scholarly journals Key Parts of Transmission Line Detection Using Improved YOLO v3

2021 ◽  
Vol 18 (6) ◽  
Author(s):  
Tu Renwei ◽  
Zhu Zhongjie ◽  
Bai Yongqiang ◽  
Gao Ming ◽  
Ge Zhifeng
Keyword(s):  
Transmission Line ◽  
Detection Accuracy ◽  
Data Sets ◽  
Feature Maps ◽  
Data Set ◽  
Detection Model ◽  
The Neural Network ◽  
Low Efficiency ◽  
The One ◽  
Detection Speed

Unmanned Aerial Vehicle (UAV) inspection has become one of main methods for current transmission line inspection, but there are still some shortcomings such as slow detection speed, low efficiency, and inability for low light environment. To address these issues, this paper proposes a deep learning detection model based on You Only Look Once (YOLO) v3. On the one hand, the neural network structure is simplified, that is the three feature maps of YOLO v3 are pruned into two to meet specific detection requirements. Meanwhile, the K-means++ clustering method is used to calculate the anchor value of the data set to improve the detection accuracy. On the other hand, 1000 sets of power tower and insulator data sets are collected, which are inverted and scaled to expand the data set, and are fully optimized by adding different illumination and viewing angles. The experimental results show that this model using improved YOLO v3 can effectively improve the detection accuracy by 6.0%, flops by 8.4%, and the detection speed by about 6.0%.

scholarly journals Transforming Retinal Photographs to Entropy Images in Deep Learning to Improve Automated Detection for Diabetic Retinopathy

2018 ◽  
Vol 2018 ◽  
pp. 1-6 ◽  
Author(s):  
Gen-Min Lin ◽  
Mei-Juan Chen ◽  
Chia-Hung Yeh ◽  
Yu-Yang Lin ◽  
Heng-Yu Kuo ◽  
...  
Keyword(s):  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Sensitivity And Specificity ◽  
Block Size ◽  
Detection Accuracy ◽  
Data Sets ◽  
Feature Maps ◽  
Data Set ◽  
Fundus Photographs ◽  
Grade 3

Entropy images, representing the complexity of original fundus photographs, may strengthen the contrast between diabetic retinopathy (DR) lesions and unaffected areas. The aim of this study is to compare the detection performance for severe DR between original fundus photographs and entropy images by deep learning. A sample of 21,123 interpretable fundus photographs obtained from a publicly available data set was expanded to 33,000 images by rotating and flipping. All photographs were transformed into entropy images using block size 9 and downsized to a standard resolution of 100 × 100 pixels. The stages of DR are classified into 5 grades based on the International Clinical Diabetic Retinopathy Disease Severity Scale: Grade 0 (no DR), Grade 1 (mild nonproliferative DR), Grade 2 (moderate nonproliferative DR), Grade 3 (severe nonproliferative DR), and Grade 4 (proliferative DR). Of these 33,000 photographs, 30,000 images were randomly selected as the training set, and the remaining 3,000 images were used as the testing set. Both the original fundus photographs and the entropy images were used as the inputs of convolutional neural network (CNN), and the results of detecting referable DR (Grades 2–4) as the outputs from the two data sets were compared. The detection accuracy, sensitivity, and specificity of using the original fundus photographs data set were 81.80%, 68.36%, 89.87%, respectively, for the entropy images data set, and the figures significantly increased to 86.10%, 73.24%, and 93.81%, respectively (all p values <0.001). The entropy image quantifies the amount of information in the fundus photograph and efficiently accelerates the generating of feature maps in the CNN. The research results draw the conclusion that transformed entropy imaging of fundus photographs can increase the machinery detection accuracy, sensitivity, and specificity of referable DR for the deep learning-based system.

scholarly journals Lightweight target detection model for embedded platform

2021 ◽  
Vol 2078 (1) ◽  
pp. 012033
Author(s):  
Yuhuan Li ◽  
Jie Wang ◽  
Baodai Shi
Keyword(s):  
Target Detection ◽  
Layer Structure ◽  
Remote Sensing Data ◽  
Detection Algorithm ◽  
Detection Accuracy ◽  
Embedded Devices ◽  
Data Set ◽  
Detection Model ◽  
Embedded Platform ◽  
Detection Speed

