Position Of Box Research Articles

An improved YOLO v3 algorithm for remote Sensing image target detection

With more and more in-depth research on deep learning algorithms in recent years, how to use deep learning method to detect remote sensing images is the key to improving the utilization efficiency of remote sensing data and realizing the transformation from data to knowledge. In this paper, an improved YOLO V3 algorithm is proposed to solve the problems of missed detection and false detection of the original YOLOv3 algorithm in remote sensing image target detection with different size and wide disparity in length and width ratio. first of all, K-means algorithm is used for clustering analysis of data set to obtain the position of anchor box; Secondly, the dilated convolution with expansion rate of 2 is used to replace the general convolution in the feature extraction part; Then four scales are used for prediction; Finally, the improved algorithm is applied to the recognition of bridges, harbors and airports. The results show that the detection performance of the algorithm is improved by about 2% compared with the original algorithm.

Stereo CenterNet-based 3D object detection for autonomous driving

Recently, three-dimensional (3D) detection based on stereo images has progressed remarkably; however, most advanced methods adopt anchor-based two-dimensional (2D) detection or depth estimation to address this problem. Nevertheless, high computational cost inhibits these methods from achieving real-time performance. In this study, we propose a 3D object detection method, Stereo CenterNet (SC), using geometric information in stereo imagery. SC predicts the four semantic key points of the 3D bounding box of the object in space and utilizes 2D left and right boxes, 3D dimension, orientation, and key points to restore the bounding box of the object in the 3D space. Subsequently, we adopt an improved photometric alignment module to further optimize the position of the 3D bounding box. Experiments conducted on the KITTI dataset indicate that the proposed SC exhibits the best speed-accuracy trade-off among advanced methods without using extra data.

InsulatorGAN: A Transmission Line Insulator Detection Model Using Multi-Granularity Conditional Generative Adversarial Nets for UAV Inspection

Insulator detection is one of the most significant issues in high-voltage transmission line inspection using unmanned aerial vehicles (UAVs) and has attracted attention from researchers all over the world. The state-of-the-art models in object detection perform well in insulator detection, but the precision is limited by the scale of the dataset and parameters. Recently, the Generative Adversarial Network (GAN) was found to offer excellent image generation. Therefore, we propose a novel model called InsulatorGAN based on using conditional GANs to detect insulators in transmission lines. However, due to the fixed categories in datasets such as ImageNet and Pascal VOC, the generated insulator images are of a low resolution and are not sufficiently realistic. To solve these problems, we established an insulator dataset called InsuGenSet for model training. InsulatorGAN can generate high-resolution, realistic-looking insulator-detection images that can be used for data expansion. Moreover, InsulatorGAN can be easily adapted to other power equipment inspection tasks and scenarios using one generator and multiple discriminators. To give the generated images richer details, we also introduced a penalty mechanism based on a Monte Carlo search in InsulatorGAN. In addition, we proposed a multi-scale discriminator structure based on a multi-task learning mechanism to improve the quality of the generated images. Finally, experiments on the InsuGenSet and CPLID datasets demonstrated that our model outperforms existing state-of-the-art models by advancing both the resolution and quality of the generated images as well as the position of the detection box in the images.

High-Precision Underwater Acoustical Localization of the Black Box Based on an Improved TDOA Algorithm

Underwater acoustic localization (UWAL) of the black box is a significant technique for transportation safety. The traditional UWAL algorithm based on the time difference of arrival (TDOA) suffers from large localization error due to the signal period drift in the scenario of high pressure and low temperature under sea. To solve this problem, an improved TDOA algorithm is proposed in this letter. We define the second-order time difference of arrival, which eliminates the drifted signal period. Utilizing the second-order time difference of arrival information, we establish the equations to solve the position of the black box. Precision analysis is presented to theoretically calculate the localization precision of the proposed algorithm and test the robustness with respect to the input errors. Experiments in different scenes validate the effectiveness of the algorithm.

