of vegetation NPP during the study period. There was an insignificant correlation of vegetation NPP with the population density and night-time light at the regional scale. Sichuan Basin exhibited the highest correlation of the vegetation NPP with population density and night-time light in all landform units at the landform unit scale. It infers that the vegetation growth in the Sichuan Basin was more associated with human activities than the rest landform units. The finding can provide a strong theoretical basis for the regional monitoring of the vegetation growth, quality assessment of the ecological environment, and benefit evaluation of forestry ecological engineering [ABSTRACT FROM AUTHOR]

Copyright of Journal of Frontiers of Computer Science & Technology is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. 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.)

Aiming at the problems of poor training effect, over-fitting phenomenon in human action recognition based on small sample scenarios, slow convergence speed of traditional confrontation generation network, and high requirements for computer performance, this paper proposed solutions from data enhancement and hyperparameter optimization. Firstly, it built the A WR1 243 radar data acquisition platform to preprocess the collected echo signals. Secondly, it used STFT for time-frequency analysis and the newly proposed FT_SSIM algorithm for data enhancement. Furthermore, it used the proposed ICAGA_CNN for classification and recognition, and compared the experiments with traditional data enhancement algorithms and hyperparameter optimization. In order to verify the proposed algorithm that has a certain generalization ability, the public KTH human body motion data set and the actual measurement data of radar were used for verification. Experimental results show that, on the one hand, the proposed algorithm effectively avoids the occurrence of over-fitting in small sample scenarios, reduced the requirements of traditional data enhancement on computer performance, and accelerated the speed of convergence. On the other hand, the proposed algorithm has better recognition accuracy, with an average recognition rate of 98. 5%. This also shows that the proposed algorithm has a good performance in radar action recognition in small sample scenarios. [ABSTRACT FROM AUTHOR]

Collective activity recognition is an important issue in the field of computer vision, with wide application and urgently to be solved. With the development of deep neural network, the width and depth of collective activity recognition and understanding are expanding. Through investigating the research literature of collective activity recognition in recent ten years, the problem definition of the collective activity recognition is determined. The existing problems and challenges of collective activity recognition are pointed out. In the framework of deep neural network, this paper describes the development of the collective activity recognition algorithm from the early stage which only classifies and recognizes the collective activity categories to the current stage which focuses more on the understanding of the details of activities in the group behavior. Then, based on the network architectures such as CNN/3DCNN, Two-Stream Network, RNN/LSTM and Transformer, the core network architecture and the main research ideas of the mainstream collective activity recognition methods are mainly introduced. The recognition performance of these algorithms on common datasets is compared. The commonly used collective activity recognition datasets labeled with multilevel labels such as collective activity types and individual activity categories are combed and compared. Through objective and fair discussion and analysis of the advantages and disadvantages of various algorithms, it is expected to prompt readers to propose new solutions or new problems of collective activity recognition. Finally, the future development of collective activity recognition is prospected, which is expected to stimulate new research directions. [ABSTRACT FROM AUTHOR]