Abstract The detection speed of target detection algorithm depends on the performance of computer equipment. Aiming at the problems of slow detection speed and difficult trade-off between detection accuracy and detection speed when the target detection model is used in embedded devices, a lightweight target detection model based on the improved Tiny YOLO-V3 is proposed. Firstly, we analyze the time consumption of each layer structure in the convolutional neural network, and do a lot of experiments and tests. Then, we compress the time-consuming structure substantially. Secondly, we propose the segmentation and fusion module to improve the detection accuracy. Finally, we use the remote sensing data set of Wuhan University for experiments, and the experimental results show that compared with Tiny YOLO-V3, the detection speed is improved by 4 times, and the accuracy is improved by 2 percentage points.

scholarly journals Multiview deep learning based on tensor decomposition and its application in fault detection of overhead contact systems

2021 ◽  
Author(s):  
Xuewu Zhang ◽  
Yansheng Gong ◽  
Chen Qiao ◽  
Wenfeng Jing
Keyword(s):  
High Speed ◽  
Tensor Decomposition ◽  
Detection Methods ◽  
Detection Accuracy ◽  
Feature Maps ◽  
Training Time ◽  
Detection Model ◽  
Railway Line ◽  
Result Show ◽  
Deep Layers

AbstractThis article mainly focuses on the most common types of high-speed railways malfunctions in overhead contact systems, namely, unstressed droppers, foreign-body invasions, and pole number-plate malfunctions, to establish a deep-network detection model. By fusing the feature maps of the shallow and deep layers in the pretraining network, global and local features of the malfunction area are combined to enhance the network's ability of identifying small objects. Further, in order to share the fully connected layers of the pretraining network and reduce the complexity of the model, Tucker tensor decomposition is used to extract features from the fused-feature map. The operation greatly reduces training time. Through the detection of images collected on the Lanxin railway line, experiments result show that the proposed multiview Faster R-CNN based on tensor decomposition had lower miss probability and higher detection accuracy for the three types faults. Compared with object-detection methods YOLOv3, SSD, and the original Faster R-CNN, the average miss probability of the improved Faster R-CNN model in this paper is decreased by 37.83%, 51.27%, and 43.79%, respectively, and average detection accuracy is increased by 3.6%, 9.75%, and 5.9%, respectively.

Data Sets

2013 ◽  
Author(s):  
James B. Elsner ◽  
Thomas H. Jagger
Keyword(s):  
Model Building ◽  
Careful Analysis ◽  
Logical Structure ◽  
Fortran Program ◽  
Data Sets ◽  
Data Set ◽  
Near Surface ◽  
Wind Speeds ◽  
The North ◽  
The One

Hurricane data originate from careful analysis of past storms by operational meteorologists. The data include estimates of the hurricane position and intensity at 6-hourly intervals. Information related to landfall time, local wind speeds, damages, and deaths, as well as cyclone size, are included. The data are archived by season. Some effort is needed to make the data useful for hurricane climate studies. In this chapter, we describe the data sets used throughout this book. We show you a work flow that includes importing, interpolating, smoothing, and adding attributes. We also show you how to create subsets of the data. Code in this chapter is more complicated and it can take longer to run. You can skip this material on first reading and continue with model building in Chapter 7. You can return here when you have an updated version of the data that includes the most recent years. Most statistical models in this book use the best-track data. Here we describe these data and provide original source material. We also explain how to smooth and interpolate them. Interpolations are needed for regional hurricane analyses. The best-track data set contains the 6-hourly center locations and intensities of all known tropical cyclones across the North Atlantic basin, including the Gulf of Mexico and Caribbean Sea. The data set is called HURDAT for HURricane DATa. It is maintained by the U.S. National Oceanic and Atmospheric Administration (NOAA) at the National Hurricane Center (NHC). Center locations are given in geographic coordinates (in tenths of degrees) and the intensities, representing the one-minute near-surface (∼10 m) wind speeds, are given in knots (1 kt = .5144 m s−1) and the minimum central pressures are given in millibars (1 mb = 1 hPa). The data are provided in 6-hourly intervals starting at 00 UTC (Universal Time Coordinate). The version of HURDAT file used here contains cyclones over the period 1851 through 2010 inclusive. Information on the history and origin of these data is found in Jarvinen et al (1984). The file has a logical structure that makes it easy to read with a FORTRAN program. Each cyclone contains a header record, a series of data records, and a trailer record.

