Average Correct Classification Rate Research Articles

Utilizing Deep Learning for Defect Inspection in Hand Tool Assembly.

The integrity of product assembly in the precision assembly industry significantly influences the quality of the final products. During the assembly process, products may acquire assembly defects due to personnel oversight. A severe assembly defect could impair the product's normal function and potentially cause loss of life or property for the user. For workpiece defect inspection, there is limited discussion on the simultaneous detection of the primary kinds of assembly anomaly (missing parts, misplaced parts, foreign objects, and extra parts). However, these assembly anomalies account for most customer complaints in the traditional hand tool industry. This is because no equipment can comprehensively inspect major assembly defects, and inspections rely solely on professionals using simple tools and their own experience. Thus, this study proposes an automated visual inspection system to achieve defect inspection in hand tool assembly. This study samples the work-in-process from three assembly stations in the ratchet wrench assembly process; an investigation of 28 common assembly defect types is presented, covering the 4 kinds of assembly anomaly in the assembly operation; also, this study captures sample images of various assembly defects for the experiments. First, the captured images are filtered to eliminate surface reflection noise from the workpiece; then, a circular mask is given at the assembly position to extract the ROI area; next, the filtered ROI images are used to create a defect-type label set using manual annotation; after this, the R-CNN series network models are applied to object feature extraction and classification; finally, they are compared with other object detection models to identify which inspection model has the better performance. The experimental results show that, if each station uses the best model for defect inspection, it can effectively detect and classify defects. The average defect detection rate (1-β) of each station is 92.64%, the average misjudgment rate (α) is 6.68%, and the average correct classification rate (CR) is 88.03%.

Deep learning-based image classification of turtles imported into Korea

Although turtles play a key role in maintaining healthy and balanced environments, these are endangered due to global trade to meet the high demand for food, medicine, and pets in Asia. In addition, imported non-native turtles have been controlled as alien invasive species in various countries, including Korea. Therefore, a rapid and accurate classification of imported turtles is needed to conserve and detect those in native ecosystems. In this study, eight Single Shot MultiBox Detector (SSD) models using different backbone networks were used to classify 36 imported turtles in Korea. The images of these species were collected from Google and were identified using morphological features. Then, these were divided into 70% for training, 15% for validation, and 15% for test sets. In addition, data augmentation was applied to the training set to prevent overfitting. Among the eight models, the Resnet18 model showed the highest mean Average Precision (mAP) at 88.1% and the fastest inference time at 0.024 s. The average correct classification rate of 36 turtles in this model was 82.8%. The results of this study could help in management of the turtle trade, specifically in improving detection of alien invasive species in the wild.

Elucidating the mechanism of action of mycotoxins through machine learning-driven QSAR models: Focus on lipid peroxidation

Understanding the mechanisms of mycotoxin toxicity is crucial for establishing effective guidelines and preventive strategies. In this study, machine learning models based on quantitative structure-activity relationship (QSAR) were employed to predict the lipid peroxidation activity of mycotoxins. Two different algorithms using Linear Discriminant Analysis (LDA) and Artificial Neural Networks (ANNs) have been trained using a dataset of 70 mycotoxins. The LDA model had an average correct classification rate of 91%, while the ANN model achieved a perfect 100% classification rate. Following an internal validation process, the models were utilized to predict mycotoxins with known lipid peroxidation activity. The machine learning models achieved an 88% correct classification rate for these mycotoxins. Finally, by utilizing classified algorithms, the study aimed to infer the mechanism of action related to lipid peroxidation for 91 unstudied mycotoxins. These models provide a fast, accurate, and cost-effective means to assess the potential toxicity and mechanism of action of mycotoxins. The findings of this study contribute to a comprehensive understanding of mycotoxin toxicology and assist researchers and toxicologists in evaluating health risks associated with mycotoxin exposure and developing appropriate preventive strategies and potential therapeutic interventions to mitigate the effects of mycotoxins.

