scholarly journals Multilabel Video Classification Model of Navigation Mark’s Lights Based on Deep Learning

2021 ◽  
Vol 2021 ◽  
pp. 1-13
Author(s):  
Xu Han ◽  
Mingyang Pan ◽  
Haipeng Ge ◽  
Shaoxi Li ◽  
Jingfeng Hu ◽  
...  
Keyword(s):  
High Speed ◽  
Temporal Dynamics ◽  
Research Direction ◽  
Classification Model ◽  
Video Frame ◽  
Video Classification ◽  
Data Set ◽  
Binary Relevance ◽  
Power Set ◽  
Intelligent Recognition

At night, buoys and other navigation marks disappear to be replaced by fixed or flashing lights. Navigation marks are seen as a set of lights in various colors rather than their familiar outline. Deciphering that the meaning of the lights is a burden to navigators, it is also a new challenging research direction of intelligent sensing of navigation environment. The study studied initiatively the intelligent recognition of lights on navigation marks at night based on multilabel video classification methods. To capture effectively the characteristics of navigation mark’s lights, including both color and flashing phase, three different multilabel classification models based on binary relevance, label power set, and adapted algorithm were investigated and compared. According to the experiment’s results performed on a data set with 8000 minutes video, the model based on binary relevance, named NMLNet, has highest accuracy about 99.23% to classify 9 types of navigation mark’s lights. It also has the fastest computation speed with least network parameters. In the NMLNet, there are two branches for the classifications of color and flashing, respectively, and for the flashing classification, an improved MobileNet-v2 was used to capture the brightness characteristic of lights in each video frame, and an LSTM is used to capture the temporal dynamics of lights. Aiming to run on mobile devices on vessel, the MobileNet-v2 was used as backbone, and with the improvement of spatial attention mechanism, it achieved the accuracy near Resnet-50 while keeping its high speed.

Factors Affecting Driver Injury Severity in the Wrong-Way Crash: Accounting for Potential Heterogeneity in Means and Variances of Random Parameters

2021 ◽  
pp. 036119812110098
Author(s):  
Miao Yu ◽  
Jinxing Shen ◽  
Changxi Ma
Keyword(s):  
High Speed ◽  
Injury Severity ◽  
Unobserved Heterogeneity ◽  
Research Effort ◽  
Random Parameter ◽  
Policy Implications ◽  
Contributing Factors ◽  
Severe Injuries ◽  
Data Set ◽  
Severity Levels

Because of the high percentage of fatalities and severe injuries in wrong-way driving (WWD) crashes, numerous studies have focused on identifying contributing factors to the occurrence of WWD crashes. However, a limited number of research effort has investigated the factors associated with driver injury-severity in WWD crashes. This study intends to bridge the gap using a random parameter logit model with heterogeneity in means and variances approach that can account for the unobserved heterogeneity in the data set. Police-reported crash data collected from 2014 to 2017 in North Carolina are used. Four injury-severity levels are defined: fatal injury, severe injury, possible injury, and no injury. Explanatory variables, including driver characteristics, roadway characteristics, environmental characteristics, and crash characteristics, are used. Estimation results demonstrate that factors, including the involvement of alcohol, rural area, principal arterial, high speed limit (>60 mph), dark-lighted conditions, run-off-road collision, and head-on collision, significantly increase the severity levels in WWD crashes. Several policy implications are designed and recommended based on findings.

scholarly journals Comparison of Spectroscopic Techniques Combined with Chemometrics for Cocaine Powder Analysis

2020 ◽  
Vol 44 (8) ◽  
pp. 851-860
Author(s):  
Joy Eliaerts ◽  
Natalie Meert ◽  
Pierre Dardenne ◽  
Vincent Baeten ◽  
Juan-Antonio Fernandez Pierna ◽  
...  
Keyword(s):  
Gas Chromatography ◽  
Near Infrared ◽  
Evaluation Criteria ◽  
Classification Model ◽  
Support Vector ◽  
Spectroscopic Techniques ◽  
Data Set ◽  
Promising Tool ◽  
Powder Analysis ◽  
Mir Spectra

