Showing papers in "Journal of Sensors in 2021"

Visual and Visual-Inertial SLAM: State of the Art, Classification, and Experimental Benchmarking

Abstract: Simultaneous Localization and Mapping is now widely adopted by many applications, and researchers have produced very dense literature on this topic. With the advent of smart devices, embedding cameras, inertial measurement units, visual SLAM (vSLAM), and visual-inertial SLAM (viSLAM) are enabling novel general public applications. In this context, this paper conducts a review of popular SLAM approaches with a focus on vSLAM/viSLAM, both at fundamental and experimental levels. It starts with a structured overview of existing vSLAM and viSLAM designs and continues with a new classification of a dozen main state-of-the-art methods. A chronological survey of viSLAM’s development highlights the historical milestones and presents more recent methods into a classification. Finally, the performance of vSLAM is experimentally assessed for the use case of pedestrian pose estimation with a handheld device in urban environments. The performance of five open-source methods Vins-Mono, ROVIO, ORB-SLAM2, DSO, and LSD-SLAM is compared using the EuRoC MAV dataset and a new visual-inertial dataset corresponding to urban pedestrian navigation. A detailed analysis of the computation results identifies the strengths and weaknesses for each method. Globally, ORB-SLAM2 appears to be the most promising algorithm to address the challenges of urban pedestrian navigation, tested with two datasets.

Kalman Filter: Historical Overview and Review of Its Use in Robotics 60 Years after Its Creation

Abstract: Due to its widespread application in the robotics field, the Kalman filter has received increased attention from researchers. This work reviews some of the modifications conducted on to this algorithm over the last years. Problems such as the consistency, convergence, and accuracy of the filter are also dealt with. Sixty years after its creation, the Kalman filter is still used in autonomous navigation processes, robot control, and trajectory tracking, among other activities. The filter is not only restricted to robotics but is also present in different fields, such as economics and medicine. In addition, the characteristics of each modification on this filter are analyzed and compared.

A Lightweight Model for Traffic Sign Classification Based on Enhanced LeNet-5 Network

Abstract: For several years, much research has focused on the importance of traffic sign recognition systems, which have played a very important role in road safety. Researchers have exploited the techniques of machine learning, deep learning, and image processing to carry out their research successfully. The new and recent research on road sign classification and recognition systems is the result of the use of deep learning-based architectures such as the convolutional neural network (CNN) architectures. In this research work, the goal was to achieve a CNN model that is lightweight and easily implemented for an embedded application and with excellent classification accuracy. We choose to work with an improved network LeNet-5 model for the classification of road signs. We trained our model network on the German Traffic Sign Recognition Benchmark (GTSRB) database and also on the Belgian Traffic Sign Data Set (BTSD), and it gave good results compared to other models tested by us and others tested by different researchers. The accuracy was 99.84% on GTSRB and 98.37% on BTSD. The lightness and the reduced number of parameters of our model (0.38 million) based on the enhanced LeNet-5 network pushed us to test our model for an embedded application using a webcam. The results we found are efficient, which emphasize the effectiveness of our method.

An Exhaustive Research on the Application of Intrusion Detection Technology in Computer Network Security in Sensor Networks

Abstract: Intrusion detection is crucial in computer network security issues; therefore, this work is aimed at maximizing network security protection and its improvement by proposing various preventive techniques. Outlier detection and semisupervised clustering algorithms based on shared nearest neighbors are proposed in this work to address intrusion detection by converting it into a problem of mining outliers using the network behavior dataset. The algorithm uses shared nearest neighbors as similarity, judges whether it is an outlier according to the number of nearest neighbors of a data point, and performs semisupervised clustering on the dataset where outliers are deleted. In the process of semisupervised clustering, vast prior knowledge is added, and the dataset is clustered according to the principle of graph segmentation. The novelty of the proposed algorithm lies in outlier detection while effectively avoiding the dependence on parameters, thus eliminating the influence of outliers on clustering. This article uses real datasets: lypmphography and glass for simulation purposes. The simulation results show that the algorithm proposed in this paper can effectively detect outliers and has a good clustering effect. Furthermore, the experimentation reveals that the outlier detection-based SCA-SNN algorithm has the best practical effect on the dataset without outliers, clearly validating the clustering performance of the outlier detection-based SCA-SNN algorithm. Furthermore, compared to the other state-of-the-art anomaly detection method, it was revealed that the anomaly detection technology based on outlier mining does not require a training process. Thus, they overcome the current anomaly detection problems caused due to incomplete normal patterns in training samples.