scholarly journals Automatic Detection of Transformer Components in Inspection Images Based on Improved Faster R-CNN

Energies ◽  
2018 ◽  
Vol 11 (12) ◽  
pp. 3496 ◽  
Author(s):  
Ziquan Liu ◽  
Huifang Wang
Keyword(s):  
Detection Accuracy ◽  
Training Process ◽  
Feature Maps ◽  
Size Change ◽  
Ordinary Objects ◽  
Detection Model ◽  
Significant Difference ◽  
High Detection ◽  
Practical Engineering ◽  
Improved Model

To detect the categories and positions of various transformer components in inspection images automatically, this paper proposes a transformer component detection model with high detection accuracy, based on the structure of Faster R-CNN. In consideration of the significant difference in component sizes, double feature maps are used to adapt to the size change, by adjusting two weights dynamically according to the object size. Moreover, different from the detection of ordinary objects, there is abundant useful information contained in the relative positions between components. Thus, the relative position features are defined and introduced to the refinement of the detection results. Then, the training process and detection process are proposed specifically for the improved model. Finally, an experiment is given to compare the accuracy and efficiency of the improved model and the original Faster R-CNN, along with other object detection models. Results show that the improved model has an obvious advantage in accuracy, and the efficiency is significantly higher than that of manual detection, which suggests that the model is suitable for practical engineering applications.

scholarly journals ISSD: Improved SSD for Insulator and Spacer Online Detection Based on UAV System

Sensors ◽  
2020 ◽  
Vol 20 (23) ◽  
pp. 6961
Author(s):  
Xuan Liu ◽  
Yong Li ◽  
Feng Shuang ◽  
Fang Gao ◽  
Xiang Zhou ◽  
...  
Keyword(s):  
Capture Rate ◽  
Aerial Images ◽  
Detection Accuracy ◽  
Single Shot ◽  
Online Detection ◽  
Feature Maps ◽  
Fast Detection ◽  
Model Size ◽  
Small Model ◽  
Detection Speed

In power inspection tasks, the insulator and spacer are important inspection objects. UAV (unmanned aerial vehicle) power inspection is becoming more and more popular. However, due to the limited computing resources carried by a UAV, a lighter model with small model size, high detection accuracy, and fast detection speed is needed to achieve online detection. In order to realize the online detection of power inspection objects, we propose an improved SSD (single shot multibox detector) insulator and spacer detection algorithm using the power inspection images collected by a UAV. In the proposed algorithm, the lightweight network MnasNet is used as the feature extraction network to generate feature maps. Then, two multiscale feature fusion methods are used to fuse multiple feature maps. Lastly, a power inspection object dataset containing insulators and spacers based on aerial images is built, and the performance of the proposed algorithm is tested on real aerial images and videos. Experimental results show that the proposed algorithm can efficiently detect insulators and spacers. Compared with existing algorithms, the proposed algorithm has the advantages of small model size and fast detection speed. The detection accuracy can achieve 93.8%. The detection time of a single image on TX2 (NVIDIA Jetson TX2) is 154 ms and the capture rate on TX2 is 8.27 fps, which allows realizing online detection.