Abstract Spectroscopic techniques combined with chemometrics are a promising tool for analysis of seized drug powders. In this study, the performance of three spectroscopic techniques [Mid-InfraRed (MIR), Raman and Near-InfraRed (NIR)] was compared. In total, 364 seized powders were analyzed and consisted of 276 cocaine powders (with concentrations ranging from 4 to 99 w%) and 88 powders without cocaine. A classification model (using Support Vector Machines [SVM] discriminant analysis) and a quantification model (using SVM regression) were constructed with each spectral dataset in order to discriminate cocaine powders from other powders and quantify cocaine in powders classified as cocaine positive. The performances of the models were compared with gas chromatography coupled with mass spectrometry (GC–MS) and gas chromatography with flame-ionization detection (GC–FID). Different evaluation criteria were used: number of false negatives (FNs), number of false positives (FPs), accuracy, root mean square error of cross-validation (RMSECV) and determination coefficients (R2). Ten colored powders were excluded from the classification data set due to fluorescence background observed in Raman spectra. For the classification, the best accuracy (99.7%) was obtained with MIR spectra. With Raman and NIR spectra, the accuracy was 99.5% and 98.9%, respectively. For the quantification, the best results were obtained with NIR spectra. The cocaine content was determined with a RMSECV of 3.79% and a R2 of 0.97. The performance of MIR and Raman to predict cocaine concentrations was lower than NIR, with RMSECV of 6.76% and 6.79%, respectively and both with a R2 of 0.90. The three spectroscopic techniques can be applied for both classification and quantification of cocaine, but some differences in performance were detected. The best classification was obtained with MIR spectra. For quantification, however, the RMSECV of MIR and Raman was twice as high in comparison with NIR. Spectroscopic techniques combined with chemometrics can reduce the workload for confirmation analysis (e.g., chromatography based) and therefore save time and resources.

scholarly journals An efficient technique for CT scan images classification of COVID-19

2020 ◽  
pp. 1-14
Author(s):  
Esraa Hassan ◽  
Noha A. Hikal ◽  
Samir Elmuogy
Keyword(s):  
Neural Network ◽  
Diagnostic Tool ◽  
Deep Neural Network ◽  
Data Augmentation ◽  
Performance Metrics ◽  
Classification Model ◽  
Data Set ◽  
Training Models ◽  
The Earth ◽  
Diagnostic Time

Nowadays, Coronavirus (COVID-19) considered one of the most critical pandemics in the earth. This is due its ability to spread rapidly between humans as well as animals. COVID_19 expected to outbreak around the world, around 70 % of the earth population might infected with COVID-19 in the incoming years. Therefore, an accurate and efficient diagnostic tool is highly required, which the main objective of our study. Manual classification was mainly used to detect different diseases, but it took too much time in addition to the probability of human errors. Automatic image classification reduces doctors diagnostic time, which could save human’s life. We propose an automatic classification architecture based on deep neural network called Worried Deep Neural Network (WDNN) model with transfer learning. Comparative analysis reveals that the proposed WDNN model outperforms by using three pre-training models: InceptionV3, ResNet50, and VGG19 in terms of various performance metrics. Due to the shortage of COVID-19 data set, data augmentation was used to increase the number of images in the positive class, then normalization used to make all images have the same size. Experimentation is done on COVID-19 dataset collected from different cases with total 2623 where (1573 training,524 validation,524 test). Our proposed model achieved 99,046, 98,684, 99,119, 98,90 In terms of Accuracy, precision, Recall, F-score, respectively. The results are compared with both the traditional machine learning methods and those using Convolutional Neural Networks (CNNs). The results demonstrate the ability of our classification model to use as an alternative of the current diagnostic tool.