College English Teaching Quality Evaluation System Based on Information Fusion and Optimized RBF Neural Network Decision Algorithm

Abstract: In the process of deepening and developing the current higher education reform, people pay more and more attention to the research of college English education. The key to improve the college English education is to improve the quality of education, and learning evaluation is the key measure to improve the quality of education and training. This paper mainly studies the college English teaching quality evaluation system based on information fusion and optimized RBF neural network decision algorithm. This paper analyzes the main problems and complexity of creating an ideal learning quality evaluation system. On the basis of analyzing the advantages and disadvantages of the previous learning quality evaluation methods, this paper summarizes the existing learning quality evaluation methods and puts forward some suggestions according to the existing evaluation methods. A learning quality evaluation model based on RBF algorithm of neural network is proposed. RBF regularization network method, RBF neural network decision algorithm, and experimental investigation method are used to study the college English teaching quality evaluation system based on information fusion and optimization of RBF neural network decision algorithm. By innovating teaching methods and enriching teaching means, college students’ thirst for English knowledge can be aroused, and teachers’ teaching level can be improved. The results show that 50% of college students think that the level of college English teaching is average and needs to be improved. In the performance evaluation system of college English teaching quality based on information fusion and optimized RBF neural network decision algorithm, it is necessary to establish a learning evaluation system, monitor the learning quality in real time, find problems and improve them in time, and recognize the current situation of education.

FBG-Based Sensing for Structural Health Monitoring of Road Infrastructure

Abstract: Public road infrastructure is developed all around the world. To save resources, ensure public safety, and provide longer-lasting road infrastructure, structural health monitoring (SHM) applications for roads have to be researched and developed. Asphalt is one of the largest used surface materials for the road building industry. This material also provides relatively easy fiber optical sensor technology installment, which can be effectively used for SHM applications—road infrastructure monitoring as well as for resource optimization when road building or their repairs are planned. This article focuses on the research of the fiber Bragg grating (FBG) optical temperature and strain sensor applications in road SHM, which is part of the greater interdisciplinary research project started at the Riga Technical University in the year 2017. Experimental work described in this article was realized in one of the largest Latvian road sites where the FBG strain and temperature sensors were installed into asphalt pavement, and experiments were carried out in two main scenarios. Firstly, in a controlled environment with a calibrated falling weight deflectometer (FWD) to test the installed FBG sensors. Secondly, by evaluating the real-time traffic impact on the measured strain and temperature, where different types of vehicles passed the asphalt span in which the sensors were embedded. The findings in this research illustrate that by gathering and combining data from calibrated FWD measurements, measurements from embedded FBG optical sensors which were providing the essential information of how the pavement structure could sustain the load and information about the traffic intensity on the specific road section, and the structural life of the pavement can be evaluated and predicted. Thus, it enables the optimal pavement future design for necessary requirements and constraints as well as efficient use, maintenance, and timely repairs of the public roads, directly contributing to the overall safety of our transportation system.

Indoor Detection and Tracking of People Using mmWave Sensor

Abstract: This paper proposes a new indoor people detection and tracking system using a millimeter-wave (mmWave) radar sensor. Firstly, a systematic approach for people detection and tracking is presented—a static clutter removal algorithm used for removing mmWave radar data’s static points. Two efficient clustering algorithms are used to cluster and identify people in a scene. The recursive Kalman filter tracking algorithm with data association is used to track multiple people simultaneously. Secondly, a fast indoor people detection and tracking system is designed based on our proposed algorithms. The method is lightweight enough for scalability and portability, and we can execute it in real time on a Raspberry Pi 4. Finally, the proposed method is validated by comparing it with the Texas Instruments (TI) system. The proposed system’s experimental accuracy ranged from 98% (calculated by misclassification errors) for one person to 65% for five people. The average position errors at positions 1, 2, and 3 are 0.2992 meters, 0.3271 meters, and 0.3171 meters, respectively. In comparison, the Texas Instruments system had an experimental accuracy ranging from 96% for one person to 45% for five people. The average position errors at positions 1, 2, and 3 are 0.3283 meters, 0.3116 meters, and 0.3343 meters, respectively. The proposed method’s advantage is demonstrated in terms of tracking accuracy, computation time, and scalability.