Evaluation of medicinal plants using laser-induced breakdown spectroscopy (LIBS) combined with chemometric techniques.

Medicinal plants play a vital role in herbal medical field and allopathic medicine field industry. Chemical and spectroscopic studies of Taraxacum officinale, Hyoscyamus niger, Ajuga bracteosa, Elaeagnus angustifolia, Camellia sinensis, and Berberis lyceum are conducted in this paper by using a 532-nm Nd:YAG laser in an open air environment. These medicinal plant's leaves, roots, seed, and flowers are used to treat a range of diseases by the locals. It is crucial to be able to distinguish between beneficial and detrimental metal elements in these plants. We demonstrated how various elements are categorized and how roots, leaves, seeds and flowers of same plants differ from each other on the basis of elemental analysis. Furthermore, for classification purpose, different classification models, partial least square discriminant analysis (PLS-DA), k-nearest neighbors (kNN), and principal component analysis (PCA) are used. We found silicon (Si), aluminum (Al), iron (Fe), copper (Cu), calcium (Ca), magnesium (Mg), sodium (Na), potassium (K), manganese (Mn), phosphorous (P), and vanadium (V) in all of the medicinal plant samples with a molecular form of carbon and nitrogen band. We detected Ca, Mg, Si, and P as primary components in all of the plant samples, as well as V, Fe, Mn, Al, and Ti as essential medicinal metals, and additional trace elements like Si, Sr, and Al. The result's findings show that the PLS-DA classification model with single normal variate (SNV) preprocessing method is the most effective classification model for different types of plant samples. The average correct classification rate obtained for PLS-DA with SNV is 95%. Moreover, laser-induced breakdown spectroscopy (LIBS) was successfully employed to perform rapid, sensitive, and quantitative trace element analysis on medicinal herbs and plant samples.

LiDAR Vehicle Point Cloud Reconstruction Framework for Axle-Based Classification

As Light Detection and Range (LiDAR) technology rapidly advances, it is becoming increasingly viable as a solution to collect vehicle classification data. The main challenge of LiDAR compared with video in vehicle classification lies in its resolution, which limits the ability of LiDAR-based models to classify vehicles in detail from a single captured frame. This paper proposes a novel framework for reconstructing the vehicle point clouds generated by a roadside LiDAR sensor, by considering ground plane constraints and developing a bootstrap aggregating deep neural network (bagging DNN) model to classify the reconstructed vehicle point clouds, according to the US Federal Highway Administration (FHWA) axle-based vehicle classification scheme. First, a probabilistic-based registration algorithm is used to estimate the transformation matrix between consecutive frames of each vehicle point cloud. Then, a multiway registration is conducted to fine-tune the estimated transformation matrices to rebuild the 3D model of each moving vehicle. Second, key features are extracted from the reconstructed vehicle models and fed into a bagging DNN model to provide classifications based on the FHWA scheme. The classification model with the reconstruction framework outperforms the latest developed LiDAR-based FHWA classification model in terms of both accuracy and robustness. The model has an 83 percent average correct classification rate (CCR) on the test set. Remarkably, the proposed model can accurately distinguish Classes 5 and 8 trucks, which have overlapping axle configurations, with a 97 percent and a 90 percent CCR, respectively.