EFFECTS OF NONSINGULAR PREPROCESSING ON FEEDFORWARD NETWORK TRAINING

2005 ◽  
Vol 19 (02) ◽  
pp. 217-247 ◽  
Author(s):  
CHANGHUA YU ◽  
MICHAEL T. MANRY ◽  
JIANG LI
Keyword(s):  
Back Propagation ◽  
Original Data ◽  
Data Sets ◽  
Training Algorithm ◽  
Feedforward Network ◽  
Data Set ◽  
Network Training ◽  
The Neural Network ◽  
Hidden Layer ◽  
Theoretical Analyses

In the neural network literature, many preprocessing techniques, such as feature de-correlation, input unbiasing and normalization, are suggested to accelerate multilayer perceptron training. In this paper, we show that a network trained with an original data set and one trained with a linear transformation of the original data will go through the same training dynamics, as long as they start from equivalent states. Thus preprocessing techniques may not be helpful and are merely equivalent to using a different weight set to initialize the network. Theoretical analyses of such preprocessing approaches are given for conjugate gradient, back propagation and the Newton method. In addition, an efficient Newton-like training algorithm is proposed for hidden layer training. Experiments on various data sets confirm the theoretical analyses and verify the improvement of the new algorithm.

The implications of the revised code of corporate governance on firm performance

2015 ◽  
Vol 5 (3) ◽  
pp. 350-380 ◽  
Author(s):  
Abdifatah Ahmed Haji ◽  
Sanni Mubaraq
Keyword(s):  
Corporate Governance ◽  
Firm Performance ◽  
Ownership Structure ◽  
Market Performance ◽  
Negative Relationship ◽  
Future Research ◽  
Data Sets ◽  
Data Set ◽  
Content Type ◽  
The One

Purpose – The purpose of this paper is to examine the impact of corporate governance and ownership structure attributes on firm performance following the revised code on corporate governance in Malaysia. The study presents a longitudinal assessment of the compliance and implications of the revised code on firm performance. Design/methodology/approach – Two data sets consisting of before (2006) and after (2008-2010) the revised code are examined. Drawing from the largest companies listed on Bursa Malaysia (BM), the first data set contains 92 observations in the year 2006 while the second data set comprises of 282 observations drawn from the largest companies listed on BM over a three-year period, from 2008-2010. Both accounting (return on assets and return on equity) and market performance (Tobin’s Q) measures were used to measure firm performance. Multiple and panel data regression analyses were adopted to analyze the data. Findings – The study shows that there were still cases of non-compliance to the basic requirements of the code such as the one-third independent non-executive director (INDs) requirement even after the revised code. While the regression models indicate marginal significance of board size and independent directors before the revised code, the results indicate all corporate governance variables have a significant negative relationship with at least one of the measures of corporate performance. Independent chairperson, however, showed a consistent positive impact on firm performance both before and after the revised code. In addition, ownership structure elements were found to have a negative relationship with either accounting or market performance measures, with institutional ownership showing a consistent negative impact on firm performance. Firm size and leverage, as control variables, were significant in determining corporate performance. Research limitations/implications – One limitation is the use of separate measures of corporate governance attributes, as opposed to a corporate governance index (CGI). As a result, the study constructs a CGI based on the recommendations of the revised code and proposes for future research use. Practical implications – Some of the largest companies did not even comply with basic requirements such as the “one-third INDs” mandatory requirement. Hence, the regulators may want to reinforce the requirements of the code and also detail examples of good governance practices. The results, which show a consistent positive relationship between the presence of an independent chairperson and firm performance in both data sets, suggest listed companies to consider appointing an independent chairperson in the corporate leadership. The regulatory authorities may also wish to note this phenomenon when drafting any future corporate governance codes. Originality/value – This study offers new insights of the implications of regulatory changes on the relationship between corporate governance attributes and firm performance from the perspective of a developing country. The development of a CGI for future research is a novel approach of this study.

scholarly journals GLEAM v3: satellite-based land evaporation and root-zone soil moisture

2016 ◽  
Author(s):  
Brecht Martens ◽  
Diego G. Miralles ◽  
Hans Lievens ◽  
Robin van der Schalie ◽  
Richard A. M. de Jeu ◽  
...  
Keyword(s):  
Soil Moisture ◽  
Eddy Covariance ◽  
Optical Depth ◽  
Root Zone ◽  
European Space Agency ◽  
Data Sets ◽  
Data Set ◽  
Previous Version ◽  
The One