Electrochemical Sensor Based on Nanodiamonds and Manioc Starch for Detection of Tetracycline

Abstract: The use of nanostructured materials is already well-known as a powerful tool in the development of electrochemical sensors. Among several immobilization strategies of nanomaterials in the development of electrochemical sensors, the use of low-cost and environmentally friendly polymeric materials is highlighted. In this context, a new nanostructured biocomposite electrode is proposed as an electrochemical sensor for the analysis and determination of tetracycline. The composite electrode consists of a modified glassy carbon electrode (GCE) with a nanodiamond-based (ND) and manioc starch biofilm (MS), called ND-MS/GCE. The proposed sensor showed better electrochemical performance in the presence of tetracycline in comparison to the unmodified electrode, which was attributed to the increase in the electroactive surface area due to the presence of nanodiamonds. A linear dynamic range from to mol L−1 and a limit of detection of mol L−1 were obtained for the proposed sensor. ND-MS/GCE exhibited high repeatability and reproducibility for successive measurements with a relative standard deviation (RSD) of 6.3% and 1.5%, respectively. The proposed electrode was successfully applied for the detection of tetracycline in different kinds of water samples, presenting recoveries ranging from 86 to 112%.

A Convolutional Neural Network-Based Classification and Decision-Making Model for Visible Defect Identification of High-Speed Train Images

Abstract: In high-speed train safety inspection, two changed images which are derived from corresponding parts of the same train and photographed at different times are needed to identify whether they are defects. The critical challenge of this change classification task is how to make a correct decision by using bitemporal images. In this paper, two convolutional neural networks are presented to perform this task. Distinct from traditional classification tasks which simply group each image into different categories, the two presented networks are capable of inherently detecting differences between two images and further identifying changes by using a pair of images. In doing so, even in the case that abnormal samples of specific components are unavailable in training, our networks remain capable to make inference as to whether they become abnormal using change information. This proposed method can be used for recognition or verification applications where decisions cannot be made with only one image (state). Equipped with deep learning, this method can address many challenging tasks of high-speed train safety inspection, in which conventional methods cannot work well. To further improve performance, a novel multishape training method is introduced. Extensive experiments demonstrate that the proposed methods perform well.

Static Force Measurement Using Piezoelectric Sensors

Abstract: In force measurement applications, a piezoelectric force sensor is one of the most popular sensors due to its advantages of low cost, linear response, and high sensitivity. Piezoelectric sensors effectively convert dynamic forces to electrical signals by the direct piezoelectric effect, but their use has been limited in measuring static forces due to the easily neutralized surface charge. To overcome this shortcoming, several static (either pure static or quasistatic) force sensing techniques using piezoelectric materials have been developed utilizing several unique parameters rather than just the surface charge produced by an applied force. The parameters for static force measurement include the resonance frequency, electrical impedance, decay time constant, and capacitance. In this review, we discuss the detailed mechanism of these piezoelectric-type, static force sensing methods that use more than the direct piezoelectric effect. We also highlight the challenges and potentials of each method for static force sensing applications.

Hierarchical Routing Protocols for Wireless Sensor Networks: Functional and Performance Analysis

Abstract: Wireless sensor network (WSN) comprises of spatially distributed autonomous sensors to monitor physical or environmental conditions, such as temperature, sound, and pressure, and to cooperatively forward the collected information to the destination through the network infrastructure. As sensor nodes are energy constraint devices, therefore, the importance of energy efficient routing protocols has been increased. In order to minimize energy consumption, recently, a number of hierarchical routing protocols are proposed. For instance, LEACH is an elementary hierarchical routing protocol that employs clustering technique to achieve energy efficiency. A lot of research work has been performed to remove shortcomings and to improve the performance of hierarchical routing protocols. Therefore, a comprehensive review is required which can review state-of-the-art technologies, analyze functional and performance aspects, and highlight hierarchical routing protocol issues and challenges in WSNs. This paper proposes a taxonomy for the classification of existing hierarchical routing protocols for WSNs and analyzes the functionality and performance of existing hierarchical routing protocols. Moreover, it compares existing routing protocols to highlight key technological differences and provides performance comparison for the selected LEACH based routing protocols. Finally, the paper spotlights issues and challenges in existing routing protocols of WSNs, which can assist in future research for the selection of appropriate research domain and provide guidance in selection of energy efficient techniques in the design of energy efficient of routing protocols for WSNs.

Bearing Fault Diagnosis Based on Multiscale Convolutional Neural Network Using Data Augmentation

Abstract: Bearings are one of the most important parts of a rotating machine. Bearing failure can lead to mechanical failure, financial loss, and even personal injury. In recent years, various deep learning techniques have been used to diagnose bearing faults in rotating machines. However, deep learning technology has a data imbalance problem because it requires huge amounts of data. To solve this problem, we used data augmentation techniques. In addition, Convolutional Neural Network, one of the deep learning models, is a method capable of performing feature learning without prior knowledge. However, since conventional fault diagnosis based on CNN can only extract single-scale features, not only useful information may be lost but also domain shift problems may occur. In this paper, we proposed a Multiscale Convolutional Neural Network (MSCNN) to extract more powerful and differentiated features from raw signals. MSCNN can learn more powerful feature expression than conventional CNN through multiscale convolution operation and reduce the number of parameters and training time. The proposed model proved better results and validated the effectiveness of the model compared to 2D-CNN and 1D-CNN.