scholarly journals Simultaneous spatiotemporal super-resolution and multi-parametric fluorescence microscopy

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Jagadish Sankaran ◽  
Harikrushnan Balasubramanian ◽  
Wai Hoh Tang ◽  
Xue Wen Ng ◽  
Adrian Röllin ◽  
...  
Keyword(s):  
Single Molecule ◽  
Temporal Dynamics ◽  
Growth Factor Receptor ◽  
Super Resolution ◽  
Actin Binding ◽  
Biological Knowledge ◽  
Data Set ◽  
Epidermal Growth ◽  
Molecular Brightness ◽  
Super Resolution Microscopy

AbstractSuper-resolution microscopy and single molecule fluorescence spectroscopy require mutually exclusive experimental strategies optimizing either temporal or spatial resolution. To achieve both, we implement a GPU-supported, camera-based measurement strategy that highly resolves spatial structures (~100 nm), temporal dynamics (~2 ms), and molecular brightness from the exact same data set. Simultaneous super-resolution of spatial and temporal details leads to an improved precision in estimating the diffusion coefficient of the actin binding polypeptide Lifeact and corrects structural artefacts. Multi-parametric analysis of epidermal growth factor receptor (EGFR) and Lifeact suggests that the domain partitioning of EGFR is primarily determined by EGFR-membrane interactions, possibly sub-resolution clustering and inter-EGFR interactions but is largely independent of EGFR-actin interactions. These results demonstrate that pixel-wise cross-correlation of parameters obtained from different techniques on the same data set enables robust physicochemical parameter estimation and provides biological knowledge that cannot be obtained from sequential measurements.

scholarly journals Mapping the Spatial-Temporal Dynamics of Vegetation Response Lag to Drought in a Semi-Arid Region

2019 ◽  
Vol 11 (16) ◽  
pp. 1873 ◽  
Author(s):  
Li Hua ◽  
Huidong Wang ◽  
Haigang Sui ◽  
Brian Wardlow ◽  
Michael J. Hayes ◽  
...  
Keyword(s):  
Vegetation Index ◽  
Temporal Dynamics ◽  
Risk Mitigation ◽  
Scientific Information ◽  
Gaussian Function ◽  
Drought Risk ◽  
Severe Drought ◽  
Similar Response ◽  
Vegetation Types ◽  
Data Set

Drought, as an extreme climate event, affects the ecological environment for vegetation and agricultural production. Studies of the vegetative response to drought are paramount to providing scientific information for drought risk mitigation. In this paper, the spatial-temporal pattern of drought and the response lag of vegetation in Nebraska were analyzed from 2000 to 2015. Based on the long-term Daymet data set, the standard precipitation index (SPI) was computed to identify precipitation anomalies, and the Gaussian function was applied to obtain temperature anomalies. Vegetation anomaly was identified by dynamic time warping technique using a remote sensing Normalized Difference Vegetation Index (NDVI) time series. Finally, multilayer correlation analysis was applied to obtain the response lag of different vegetation types. The results show that Nebraska suffered severe drought events in 2002 and 2012. The response lag of vegetation to drought typically ranged from 30 to 45 days varying for different vegetation types and human activities (water use and management). Grasslands had the shortest response lag (~35 days), while forests had the longest lag period (~48 days). For specific crop types, the response lag of winter wheat varied among different regions of Nebraska (35–45 days), while soybeans, corn and alfalfa had similar response lag times of approximately 40 days.

scholarly journals Research on Real-Time Multiple Single Garbage Classification Based on Convolutional Neural Network

2020 ◽  
Vol 2020 ◽  
pp. 1-6
Author(s):  
Jian-ye Yuan ◽  
Xin-yuan Nan ◽  
Cheng-rong Li ◽  
Le-le Sun
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Real Time ◽  
Environmental Science ◽  
Fine Tuning ◽  
Classification Model ◽  
Classification Models ◽  
Data Set ◽  
Classification Tasks ◽  
Accuracy Rates