Characterization of leaf surface phenotypes based on light interaction

BackgroundLeaf surface phenotypes can indicate plant health and relate to a plant’s adaptations to environmental stresses. Identifying these phenotypes using non-invasive techniques can assist in high-throughput phenotyping and can improve decision making in plant breeding. Identification of these surface phenotypes can also assist in stress identification. Incorporating surface phenotypes into leaf optical modelling can lead to improved biochemical parameter retrieval and species identification.ResultsIn this paper, leaf surface phenotypes are characterized for 349 leaf samples based on polarized light reflectance measured at Brewster’s Angle, and microscopic observation. Four main leaf surface phenotypes (glossy wax, glaucous wax, high trichome density, and glabrous) were identified for the leaf samples. The microscopic and visual observations of the phenotypes were used as ground truth for comparison with the spectral classification. In addition to surface classification, the microscope images were used to assess cell size, shape, and cell cap aspect ratios; these surface attributes were not found to correlate significantly with spectral measurements obtained in this study. Using a quadratic discriminant analysis function, a series of 10,000 classifications were run with the data randomly split between training and testing datasets, with 150 and 199 samples, respectively. The average correct classification rate was 72.9% with a worst-case classification of 60.3%.ConclusionsLeaf surface phenotypes were successfully correlated with spectral measurements that can be obtained remotely. Remote identification of these surface phenotypes will improve leaf optical modelling and biochemical parameter estimations. Phenotyping of leaf surfaces can inform plant breeding decisions and assist with plant health monitoring.

E-Eye-Based Approach for Traceability and Annuality Compliance of Lentils

In recent years, thanks to their numerous nutritional benefits, legumes have been rediscovered and have attracted interest from many consumers. However, these products, the most valuable ones traditionally produced in smaller communities in particular, can be objects of fraud; this is the case of Italian lentils, which, being a dry product, have a fairly long shelf life, but, due to the minimal visual changes that can affect them, it is possible that expired lentils may be sold alongside edible ones. The present work aims at creating a non-destructive method for classifying Italian lentils according to their harvest year and origin, and for discriminating between expired and edible ones. In order to achieve this goal, Red-Green-Blue (RGB) imaging, which could be considered as a sort of e-eye and represents a cutting-edge, rapid, and effective analytical method, was used in combination with a discriminant classifier (Sequential Preprocessing through ORThogonalization-Linear Discriminant Analysis, SPORT-LDA) to create novel testing models. The SPORT-LDA models built to discriminate the different geographical origins provided an average correct classification rate on the test set of about 88%, whereas an overall 90% accuracy was obtained (on the test samples) by the SPORT-LDA model built to recognize whether a sample was still within its expiry date or not.

Metaheuristic algorithms in visible and near infrared spectra to detect excess nitrogen content in tomato plants

Chemical fertilizers are widely applied in agriculture to achieve high yield, enhance produce quality and build resistance to diseases; in our case the plant being tomato ( Solanum lycopersicum L. var. Royal). However, the acidity, size and taste of tomato fruits could change with excess nitrogen (N) application. The present study aims at the early detection of nitrogen-rich tomato leaves using hyperspectral imaging techniques in the visible and near infrared (Vis-NIR) spectrum, in order to improve plant nutrition composition at an early growth stage. A 30% over-dose of nitrogen was applied to half of the tomato pots. Five leaves were randomly collected from each pot for 3 days (classes D0, D1, D2 and D3), and images were captured with a hyperspectral camera. A metaheuristic approach of artificial neural networks and the firefly algorithm (ANN-FA) was used to determine the most discriminative wavelengths. Afterwards, a combination of ANN and particle swarm optimization (ANN-PSO) was used to classify tomato leaves into the four classes. The training/classification process was repeated 200 times, and results indicated that the proposed approach was able to detect the excess of nitrogen even at the first day (D1), with a precision of 92.9%. Considering all the classes, the average correct classification rate was 92.6%, while the best execution achieved 95.5% accuracy. Thus, the method showed a high performance for practical uses.

Financial distress prediction using integrated Z-score and multilayer perceptron neural networks

The COVID-19 pandemic led to a great deal of financial uncertainty in the stock market. An initial drop in March 2020 was followed by unexpected rapid growth over 2021. Therefore, financial risk forecasting continues to be a central issue in financial planning, dealing with new types of uncertainty. This paper presents a stock market forecasting model combining a multi-layer perceptron artificial neural network (MLP-ANN) with the traditional Altman Z-Score model. The contribution of the paper is presentation of a new hybrid enterprise crisis warning model combining Z-score and MLP-ANN models. The new hybrid default prediction model is demonstrated using Chinese data. The results of empirical analysis show that the average correct classification rate of thew hybrid neural network model (99.40%) is higher than that of the Altman Z-score model (86.54%) and of the pure neural network method (98.26%). Our model can provide early warning signals of a company's deteriorating financial situation to managers and other related personnel, investors and creditors, government regulators, financial institutions and analysts and others so that they can take timely measures to avoid losses.