Face Detection and Recognition Algorithm in Digital Image Based on Computer Vision Sensor

Abstract: With the continuous innovation of network technology, various kinds of convenient network technologies have grown, and human dependence on network technology has gradually increased, which has resulted in the importance of network information security issues. With the continuous development of my country’s industrialization, the application of sensors is becoming more and more extensive, for example, the security vulnerabilities and defects in the operating system itself. Traditional sensors can “perceive” a certain thing or signal, convert it into an electrical signal and record it, and then use a conversion circuit to output the electrical signal into a value or other display form that is conducive to observation. Nowadays, sensors have been further developed. Based on the original “perception” function, combined with computer technology, it integrates data storage, data processing, data communication, and other functions, so that it has analysis functions and can better display information. The technical level has reached a new level. Early intelligent recognition mainly used the uniqueness of finger and palm lines to scan and contrast, but due to some weather reasons or skin texture constraints caused by skin texture, these methods showed certain limitations. This paper proposes a new computer vision-based algorithm from face detection technology and face recognition technology. In the face detection technology, it is mainly introduced from the OpenCV method. Face recognition technology is improved in practical applications through the Seetaface method and YouTu method. At the same time, using the contrast experiment, the detection and recognition rates under the three different requirements of side face detection, occlusion detection, and facial exaggerated expression are compared, and the accuracy of each method is improved. The results show that each case is compared in each case. The advantages and disadvantages of the algorithm effectively verify the effectiveness of the method.

Real-Time Object Detection for LiDAR Based on LS-R-YOLOv4 Neural Network

Abstract: Recently, self-driving cars became a big challenge in the automobile industry After the DARPA challenge, which introduced the design of a self-driving system that can be classified as SAR Level 3 or higher levels, driven to focus on self-driving cars more Later on, using these introduced design models, a lot of companies started to design self-driving cars Various sensors, such as radar, high-resolution cameras, and LiDAR are important in self-driving cars to sense the surroundings LiDAR acts as an eye of a self-driving vehicle, by offering 64 scanning channels, 269° vertical field view, and a high-precision 360° horizontal field view in real-time The LiDAR sensor can provide 360° environmental depth information with a detection range of up to 120 meters In addition, the left and right cameras can further assist in obtaining front image information In this way, the surrounding environment model of the self-driving car can be accurately obtained, which is convenient for the self-driving algorithm to perform route planning It is very important for self-driving to avoid the collision LiDAR provides both horizontal and vertical field views and helps in avoiding collision In an online website, the dataset provides different kinds of data like point cloud data and color images which helps this data to use for object recognition In this paper, we used two types of publicly available datasets, namely, KITTI and PASCAL VOC Firstly, the KITTI dataset provides in-depth data knowledge for the LiDAR segmentation (LS) of objects obtained through LiDAR point clouds The performance of object segmentation through LiDAR cloud points is used to find the region of interest (ROI) on images And later on, we trained the network with the PASCAL VOC dataset used for object detection by the YOLOv4 neural network To evaluate, we used the region of interest image as input to YOLOv4 By using all these technologies, we can segment and detect objects Our algorithm ultimately constructs a LiDAR point cloud at the same time; it also detects the image in real-time

Analysis of Land Use Changes and Driving Forces in the Yanhe River Basin from 1980 to 2015

Abstract: Based on the support of RS and GIS technology, this paper analyzes the spatial and temporal variation characteristics and driving forces of land use in the Yanhe River Basin through the processing and interpretation of remote sensing images in different periods from 1980 to 2015 and the methods of the land use transfer matrix and dynamic attitude. The results show that cropland, grassland, and forest land are the three types of land use with the most obvious changes, while urban land and water body have relatively small changes in the Yanhe River Basin. The transfer between cropland, forest land, and grassland and urban land is very obvious, among which the conversion rate of cropland is the highest. During the 15 years from 2000 to 2015, the land use types of the Yanhe River Basin changed by 13.17%, with an average annual growth rate of 0.88%. The implementation of ecological restoration and governance policy is the direct driving force of land use change in the Yanhe River Basin. The results obtained in this study can provide reference basis for land use planning and management and land use structure optimization in the Yanhe River Basin in the future.