Considering that the garbage classification is urgent, a 23-layer convolutional neural network (CNN) model is designed in this paper, with the emphasis on the real-time garbage classification, to solve the low accuracy of garbage classification and recycling and difficulty in manual recycling. Firstly, the depthwise separable convolution was used to reduce the Params of the model. Then, the attention mechanism was used to improve the accuracy of the garbage classification model. Finally, the model fine-tuning method was used to further improve the performance of the garbage classification model. Besides, we compared the model with classic image classification models including AlexNet, VGG16, and ResNet18 and lightweight classification models including MobileNetV2 and SuffleNetV2 and found that the model GAF_dense has a higher accuracy rate, fewer Params, and FLOPs. To further check the performance of the model, we tested the CIFAR-10 data set and found the accuracy rates of the model (GAF_dense) are 0.018 and 0.03 higher than ResNet18 and SufflenetV2, respectively. In the ImageNet data set, the accuracy rates of the model (GAF_dense) are 0.225 and 0.146 higher than Resnet18 and SufflenetV2, respectively. Therefore, the garbage classification model proposed in this paper is suitable for garbage classification and other classification tasks to protect the ecological environment, which can be applied to classification tasks such as environmental science, children’s education, and environmental protection.

MONITORING AND CLASSIFYING THE STATE OF HARD DISKS USING RECURRENT NEURAL NETWORKS

2021 ◽  
pp. 36-43
Author(s):  
L. A. Demidova ◽  
A. V. Filatov
Keyword(s):  
Neural Networks ◽  
Recurrent Neural Networks ◽  
Experimental Studies ◽  
Machine Learning Algorithms ◽  
Classification Model ◽  
Dense Layer ◽  
Hard Disks ◽  
Destructive Testing ◽  
Data Set ◽  
State Classification

The article considers an approach to solving the problem of monitoring and classifying the states of hard disks, which is solved on a regular basis, within the framework of the concept of non-destructive testing. It is proposed to solve this problem by developing a classification model using machine learning algorithms, in particular, using recurrent neural networks with Simple RNN, LSTM and GRU architectures. To develop a classification model, a data set based on the values of SMART sensors installed on hard disks it used. It represents a group of multidimensional time series. At the same time, the structure of the classification model contains two layers of a neural network with one of the recurrent architectures, as well as a Dropout layer and a Dense layer. The results of experimental studies confirming the advantages of LSTM and GRU architectures as part of hard disk state classification models are presented.

A Novel ECG Automatic Detection Using LongShort-Term Memory Network and Internet of Things Technology

2021 ◽  
Vol 11 (6) ◽  
pp. 1592-1598
Author(s):  
Xufei Liu
Keyword(s):  
Machine Learning ◽  
Cardiovascular Diseases ◽  
Internet Of Things ◽  
Network Structure ◽  
Recognition Rate ◽  
Classification Model ◽  
Data Set ◽  
Public Data ◽  
Lstm Network ◽  
Internet Of Things Technology

The early detection of cardiovascular diseases based on electrocardiogram (ECG) is very important for the timely treatment of cardiovascular patients, which increases the survival rate of patients. ECG is a visual representation that describes changes in cardiac bioelectricity and is the basis for detecting heart health. With the rise of edge machine learning and Internet of Things (IoT) technologies, small machine learning models have received attention. This study proposes an ECG automatic classification method based on Internet of Things technology and LSTM network to achieve early monitoring and early prevention of cardiovascular diseases. Specifically, this paper first proposes a single-layer bidirectional LSTM network structure. Make full use of the timing-dependent features of the sampling points before and after to automatically extract features. The network structure is more lightweight and the calculation complexity is lower. In order to verify the effectiveness of the proposed classification model, the relevant comparison algorithm is used to verify on the MIT-BIH public data set. Secondly, the model is embedded in a wearable device to automatically classify the collected ECG. Finally, when an abnormality is detected, the user is alerted by an alarm. The experimental results show that the proposed model has a simple structure and a high classification and recognition rate, which can meet the needs of wearable devices for monitoring ECG of patients.