Multi-Stream Convolutional Neural Network-Based Wearable, Flexible Bionic Gesture Surface Muscle Feature Extraction and Recognition.

Surface electromyographic (sEMG) signals are weak physiological electrical signals, which are highly susceptible to coupling external noise and cause major difficulties in signal acquisition and processing. The study of using sEMG signals to analyze human motion intention mainly involves data preprocessing, feature extraction, and model classification. Feature extraction is an extremely critical part; however, this often involves many manually designed features with specialized domain knowledge, so the experimenter will spend time and effort on feature extraction. To address this problem, deep learning methods that can automatically extract features are applied to the sEMG-based gesture recognition problem, drawing on the success of deep learning for image classification. In this paper, sEMG is captured using a wearable, flexible bionic device, which is simple to operate and highly secure. A multi-stream convolutional neural network algorithm is proposed to enhance the ability of sEMG to characterize hand actions in gesture recognition. The algorithm virtually augments the signal channels by reconstructing the sample structure of the sEMG to provide richer input information for gesture recognition. The methods for noise processing, active segment detection, and feature extraction are investigated, and a basic method for gesture recognition based on the combination of multichannel sEMG signals and inertial signals is proposed. Suitable filters are designed for the common noise in the signal. An improved moving average method based on the valve domain is used to reduce the segmentation error rate caused by the short resting signal time in continuous gesture signals. In this paper, three machine learning algorithms, K-nearest neighbor, linear discriminant method, and multi-stream convolutional neural network, are used for hand action classification experiments, and the effectiveness of the multi-stream convolutional neural network algorithm is demonstrated by comparison of the results. To improve the accuracy of hand action recognition, a final 10 gesture classification accuracy of up to 93.69% was obtained. The separability analysis showed significant differences in the signals of the two cognitive-behavioral tasks when the optimal electrode combination was used. A cross-subject analysis of the test set subjects illustrated that the average correct classification rate using the pervasive electrode combination could reach 93.18%.

Truck body type classification using a deep representation learning ensemble on 3D point sets

Understanding the spatiotemporal distribution of commercial vehicles is essential for facilitating strategic pavement design, freight planning, and policy making. Hence, transportation agencies have been increasingly interested in collecting truck body configuration data due to its strong association with industries and freight commodities, to better understand their distinct operational characteristics and impacts on infrastructure and the environment. The rapid advancement of Light Detection and Ranging (LiDAR) technology has facilitated the development of non-intrusive detection solutions that are able to accurately classify truck body types in detail. This paper proposes a new truck classification method using a LiDAR sensor oriented to provide a wide field-of-view of roadways. In order to enrich the sparse point cloud obtained from the sensor, point clouds originating from the same truck across consecutive frames were grouped and combined using a two-stage vehicle reconstruction framework to generate a dense three-dimensional (3D) point cloud representation of each truck. Subsequently, PointNet – a deep representation learning algorithm – was adopted to train the classification model from reconstructed point clouds. The model utilizes low-level features extracted from the 3D point clouds and detects key features associated with each truck class. Finally, model ensemble techniques were explored to reduce the generalization error by averaging the results of seven PointNet models and further enhancing the overall model performance. The optimal number of models in the ensemble was determined through a comprehensive sensitivity analysis with the consideration of the average correct classification rate (CCR), the variability of the prediction results, and the computation efficiency. The developed model is capable of distinguishing passenger vehicles and 29 different truck body configurations with an average CCR of 83 percent. The average correct classification rate of the developed method on the test dataset was 90 percent for trucks pulling a large trailer(s).

Microbial source tracking of fecal contamination in Laguna Lake, Philippines using the library-dependent method, rep-PCR.