Fuzzy Control-Based Energy-Aware Routing Protocol for Wireless Body Area Networks

Abstract: Advances in medical and communication technologies have empowered the development of Wireless Body Area Networks (WBANs). WBANs interconnect with miniature sensors placed on the human body to enable medical monitoring of patient health. However, the limited battery capacity, delay, and reliability of data transmission have brought challenges to the wider application of WBAN. Minimum consumption of energy and maximum satisfaction with the QoS requirements are essential design aims of the WBAN schemes. Therefore, a fuzzy control-based energy-aware routing protocol (EARP) is proposed in this paper, the proposed protocol establishes a fuzzy control model composed of remaining node energy and link quality, and the best forwarder node is determined by the processes of fuzzification, fuzzy inference, and defuzzification. The simulation results showed that compared with the performance of the existing EERDT and M-TSIMPLE protocols, the proposed EARP has better performance, including extending network lifetime and improving the reliability of data transmission.

Nondestructive Testing for Corrosion Evaluation of Metal under Coating

Abstract: The use of steel has grown rapidly over the past decades. However, corrosion under coating detection still presents challenges for nondestructive testing (NDT) techniques. One of such challenges is the lift-off introduced by complex structures. Inaccessibility due to structure leads corrosion to be undetected, which can lead to catastrophic failure. Furthermore, lift-off effects reduce the sensitivities. The limitations of existing NDT techniques heighten the need for novel approaches to the characterization of corrosion. This paper begins with a discussion of the challenges associated with corrosion detection of metal under coating. Secondly, reviews are given of the most NDT methods used for the detection of corrosion under coating. The different techniques based on nondestructive testing methods such as ultrasonic, acoustic, electromagnetic, radiographic, and thermographic have been detailed out. This review presents the significance and advantages provided by the emerging NDT techniques. In the end, the trends and identified problems are summarized.

A Novel Deep Convolutional Neural Network Based on ResNet-18 and Transfer Learning for Detection of Wood Knot Defects

Abstract: Wood defects are quickly identified from an optical image based on deep learning methodology, which effectively improves wood utilization. Traditional neural network techniques have not yet been employed for wood defect detection due to long training time, low recognition accuracy, and nonautomatical extraction of defect image features. In this work, a model (so-called ReSENet-18) for wood knot defect detection that combined deep learning and transfer learning is proposed. The “squeeze-and-excitation” (SE) module is firstly embedded into the “residual basic block” structure for a “SE-Basic-Block” module construction. This model has the advantages of the features that are extracted in the channel dimension, and it is fused in multiscale with original features. Instantaneously, the fully connected layer is replaced with a global average pooling; consequently, the model parameters could be reduced effectively. The experimental results show that the accuracy has reached 99.02%, meanwhile the training time is also reduced. It shows that the proposed deep convolutional neural network based on ReSENet-18 combined with transfer learning can improve the accuracy of defect recognition and has a potential application in the detection of wood knot defects.

Weak-Light Image Enhancement Method Based on Adaptive Local Gamma Transform and Color Compensation

Abstract: In weak-light environments, images suffer from low contrast and the loss of details. Traditional image enhancement models are usually failure to avoid the issue of overenhancement. In this paper, a simple and novel correction method is proposed based on an adaptive local gamma transformation and color compensation, which is inspired by the illumination reflection model. Our proposed method converts the source image into YUV color space, and the component is estimated with a fast guided filter. The local gamma transform function is used to improve the brightness of the image by adaptively adjusting the parameters. Finally, the dynamic range of the image is optimized by a color compensation mechanism and a linear stretching strategy. By comparing with the state-of-the-art algorithms, it is demonstrated that the proposed method adaptively reduces the influence of uneven illumination to avoid overenhancement and improve the visual effect of low-light images.