Candidate box fusion based approach to adjust position of the candidate box for object detection

The method of object detection has been applied to all aspects in our lives. Although object detection methods based on deep learning have been widely used in various fields, there are still some overlooked problems in the candidate box selection stage. The detection results of traditional candidate box selection methods can only select a relatively optimal maximum candidate box. If the maximum candidate box is still not accurate enough, this type of methods will not be able to do adjust it. To solve this problem, an object detection method based on the multiple candidate box fusion is proposed. The method can not only retain the maximum candidate box and delete the non-maximum candidate box, but also adjust the position of the maximum candidate box again. Thereby a more accurate maximum candidate box can be obtained. In order to verify the generalization ability of the method, the candidate box fusion method is combined with the two object detection frameworks: faster R-CNN model and YOLOv3 model. The results of these experiments prove that the proposed method can achieve higher detection accuracy and complete the object detection task more effectively.

Study on the position of load box for pile foundation of Qilu Yellow River Approach Bridge

In theory, the balance point position of O-cell pile test is determined and unique, and the load box should be located at this point. But in practical projects, it is difficult to determine the balance point position accurately. The pile foundation of Qilu Yellow River Approach Bridge bears a large load. In order to reduce the risk caused by the position of load box in O-cell pile test, the double load box technology is adopted in this pile test project. According to the pile design, the soil properties and the loading sequence, the bearing capacity of the pile foundation before and after grouting is estimated. The pile test results show that the project is a successful example for the double load box technology.

The Role of Gender and Familiarity in a Modified Version of the Almeria Boxes Room Spatial Task.

Individual factors like gender and familiarity can affect the kind of environmental representation that a person acquires during spatial navigation. Men seem to prefer relying on map-like survey representations, while women prefer using sequential route representations. Moreover, a good familiarity with the environment allows more complete environmental representations. This study was aimed at investigating gender differences in two different object-position learning tasks (i.e., Almeria Boxes Tasks) assuming a route or a survey perspective also considering the role of environmental familiarity. Two groups of participants had to learn the position of boxes placed in a virtual room. Participants had several trials, so that familiarity with the environment could increase. In both tasks, the effects of gender and familiarity were found, and only in the route perspective did an interaction effect emerge. This suggests that gender differences can be found regardless of the perspective taken, with men outperforming women in navigational tasks. However, in the route task, gender differences appeared only at the initial phase of learning, when the environment was unexplored, and disappeared when familiarity with the environment increased. This is consistent with studies showing that familiarity can mitigate gender differences in spatial tasks, especially in more complex ones.

A Comparison of Single-Entry and Multiple-Entry Casing Impingement Manifolds for Active Thermal Tip Clearance Control

In this paper, we compare using computational fluid dynamics the aero-thermal performance of two candidate casing manifolds for supplying an impingement-actuated active tip clearance control system for an aero-engine high-pressure turbine. The two geometries are (a) single-entry: an annular manifold fed at one circumferential location; (b) multiple-entry: a casing manifold split into four annular sectors, with each sector supplied separately from an annular ring main. Both the single-entry and multiple-entry systems analysed in this paper are idealised versions of active clearance control systems in current production engines. Aero-thermal performance is quantitatively assessed on the basis of the heat transfer coefficient distribution, driving temperature difference for heat transfer between the jet and casing wall and total pressure loss within the high-pressure turbine active clearance control system. We predict that the mean heat transfer coefficient (defined with respect to the inlet temperature and local wall temperature) of the single-entry active clearance control system is 77% greater than the multiple-entry system, primarily because the coolant in the multiple-entry case picks up approximately 40 K of temperature from the ring main walls, and secondarily because the average jet Reynolds number of impingement holes in the single-entry system is 1.2 times greater than in the multiple-entry system. The multiple-entry system exhibits many local hot and cold spots, depending on the position of the transfer boxes, while the single-entry case has a more predictable aero-thermal field across the system. The multiple-entry feed system uses an average of 20% of the total available pressure drop, while the feed system for the single-entry geometry uses only 2% of the total available pressure drop. From the aero-thermal results of this computational study, and in consideration of holistic aero-engine design factors, we conclude that a single-entry system is closer to an optimal solution than a multiple-entry system.