Laguna Lake is an economically important resource in the Philippines, with reports of declining water quality due to fecal pollution. Currently, monitoring methods rely on counting fecal indicator bacteria, which does not supply information on potential sources of contamination. In this study, we predicted sources of Escherichia coli in lake stations and tributaries by establishing a fecal source library composed of rep-PCR DNA fingerprints of human, cattle, swine, poultry, and sewage samples (n = 1,408). We also evaluated three statistical methods for predicting fecal contamination sources in surface waters. Random forest (RF) outperformed k-nearest neighbors and discriminant analysis of principal components in terms of average rates of correct classification in two- (84.85%), three- (82.45%), and five-way (74.77%) categorical splits. Overall, RF exhibited the most balanced prediction, which is crucial for disproportionate libraries. Source tracking of environmental isolates (n = 332) revealed the dominance of sewage (47.59%) followed by human sources (29.22%), poultry (12.65%), swine (7.23%), and cattle (3.31%) using RF. This study demonstrates the promising utility of a library-dependent method in augmenting current monitoring systems for source attribution of fecal contamination in Laguna Lake. This is also the first known report of microbial source tracking using rep-PCR conducted in surface waters of the Laguna Lake watershed.

Transcranial Magnetic Stimulation Indices of Cortical Excitability Enhance the Prediction of Response to Pharmacotherapy in Late-Life Depression

BackgroundOlder adults with late-life depression (LLD) often experience incomplete or lack of response to first-line pharmacotherapy. The treatment of LLD could be improved using objective biological measures to predict response. Transcranial magnetic stimulation (TMS) can be used to measure cortical excitability, inhibition, and plasticity, which have been implicated in LLD pathophysiology and associated with brain stimulation treatment outcomes in younger adults with depression. TMS measures have not yet been investigated as predictors of treatment outcomes in LLD or pharmacotherapy outcomes in adults of any age with depression. MethodsWe assessed whether pretreatment single-pulse and paired-pulse TMS measures, combined with clinical and demographic measures, predict venlafaxine treatment response in 76 outpatients with LLD. We compared the predictive performance of machine learning models including or excluding TMS predictors. ResultsTwo single-pulse TMS measures predicted venlafaxine response: cortical excitability (neuronal membrane excitability) and the variability of cortical excitability (dynamic fluctuations in excitability levels). In cross-validation, models using a combination of these TMS predictors, clinical markers of treatment resistance, and age classified patients with 73% ± 11% balanced accuracy (average correct classification rate of responders and nonresponders; permutation testing, p < .005); these models significantly outperformed (corrected t test, p = .025) models using clinical and demographic predictors alone (60% ± 10% balanced accuracy). ConclusionsThese preliminary findings suggest that single-pulse TMS measures of cortical excitability may be useful predictors of response to pharmacotherapy in LLD. Future studies are needed to confirm these findings and determine whether combining TMS predictors with other biomarkers further improves the accuracy of predicting LLD treatment outcome.

Assessment and Localization of Structural Damage in r/c Structures through Intelligent Seismic Signal Processing

ABSTRACT In this work, a novel approach in post-earthquake structural damage estimation is investigated. The approach is formulated as a problem of both damage approximation and localization. The inter-story drift ratio and the global damage index of Park/Ang (DIG,PA) are the estimated damage indicators for each floor of the structure. Artificial neural networks (ANNs), random forests (RFs), support vector machines (SVMs) with linear and radial basis function (RBF) kernels and adaptive neuro-fuzzy inference systems (ANFISs) are tested to predict the seismic damage state of each floor of an 8-storey reinforced concrete (r/c) building subjected to 155 natural and artificially generated seismic accelerograms. The damage potential of the accelerograms is described by three seismic parameters extracted from the response of the structure. The set of seismic accelerograms is defined by combining two outlier detection techniques, isolation forests and Z-score, while the set of seismic parameters is confirmed by minimum redundancy maximum relevance (mRMR) feature selection algorithm. Optimization methods are used to fine-tune the performance of all networks. Results indicate RFs and ANNs among the models with optimal performances, reaching average correct classification rates of up to 96.87% and 91.87% with RFs, and 96.25% and 90.12% with ANNs, for DIG,PA and ISDR, respectively.