End-to-End Ship Detection in SAR Images for Complex Scenes Based on Deep CNNs

Abstract: Ship detection on synthetic aperture radar (SAR) imagery has many valuable applications for both civil and military fields and has received extraordinary attention in recent years. The traditional detection methods are insensitive to multiscale ships and usually time-consuming, results in low detection accuracy and limitation for real-time processing. To balance the accuracy and speed, an end-to-end ship detection method for complex inshore and offshore scenes based on deep convolutional neural networks (CNNs) is proposed in this paper. First, the SAR images are divided into different grids, and the anchor boxes are predefined based on the responsible grids for dense ship prediction. Then, Darknet-53 with residual units is adopted as a backbone to extract features, and a top-down pyramid structure is added for multiscale feature fusion with concatenation. By this means, abundant hierarchical features containing both spatial and semantic information are extracted. Meanwhile, the strategies such as soft non-maximum suppression (Soft-NMS), mix-up and mosaic data augmentation, multiscale training, and hybrid optimization are used for performance enhancement. Besides, the model is trained from scratch to avoid learning objective bias of pretraining. The proposed one-stage method adopts end-to-end inference by a single network, so the detection speed can be guaranteed due to the concise paradigm. Extensive experiments are performed on the public SAR ship detection dataset (SSDD), and the results show that the method can detect both inshore and offshore ships with higher accuracy than other mainstream methods, yielding the accuracy with an average of 95.52%, and the detection speed is quite fast with about 72 frames per second (FPS). The actual Sentinel-1 and Gaofen-3 data are utilized for verification, and the detection results also show the effectiveness and robustness of the method.

Modified Sage-Husa Adaptive Kalman Filter-Based SINS/DVL Integrated Navigation System for AUV

Abstract: This paper presents a modified Sage-Husa adaptive Kalman filter-based SINS/DVL integrated navigation system for the autonomous underwater vehicle (AUV), where DVL is employed to correct the navigation errors of SINS that accumulate over time. When negative definite items are large enough, different from the positive definiteness of noise matrices which cannot be guaranteed for the conventional Sage-Husa adaptive Kalman filter, the proposed modified Sage-Husa adaptive Kalman filter deletes the negative definite items of adaptive update laws of the noise matrix to ensure the convergence of the Sage-Husa adaptive Kalman filter. In other words, this method sacrifices some filtering precision to ensure the stability of the filter. The simulation tests are implemented to verify that expected navigation accuracy for AUV can be obtained using the proposed modified Sage-Husa adaptive Kalman filter.

A Novel Neural Network Model for Traffic Sign Detection and Recognition under Extreme Conditions

Abstract: Traffic sign detection is extremely important in autonomous driving and transportation safety systems. However, the accurate detection of traffic signs remains challenging, especially under extreme conditions. This paper proposes a novel model called Traffic Sign Yolo (TS-Yolo) based on the convolutional neural network to improve the detection and recognition accuracy of traffic signs, especially under low visibility and extremely restricted vision conditions. A copy-and-paste data augmentation method was used to build a large number of new samples based on existing traffic-sign datasets. Based on You Only Look Once (YoloV5), the mixed depth-wise convolution (MixConv) was employed to mix different kernel sizes in a single convolution operation, so that different patterns with various resolutions can be captured. Furthermore, the attentional feature fusion (AFF) module was integrated to fuse the features based on attention from same-layer to cross-layer scenarios, including short and long skip connections, and even performing the initial fusion with itself. The experimental results demonstrated that, using the YoloV5 dataset with augmentation, the precision was 71.92, which was increased by 34.56 compared with the data without augmentation, and the mean average precision mAP_0.5 was 80.05, which was increased by 33.11 compared with the data without augmentation. When MixConv and AFF were applied to the TS-Yolo model, the precision was 74.53 and 2.61 higher than that with data augmentation only, and the value of mAP_0.5 was 83.73 and 3.68 higher than that based on the YoloV5 dataset with augmentation only. Overall, the performance of the proposed method was competitive with the latest traffic sign detection approaches.

DWCA-YOLOv5: An Improve Single Shot Detector for Safety Helmet Detection

Abstract: Aiming at solving the problem that the detection methods used in the existing helmet detection research has low detection efficiency and the cumulative error influences accuracy, a new algorithm for improving YOLOv5 helmet wearing detection is proposed. First of all, we use the - means++ algorithm to improve the size matching degree of the a priori anchor box; secondly, integrate the Depthwise Coordinate Attention (DWCA) mechanism in the backbone network, so that the network can learn the weight of each channel independently and enhance the information dissemination between features, thereby strengthening the network’s ability to distinguish foreground and background. The experimental results show as follows: in the self-made safety helmet wearing detection dataset, the average accuracy rate reached 95.9%, the average accuracy of the helmet detection reached 96.5%, and the average accuracy of the worker’s head detection reached 95.2%. Making a comparison with the YOLOv5 algorithm, our model has a 3% increase in the average accuracy of helmet detection, which is in line with the accuracy requirements of helmet wearing detection in complex construction scenarios.