Bayesian Classifier for Sparsity-Promoting Feature Selection

2015 ◽  
Vol 29 (06) ◽  
pp. 1550022 ◽  
Author(s):  
Danlei Xu ◽  
Lan Du ◽  
Hongwei Liu ◽  
Penghui Wang
Keyword(s):  
Feature Selection ◽  
Synthetic Data ◽  
Original Data ◽  
Radar Data ◽  
Bayesian Classifier ◽  
Classification Model ◽  
Data Sets ◽  
Data Set ◽  
Classification Boundary ◽  
Nonlinear Mappings

A Bayesian classifier for sparsity-promoting feature selection is developed in this paper, where a set of nonlinear mappings for the original data is performed as a pre-processing step. The linear classification model with such mappings from the original input space to a nonlinear transformation space can not only construct the nonlinear classification boundary, but also realize the feature selection for the original data. A zero-mean Gaussian prior with Gamma precision and a finite approximation of Beta process prior are used to promote sparsity in the utilization of features and nonlinear mappings in our model, respectively. We derive the Variational Bayesian (VB) inference algorithm for the proposed linear classifier. Experimental results based on the synthetic data set, measured radar data set, high-dimensional gene expression data set, and several benchmark data sets demonstrate the aggressive and robust feature selection capability and comparable classification accuracy of our method comparing with some other existing classifiers.

scholarly journals Three-Dimensional (3-D) Immersed-Boundary (IB) Based Tornado Model for Computational Analysis of Disastrous Wind Load on Buildings

2021 ◽  
Author(s):  
Xixiong Guo ◽  
Jun Cao
Keyword(s):  
Wind Field ◽  
High Speed ◽  
Outer Boundary ◽  
Three Dimensional ◽  
Influence Factors ◽  
Research Direction ◽  
Wind Loading ◽  
Immersed Boundary ◽  
Design Strategies ◽  
Straight Line

This study is aimed at developing a novel computational framework that can essentially simulate a tornadic wind field and investigate the wind loadings on ground constructions. It is well known that tornado is a highly turbulent airflow that simultaneously translates, rotates and updrafts with a high speed. Tornadoes induce a significantly elevated level of wind forces if compared to a straight-line wind. A suitably designed building for a straight-line wind would fail to survive when exposed to a tornadic-like wind of the same wind speed. It is necessary to design buildings that are more resistant to tornadoes. Since the study of tornado dynamics relying on field observations and laboratory experiments is usually expensive, restrictive, and time-consuming, computer simulation mainly via the large eddy simulation (LES) method has become a more attractive research direction in shedding light on the intricate characteristics of a tornadic wind field. For numerical simulation of a tornado-building interaction scenario, it looks quite challenging to seek a set of physically-rational and meanwhile computationally-practical boundary conditions to accompany traditional CFD approaches; however, little literature can be found, as of today, in three-dimensional (3D) computational tornado dynamics study. Inspired by the development of the immersed boundary (IB) method, this study employed a re-tailored Rankine-combined vortex model (RCVM) that applies the “relative motion” principle to the translational component of tornado, such that the building is viewed as “virtually” translating towards a “pinned” rotational flow that remains time-invariant at the far field region. This revision renders a steady-state kinematic condition applicable to the outer boundary of a large tornado simulation domain, successfully circumventing the boundary condition updating process that the original RCVM would have to suffer, and tremendously accelerating the computation. Wind loading and its influence factors are comprehensively investigated and analyzed both on a single building and on a multiple-building configuration. The relation between the wind loadings and the height and shape of the building is also examined in detail. Knowledge of these loadings may lead to design strategies that can enable ground construction to be more resistant to tornadoes, reducing the losses caused by this type of disastrous weather.

Sign in / Sign up

Export Citation Format

Share Document