Characterizing building materials using multispectral imagery and LiDAR intensity data

This paper addresses the underlying bottleneck of unknown materials and material characteristics for assessing the life cycle of an existing structure or considering interventions. This is done by classifying and characterizing common building materials with two readily accessible, remote sensing technologies: multispectral imaging and Light Detection and Ranging (LiDAR). A total of 142 samples, including concrete of 3 different water/cement ratios, 2 mortar types, and 2 brick types (each type fired at 3 different temperatures) were scanned using a 5-band multispectral camera in the visible, RedEdge, and Near Infrared range and 2 laser scanners with different wavelengths. A Partial Least Squares Discriminant Analysis model was developed to classify the main materials and then the subcategories within each material type. With multispectral data, an 82.75% average correct classification rate was achieved (improving to 83.07% when combined with LiDAR intensity data), but the effect was not uniformly positive. This paper demonstrates the potential to identify building materials in a non-destructive, non-contact manner to aid in in-situ facade material labelling.

Greenness identification using visible spectral colour indices for site specific weed management

In the present study an attempt has been made to identify the green vegetation based on colour using visible spectral colour indices such as excess green index (ExG), excess red index (ExR) and excess green minus excess red index (ExGR). At first stage, the performance of colour indices were tested at four illumination intensities using the standard colour patches. The results indicated a clear separation between the ExG, ExR and ExGR values of green colour patches (foliage, yellow green & green) and soil colour patches (dark skin, moderate red & magenta) at illumination intensity of 89.04 ± 8.12 lux than 188.8 ± 6.36, 259.25 ± 12.73 and 359.28 ± 10.10 lux illumination intensities. This observation suggested that the colour indices might perform better at low lighting condition. In the second stage, the images of original plants and soil were captured at an illumination intensity of 89.04 ± 8.12 lux and classification rate at different threshold were studied. The average correct classification rate of ExGR and ExG colour indices were found to be 93.03% and 86.03% at threshold values 0 and 10, respectively. This indicates that the colour index ExGR could be successfully employed for image based classification of plant and non-plant material.

Rapid classification of commercial teas according to their origin and type using elemental content with X-ray fluorescence (XRF) spectroscopy

The authenticity of tea has become more important to the industry while the supply chains become complex. The quality and price of tea produced in different regions varies greatly. Currently, a rapid analytical method for testing the geographical origin of tea is missing. XRF is emerging as a screening technique for mineral and elemental analysis with applications in the traceability of foodstuffs, including tea. This study aims to develop a reliable multivariate classification model using XRF spectroscopy to obtain the mineral content. A total of 75 tea samples from tea producing countries throughout the world were analysed. After variable shortlisting, 18 elements were used to construct the multivariate models. Tea origin was determined by classifying the tea into 5 major geographical regions producing most of the global tea. PCA showed initial clustering in some regions, although the types of teas included in the study (black, green, white, herbal) showed no discrete cluster membership. The prediction power of each classification model developed was determined by using two multivariate classifiers, SIMCA and PLS-DA, against an independent validation set. The average overall correct classification rates of PLS-DA models were between 54-85% while the results of SIMCA models were between 70-84% resolving the poor clustering initially shown by PCA. This study demonstrated the potential of geographical origin of tea prediction using elemental contents of tea. Naturally, the classification can be linked not only to origin but to the type of tea as well. Practical applicationWholesalers and retailers need a rapid and robust screening tool to confirm the origin and type of tea they sell to consumers. X-Ray fluorescence spectroscopy proved a good technique for achieving this in commercial teas sourced worldwide. Building on multivariate models, broad classification was accomplished both in terms of origin (Asian vs non-Asian) and in tea type with zero sample preparation and low cost of analysis.