Precise Blood Glucose Sensing by Nitrogen-Doped Graphene Quantum Dots for Tight Control of Diabetes

Abstract: Graphene quantum dots (GQD) are novel fluorescent carbon nanomaterials based on a graphite structure. Thanks to extraordinary properties such as high surface area and enhanced prevalent optical properties, they have received more interest for special applications. Glucose sensing is a critical factor for the diagnosis, and treatment of diabetes plays an important role and could contribute to the monitoring of diabetes and other related parameters, which has been effectively underscoring the health society. Detecting glucose has been cultivated through different systems, for example, electrochemical or optical techniques. Novel transducers made with GQD that fluorescent coordinate methods have considered the improvement of cutting-edge glucose sensors with prevalent affectability and accommodation. Currently, detection of glucose by nitrogen-doped GQD frameworks concerning the determined objectives has been considerably considered. Here, we explored the properties of fluorescent nitrogen-doped GQD as an excellent and effective index that significantly could promote nitrogen-doped GQDs and make them an appropriate candidate for detecting glucose.

Unmanned Aircraft System- (UAS-) Based High-Throughput Phenotyping (HTP) for Tomato Yield Estimation

Abstract: Yield prediction and variety selection are critical components for assessing production and performance in breeding programs and precision agriculture Since plants integrate their genetics, surrounding environments, and management conditions, crop phenotypes have been measured over cropping seasons to represent the traits of varieties These days, UAS (unmanned aircraft system) provides a new opportunity to collect high-quality images and generate reliable phenotypic data efficiently Here, we propose high-throughput phenotyping (HTP) from multitemporal UAS images for tomato yield estimation UAS-based RGB and multispectral images were collected weekly and biweekly, respectively The shape of the features of tomatoes such as canopy cover, canopy, volume, and vegetation indices derived from UAS imagery was estimated throughout the entire season To extract time-series features from UAS-based phenotypic data, crop growth and growth rate curves were fitted using mathematical curves and first derivative equations Time-series features such as the maximum growth rate, day at a specific event, and duration were extracted from the fitted curves of different phenotypes The linear regression model produced high values even with different variable selection methods: all variables (079), forward selection (07), and backward selection (077) With factor analysis, we figured out two significant factors, growth speed and timing, related to high-yield varieties Then, five time-series phenotypes were selected for yield prediction models explaining 65 percent of the variance in the actual harvest The phenotypic features derived from RGB images played more important roles in prediction yield This research also demonstrates that it is possible to select lower-performing tomato varieties successfully The results from this work may be useful in breeding programs and research farms for selecting high-yielding and disease-/pest-resistant varieties

Environment Perception Technologies for Power Transmission Line Inspection Robots

Abstract: With the fast development of the power system, traditional manual inspection methods of a power transmission line (PTL) cannot supply the demand for high quality and dependability for power grid maintenance. Consequently, the automatic PTL inspection technology becomes one of the key research focuses. For the purpose of summarizing related studies on environment perception and control technologies of PTL inspection, technologies of three-dimensional (3D) reconstruction, object detection, and visual servo of PTL inspection are reviewed, respectively. Firstly, 3D reconstruction of PTL inspection is reviewed and analyzed, especially for the technology of LiDAR-based reconstruction of power lines. Secondly, the technology of typical object detection, including pylons, insulators, and power line accessories, is classified as traditional and deep learning-based methods. After that, their merits and demerits are considered. Thirdly, the progress and issues of visual servo control of inspection robots are also concisely addressed. For improving the automation degree of PTL robots, current problems of key techniques, such as multisensor fusion and the establishment of datasets, are discussed and the prospect of inspection robots is presented.

Research on Surface Deformation of Ordos Coal Mining Area by Integrating Multitemporal D-InSAR and Offset Tracking Technology

Abstract: Underground mining in coal mining areas will induce large-scale, large-gradient surface deformation, threatening the safety of people’s lives and property in nearby areas. Due to mining-related subsidence is characterized by fast displacement and high nonlinearity, monitoring this process by using traditional and single interferometric synthetic aperture radar (InSAR) technology is very challenging, and it cannot accurately and quantitatively calculate the deformation of the mining area. In this paper, we proposed a new method that combines both multitemporal consecutive D-InSAR and offset tracking technology to construct a complete deformation field of the coal mining area. Taking into account the accuracy of multitemporal consecutive D-InSAR in calculating small deformation areas and the ability of offset tracking to measure large deformation areas, we utilized their respective advantages to extract the surface influence range and applied an adaptive spatial filtering method to integrate their respective results for inversion of the deformation field. 12 ascending high-resolution TerraSAR-X images (2 m) from September 3, 2018, to October 26, 2019, and 39 descending Sentinel-1 TOPS SAR images from August 5, 2018, to November 4, 2019, in the Ordos Coalfield located at Inner Mongolia, China, were utilized to obtain the whole subsidence field of the working faces F6211 and F6207 during the 454-day mining period. The GPS monitoring station located in the direction of the mining surface is used to verify the accuracy of the above method; at the same time, to a certain extent, the difference between the unmanned aerial vehicle’s DSM data acquired after coal mining and the Shuttle Radar Topography Mission (STRM) DEM can qualitatively verify the accuracy of the results. Our results show that the results of TerraSAR are basically consistent with the deformation trend of GPS data, and that of Sentinel-1 have large errors compared with GPS. The maximum central subsidence reaches ~12 m in the working face F6211 and ~4 m in the working face F6207. In the working face F6207, the good agreement between GPS and TerraSAR results indicated that the method above using high-resolution SAR data could be reliable for monitoring the large deformation area in the mining field.