A novel DNN tracking algorithm for structural system identification

In the field of structural health monitoring (SHM), cameras record videos and tracking methods can be applied to calculate the structural displacement. Commercial and unmanned aerial vehicle (UAV) cameras are promising non-contact sensors owning to their high availability and easy installation. However, effective tracking methods need to be developed. In this study, we firstly propose an end-to-end vision measuring framework with a novel deep neural network (DNN) tracker, named Siamese Single Decoder Network (SiamSDN). The system requires no target installation and uses cellphone cameras. For SiamSDN, the position and scale of bounding box are formulated through statistical parameter estimation. Unlike generative trackers, SiamSDN does not require manually extracted features or pre-defined motion areas. The tracking object is solely identified in the first frame. A shaking table test of a five-storey structure is carried out to demonstrate the efficiency. Besides, a UAV is used to simulate the field test. To minimize the error caused by the vibrations of UAV, digital video stabilization (DVS) is proposed to eliminate the drifts. Videos taken by both the commercial and UAV cameras are analyzed to calculate the displacements. Comparing our DNN tracker with feature point matching approach, SiamSDN improves the displacement measuring accuracy by 66.16% and 57.54%, respectively, and the frequency characteristics are obtained precisely.

Bacterial contamination of nonsterile gloves versus hands after hand hygiene

BackgroundNonsterile gloves (NSG) are often overused, while the emphasis should lie in hand hygiene (HH). Furthermore, improper HH leads to contamination of NSG in glove boxes. The aim of this study was to compare microbial loads on hands from health-care workers (HCW) after HH to NSG and to study the influence of position and filling level of glove boxes on contamination rates. MethodsFingerprints on agar plates were made from randomly chosen HCWs directly after HH. Plates were incubated and colony-forming units counted. NSG taken from glove boxes were also sampled. Filling level and position (horizontal vs vertical) of the glove boxes were recorded. ResultsMedian colony-forming units count was similar for hands after HH (N = 107, median 1, IQR 5) and NSG (N = 185, median 1, IQR 2, P-value .33). Only few samples in both groups showed growth of pathogenic bacteria. Neither the filling level (P-value .76), nor the position of the glove box (P-value .68) had an influence on NSG contamination. ConclusionMicrobial loads of hands after HH are comparable to NSG. Filling level or position of the glove box did not influence glove contamination. Whether similar microbial counts translate into comparable nosocomial infection rates warrants further research.

High Resolution Information Reserved Anchor-Free Detection Algorithm

<p indent=0mm>Recently, object detection has gradually become a critical domain in the visual research community, while it’s also challenging to locate objects of different scales accurately for detectors. For the application of detecting objects in images, a high resolution anchor free (HOAR) detection strategy is proposed to meet the challenge of variable object scales. HOAR first inputs the image into a parallel high-resolution network with multiple paths (which is corresponding to different scales). Then the output feature map is extracted from each path as the deep feature representation of the image at each scale. Next, the dense feature pyramid (DenseFPN) is adopted to fuse the information of these feature maps to obtain the re-combined feature maps of multiple scales. Finally, the anchor-free detection sub-network is used to determine the object’s category and rectangle box position of each point on these feature maps. To verify the effectiveness of HOAR, comparative experiments are carried out on COCO dataset. First, the results on ablation study show the necessity of each module in the HOAR strategy. Second, the evaluation metric of detection, namely mAP, of HOAR reaches 40.5 on the validation set, which is significantly higher than mAP of all baseline models and some SOTA methods. In addition, the size of model parameters of HOAR strategy is significantly less than that of baseline models.

Lightweight Automatic Identification and Location Detection Model of Farmland Pests

Automatic identification and location of farmland pests are an important direction of target detection research. The wide variety of pests and the similarity between pest categories make the automatic identification of farmland pests have some problems, such as high error rate and difficult identification. In order to achieve a better target for automatic identification and location of farmland pests, this paper proposes a lightweight pest detection model, and the network is the EfficientNet proposed by Google, which achieves the detection of 26 pests, the idea based on the classical Yolo target detection algorithm. First of all, features were extracted through the lightweight backbone, and then multiscale feature fusion is performed by PANet; finally, three feature matrices with different sizes were output to predict pests of different sizes. Using CIOU as the loss function of regression prediction better reflects the relative position of the prior box and the real box. The experimental results are compared with other lightweight algorithms, and the results show that the accuracy rate of the algorithm for identification and localization of agricultural pest in this paper is the highest and could reach 93.73%. Moreover, the model is lightweight and can be deployed on low‐cost equipment, which reduces the cost of equipment and accurately predicts the status of pests and diseases in farmland. In practice, it is shown that the algorithm can effectively solve the problems of large number of pests, pest accumulation, background interference, and has strong robustness.