Detection of Mild Cognitive Impairment in an At-Risk Group of Older Adults: Can a Novel Self-Administered Serious Game-Based Screening Test Improve Diagnostic Accuracy?

Background:Literature supports the use of serious games and virtual environments to assess cognitive functions and detect cognitive decline. This promising assessment method, however, has not yet been translated into self-administered screening instruments for pre-clinical dementia.Objective:The aim of this study is to assess the performance of a novel self-administered serious game-based test, namely the Virtual Supermarket Test (VST), in detecting mild cognitive impairment (MCI) in a sample of older adults with subjective memory complaints (SMC), in comparison with two well-established screening instruments, the Montreal Cognitive Assessment (MoCA) and the Mini-Mental State Examination (MMSE).Methods:Two groups, one of healthy older adults with SMC (N = 48) and one of MCI patients (N = 47) were recruited from day centers for cognitive disorders and administered the VST, the MoCA, the MMSE, and an extended pencil and paper neuropsychological test battery.Results:The VST displayed a correct classification rate (CCR) of 81.91% when differentiating between MCI patients and older adults with SMC, while the MoCA displayed of CCR of 72.04% and the MMSE displayed a CCR of 64.89%.Conclusion:The three instruments assessed in this study displayed significantly different performances in differentiating between healthy older adults with SMC and MCI patients. The VST displayed a good CCR, while the MoCA displayed an average CCR and the MMSE displayed a poor CCR. The VST appears to be a robust tool for detecting MCI in a population of older adults with SMC.

Falling Detection Research Based on Elderly Behavior Infrared Video Image Contours Ellipse Fitting

Throughout the world, the proportion of the elders in the total population is increasing dramatically, and home-based care has become the most important form of old-age care. Falling is the most common cause of accidents among the elders at home that poses a huge threat to their health and lives. In order to protect the privacy of the elders an accidental falling detection algorithm for the elders in the home has been proposed in this paper. First, contour-based infrared motion video images are used instead of high-definition cameras to collect the elderly behaviors to protect their privacy. Second, ellipse fitting is performed on the infrared video images of the five behaviors including standing, sitting, squatting, bending and falling. The five geometric characteristic variables of the contour-fitting ellipses including the number of ellipses, centroid positions, ellipsoidal areas, horizontal inclinations and long-short axis ratios of the images, have been extracted. Next, an LSTM model is established using the above variables as inputs for feature extraction and classification. Finally, infrared video images of different types of active behaviors of the elders aged from 50 to 70 years have been selected as IFD database for classification detection. Sixty percent of the IFD images are used as training datasets, and 40% of the IFD images are used as test datasets, and compared with the classification detection of URFD datasets which contains optical RGB HD video images of the different behaviors. The experimental results show the effectiveness of the algorithm proposed in this paper which combines the contour ellipse fitting of the infrared video images and the LSTM feature extraction. The average correct classification rate of the normal and falling down behaviors of the elders is above 95%, which is comparable to the optical RGB datasets. The precision of behavior recognition can effectively protect the privacy of the elders, and provide protection for the accidental falling detection of the elders living alone.

Data driven leakage diagnosis for oil pipelines: An integrated approach of factor analysis and deep neural network classifier

This paper proposes a novel data-based leakage diagnosis method for big datasets, which identifies the leak occurrence, its size, and its location. Different statistical features are used to express the changes in flow and pressure signals at different leakage scenarios. To improve the performances of the leakage diagnosis approach, factor analysis (FA) is employed for dimension reduction purposes. The optimal features of both pressure and flow signals are then fed as input vectors to a deep neural network (DNN) classifier. The proposed leakage diagnosis method has been applied to the first 20 km of the Golkhari-Binak oil pipeline, located in Iran. The leakage isolation accuracy has been compared with some related works. Simulation results show that the proposed method significantly outperforms others with the average correct classification rate (CCR) of about 98%.