BLNN: Multiscale Feature Fusion-Based Bilinear Fine-Grained Convolutional Neural Network for Image Classification of Wood Knot Defects

Abstract: Wood defects are quickly identified from an optical image based on deep learning methodology, which effectively improves the wood utilization. The traditional neural network technique is unemployed for the wood defect detection of optical image used, which results from a long training time, low recognition accuracy, and nonautomatic extraction of defect image features. In this paper, a wood knot defect detection model (so-called BLNN) combined deep learning is reported. Two subnetworks composed of convolutional neural networks are trained by Pytorch. By using the feature extraction capabilities of the two subnetworks and combining the bilinear join operation, the fine-grained features of the image are obtained. The experimental results show that the accuracy has reached up 99.20%, and the training time is obviously reduced with the speed of defect detection about 0.0795 s/image. It indicates that BLNN has the ability to improve the accuracy of defect recognition and has a potential application in the detection of wood knot defects.

Vision-Based System for Assisting Blind People to Wander Unknown Environments in a Safe Way

Abstract: Vision is the principal source of information of the surrounding world. It facilitates our movement and development of everyday activities. In this sense, blind people have great difficulty for moving, especially in unknown environments, which reduces their autonomy and puts them at risk of suffering an accident. Electronic Travel Aids (ETAs) have emerged and provided outstanding navigation assistance for blind people. In this work, we present the methodology followed for implementing a stereo vision-based system that assists blind people to wander unknown environments in a safe way, by sensing the world, segmenting the floor in 3D, fusing local 2D grids considering the camera tracking, creating a global occupancy 2D grid, reacting to close obstacles, and generating vibration patterns with an haptic belt. For segmenting the floor in 3D, we evaluate normal vectors and orientation of the camera obtained from depth and inertial data, respectively. Next, we apply RANSAC for computing efficiently the equation of the supporting plane (floor). The local grids are fused, obtaining a global map with data of free and occupied areas along the whole trajectory. For parallel processing of dense data, we leverage the capacity of the Jetson TX2, achieving high performance, low power consumption, and portability. Finally, we present experimental results obtained with ten (10) participants, in different conditions, with obstacles of different height, hanging obstacles, and dynamic obstacles. These results show high performance and acceptance by the participants, highlighting the easiness to follow instructions and the short period of training.

Traffic and Energy Aware Optimization for Congestion Control in Next Generation Wireless Sensor Networks

Abstract: Congestion in wireless sensor networks (WSNs) is an unavoidable issue in today’s scenario, where data traffic increased to its aggregated capacity of the channel. The consequence of this turns in to overflowing of the buffer at each receiving sensor nodes which ultimately drops the packets, reduces the packet delivery ratio, and degrades throughput of the network, since retransmission of every unacknowledged packet is not an optimized solution in terms of energy for resource-restricted sensor nodes. Routing is one of the most preferred approaches for minimizing the energy consumption of nodes and enhancing the throughput in WSNs, since the routing problem has been proved to be an NP-hard and it has been realized that a heuristic-based approach provides better performance than their traditional counterparts. To tackle all the mentioned issues, this paper proposes an efficient congestion avoidance approach using Huffman coding algorithm and ant colony optimization (ECA-HA) to improve the network performance. This approach is a combination of traffic-oriented and resource-oriented optimization. Specially, ant colony optimization has been employed to find multiple congestion-free alternate paths. The forward ant constructs multiple congestion-free paths from source to sink node, and backward ant ensures about the successful creation of paths moving from sink to source node, considering energy of the link, packet loss rate, and congestion level. Huffman coding considers the packet loss rate on different alternate paths discovered by ant colony optimization for selection of an optimal path. Finally, the simulation result presents that the proposed approach outperforms the state of the art approaches in terms of average energy consumption, delay, and throughput and packet delivery ratio.