Shape Similarity Intersection-Over-Union Loss Hybrid Model for Detection of Synthetic Aperture Radar Small Ship Objects in Complex Scenes

With the continuous development and utilization of marine environments, the demand for accurate identification of ship targets at sea is increasing in both military and civilian fields. Synthetic aperture radar (SAR) is used to detect ship targets at sea and can provide 24-h detection under any weather conditions. Deep-learning models enable the effective detection of ship targets using SAR images; however, the recognition accuracy may be low or false positives may occur in complex scenarios wherein it is difficult to detect the ship targets. Current target-detection tasks include target classification and positioning through bounding-box regression. Herein, a regression loss function is derived to calculate the position of the bounding box, and intersection over union (IoU) is applied to estimate the positioning accuracy. As a result, a gap exists between the commonly used positioning losses for regressing the parameters of a bounding box and the optimization of these metric values. Therefore, the proposed hybrid model combines classification, localization, and segmentation with a novel multitask loss function for boundary-box localization based on the improved IoU. This solves the problem of inconsistency between training and evaluation and improves the positioning accuracy. Experiments were conducted using the SAR dataset for ship detection; the dataset was labeled by SAR experts and included multiscale ship chips with a resolution of 256 pixels in both range and azimuth. In summary, the experimental results indicate that the proposed hybrid model could improve the detection accuracy in complex scenarios, and its false-positive rate is significantly lower than those of the other models.

Study on Dust Turbulence-Chemical Agglomeration for Electrostatic Precipitation Technology

Due to the turbulent mixing between agglomeration agent and dust, the number of collisions per unit time between particles and droplets increases with the particle size in the process of chemical agglomeration, and the agglomeration efficiency increases accordingly. Turbulent agglomeration can promote the agglomeration effect of subsequent chemical agglomeration by strengthening the collision between particles and agglomerant droplets. The author used chemical agglomeration and turbulent mixing to cooperate in order to improve the efficiency of dust removal. Turbulent mixing can promote chemical agglomeration from agglomeration effect and dust removal efficiency, which can greatly improve electric dust removal technologyTurbulent mixing is the most intense at the inlet box position, and the agglomeration effect is the best. Turbulent mixing synergistic effect has an effect on dust removal efficiency. Compared with the three curves, it can be seen that the dust removal efficiency increases rapidly with the increase of agglomeration concentration, the curve trend changes obviously. The dust removal efficiency can reach 98.36 % and it is the highest in the middle section when the wind speed is greater than 11.2m/s. Through the experiment, the turbulent mixing and chemical coagulation method has a good application prospect in the electrostatic precipitators.

Automatic Microplot Localization Using UAV Images and a Hierarchical Image-Based Optimization Method.

To develop new crop varieties and monitor plant growth, health, and traits, automated analysis of aerial crop images is an attractive alternative to time-consuming manual inspection. To perform per-microplot phenotypic analysis, localizing and detecting individual microplots in an orthomosaic image of a field are major steps. Our algorithm uses an automatic initialization of the known field layout over the orthomosaic images in roughly the right position. Since the orthomosaic images are stitched from a large number of smaller images, there can be distortion causing microplot rows not to be entirely straight and the automatic initialization to not correctly position every microplot. To overcome this, we have developed a three-level hierarchical optimization method. First, the initial bounding box position is optimized using an objective function that maximizes the level of vegetation inside the area. Then, columns of microplots are repositioned, constrained by their expected spacing. Finally, the position of microplots is adjusted individually using an objective function that simultaneously maximizes the area of the microplot overlapping vegetation, minimizes spacing variance between microplots, and maximizes each microplot's alignment relative to other microplots in the same row and column. The orthomosaics used in this study were obtained from multiple dates of canola and wheat breeding trials. The algorithm was able to detect 99.7% of microplots for canola and 99% for wheat. The automatically segmented microplots were compared to ground truth segmentations, resulting in an average DSC of 91.2% and 89.6% across all microplots and orthomosaics in the canola and wheat datasets.