Abstract. The Global Land Evaporation Amsterdam Model (GLEAM) is a set of algorithms dedicated to the estimation of terrestrial evaporation and root-zone soil moisture from satellite data. Ever since its development in 2011, the model has been regularly revised aiming at the optimal incorporation of new satellite-observed geophysical variables, and improving the representation of physical processes. In this study, the next version of this model (v3) is presented. Key changes relative to the previous version include: (1) a revised formulation of the evaporative stress, (2) an optimized drainage algorithm, and (3) a new soil moisture data assimilation system. GLEAM v3 is used to produce three new data sets of terrestrial evaporation and root-zone soil moisture, including a 35-year data set spanning the period 1980–2014 (v3.0a, based on satellite-observed soil moisture, vegetation optical depth and snow water equivalents, reanalysis air temperature and radiation, and a multi-source precipitation product), and two fully satellite-based data sets. The latter two share most of their forcing, except for the vegetation optical depth and soil moisture products, which are based on observations from different passive and active C- and L-band microwave sensors (European Space Agency Climate Change Initiative data sets) for the first data set (v3.0b, spanning the period 2003–2015) and observations from the Soil Moisture and Ocean Salinity satellite in the second data set (v3.0c, spanning the period 2011–2015). These three data sets are described in detail, compared against analogous data sets generated using the previous version of GLEAM (v2), and validated against measurements from 64 eddy-covariance towers and 2338 soil moisture sensors across a broad range of ecosystems. Results indicate that the quality of the v3 soil moisture is consistently better than the one from v2: average correlations against in situ surface soil moisture measurements increase from 0.61 to 0.64 in case of the v3.0a data set and the representation of soil moisture in the second layer improves as well, with correlations increasing from 0.47 to 0.53. Similar improvements are observed for the two fully satellite-based data sets. Despite regional differences, the quality of the evaporation fluxes remains overall similar as the one obtained using the previous version of GLEAM, with average correlations against eddy-covariance measurements between 0.78 and 0.80 for the three different data sets. These global data sets of terrestrial evaporation and root-zone soil moisture are now openly available at http://GLEAM.eu and may be used for large-scale hydrological applications, climate studies and research on land-atmosphere feedbacks.

scholarly journals ForkDec: Accurate Detection for Selfish Mining Attacks

2021 ◽  
Vol 2021 ◽  
pp. 1-8
Author(s):  
Zhaojie Wang ◽  
Qingzhe Lv ◽  
Zhaobo Lu ◽  
Yilei Wang ◽  
Shengjie Yue
Keyword(s):  
Neural Network ◽  
Incentive Mechanism ◽  
Attack Detection ◽  
Detection Accuracy ◽  
Data Set ◽  
Incentive Compatible ◽  
The Neural Network ◽  
Security Challenges ◽  
Excellent Research ◽  
Fully Connected

Incentive mechanism is the key to the success of the Bitcoin system as a permissionless blockchain. It encourages participants to contribute their computing resources to ensure the correctness and consistency of user transaction records. Selfish mining attacks, however, prove that Bitcoin’s incentive mechanism is not incentive-compatible, which is contrary to traditional views. Selfish mining attacks may cause the loss of mining power, especially those of honest participants, which brings great security challenges to the Bitcoin system. Although there are a series of studies against selfish mining behaviors, these works have certain limitations: either the existing protocol needs to be modified or the detection effect for attacks is not satisfactory. We propose the ForkDec, a high-accuracy system for selfish mining detection based on the fully connected neural network, for the purpose of effectively deterring selfish attackers. The neural network contains a total of 100 neurons (10 hidden layers and 10 neurons per layer), learned on a training set containing about 200,000 fork samples. The data set, used to train the model, is generated by a Bitcoin mining simulator that we preconstructed. We also applied ForkDec to the test set to evaluate the attack detection and achieved a detection accuracy of 99.03%. The evaluation experiment demonstrates that ForkDec has certain application value and excellent research prospects.

Sign in / Sign up

Export Citation Format

Share Document