Starting a search: Exploring the differences in two digital libraries

AbstractWe conducted an eye‐tracking study in two digital libraries to examine how users orientate themselves on the digital library homepages, how fast and by which actions they initiate the search process. Results showed that participants typically started their task by querying. However, there were some differences in the elements they fixated on first, the way they started the search and the time they needed to orient on the page. These differences seem to have been caused by the homepage layout, especially by the search box position. This shows eye‐tracking can be used to examine first actions in digital libraries and could be useful for designing better interfaces.

Liquid Thermo-Responsive Smart Window Derived from Hydrogel

Summary Buildings account for 40% of global energy consumption, while windows are the least energy-efficient part of buildings. Conventional smart windows only regulate solar transmission. For the first time, we developed high thermal energy storage thermo-responsive smart window (HTEST smart window) by trapping the hydrogel-derived liquid within glasses. The excellent thermo-responsive optical property (90% of luminous transmittance and 68.1% solar modulation) together with outstanding specific heat capacity of liquid gives the HTEST smart window excellent energy conservation performance. Simulations suggested that HTEST window can cut off 44.6% heating, ventilation, and air-conditioning (HVAC) energy consumption compared with the normal glass in Singapore. In outdoor demonstrations, the HTEST smart window showed promising energy-saving performance in summer daytime. Compared with conventional energy-saving glasses, which need expensive equipment, the thermo-responsive liquid-trapped structure offers a disruptive strategy of easy fabrication, good uniformity, and scalability, together with soundproof functionality that opens an avenue for energy-saving buildings and greenhouses.

Real-time detection tracking and recognition algorithm based on multi-target faces

At present, face recognition algorithms are facing some problems with poor face tracking and low real-time performance in multi-target recognition scenarios. This paper details a multi-target face real-time detection tracking and recognition algorithm, including three methods of fast-tracking, fast detection, and quick recognition. The first step offers a new network based on GOTURN for achieving fast face tracking. The prior information of the previous frame image used to predict the position of the face boxes at the current frame. The second step is based on MTCNN for face detection, using the prior information of the present structure to avoid generating massive of invalid candidate boxes, thereby achieving rapid detection of faces. Finally, fast face recognition realized by reduced MobileFaceNet. By avoiding repeated exposure and repeated identification of the same target, the algorithm successfully transforms a multi-target scene into a single-target scene. On the OTB2015 and 300_VW test sets, the evaluation trackers tracked faces with an accuracy rate of 92.2% and 99.6% respectively. On the Xiph test set, multi-target detection and tracking face speed reached 102fps on the CPU. Compared with the original MobileFaceNet, the streamlined network has an accuracy rate of 99.1% on LFW, the feature extraction speed increased by 25%, and the model size reduced by 45%. Experimental results show that the algorithm has high recognition accuracy and real-time performance in multi-target recognition scenes.

The Discussion on the Application of the Electrical Construction Technology of Prefabricated Buildings

The acceleration of the development of economy and the construction of urbanization has promoted the development of the construction industry, and the construction technology of various construction projects has been continuously innovated. In the past, the building construction often suffered from disadvantages such as high energy consumption and generating large amounts of garbage. Modern construction techniques are actively improving these problems. Prefabricated construction techniques are widely used in the building construction, such as electrical pipeline construction, which makes electrical engineering construction more convenient and environmentally friendly. This paper mainly discusses the construction application of prefabricated electricity in the construction from the aspects of pipeline embedment, lightning protection grounding, the position of electric boxes and so on.

The Discussion on the Application of the Electrical Construction Technology of Prefabricated Buildings

The acceleration of the development of economy and the construction of urbanization has promoted the development of the construction industry, and the construction technology of various construction projects has been continuously innovated. In the past, the building construction often suffered from disadvantages such as high energy consumption and generating large amounts of garbage. Modern construction techniques are actively improving these problems. Prefabricated construction techniques are widely used in the building construction, such as electrical pipeline construction, which makes electrical engineering construction more convenient and environmentally friendly. This paper mainly discusses the construction application of prefabricated electricity in the construction from the aspects of pipeline embedment, lightning protection grounding, the position of electric boxes and so on.