Automatic Detection Method Research Articles

An ensemble method for automatic real-time detection and evaluation of oil and gas leakage in subsea pipelines based on 3D real-time sonar system

An ensemble method for automatic real-time detection and evaluation of oil and gas leakage in subsea pipelines based on 3D real-time sonar system

TBINet: fabric defect detection based on a top-down and bottom-up inference network

A key aspect of textile production is the use of automatic defect detection methods for quality control, which is a critical part of the current fabric defect detection system. Benefiting from the explosive development of convolutional neural networks, it has been proved to be feasible to apply it to salient object detection model in fabric defect detection, but how to learn more robust features and better feature fusion for saliency are still complicated tasks. This article presents a novel saliency detection method for fabric defect detection based on both top-down and bottom-up saliency inference combined with two-way information. The top-down process is to infer high-level saliency by gradually using higher-level and richer semantic features on the basis of the backbone. Then a self-fusion enhanced representation module is proposed, which simulates the parallel processing mechanism with residual connection in a top-down manner to generate robust features and effectively characterize the texture features of complex fabrics. In addition, in the down-top process, the interactive feature fusion module is designed, which is used for coarse-to-fine saliency estimation, where the high-level saliency is gradually integrated with finer lower-level features to obtain a fine-grained result. Finally, fabric defect detection is localized by segmenting the generated saliency map. Extensive experimental results on two kinds of fabric image datasets have demonstrated that the proposed TBINet can localize the defect region with high accuracy, and perform favorably against seven state-of-the-art methods on eight widely tested metrics.

Differential evolution-based convolutional neural networks: An automatic architecture design method for intrusion detection in industrial control systems

Industrial control systems (ICSs) are facing serious and evolving security threats because of a variety of malicious attacks. Deep learning-based intrusion detection systems (IDSs) have been widely considered as one of promising security solutions for ICSs, but these deep neural networks for IDSs in ICSs have been designed manually, which are extremely dependent on expert experience with numerous model parameters. This paper makes the first attempt to develop an automatic architecture design method of convolutional neural networks (CNNs) based on differential evolution (abbreviated as DE-CNN) for the intrusion detection issue in ICSs. The first phase of the proposed DE-CNN is the off-line architecture optimization of the CNNs constructed by three basic units such as ResNetBlockUnit, DenseNetBlockUnit, and PoolingUnit, including encoding the architecture parameters of a CNN as a population, evaluating the fitness of the population by the validation accuracy and the number of CNN model parameters, implementing the evolutionary process including mutation and crossover operations, and selecting the best individual from the population. Then, the optimal CNN model obtained by the off-line optimization of DE-CNN is deployed for the online IDSs. The experimental results on two intrusion detection datasets in ICSs including SWaT and WADI have demonstrated the superiority of the proposed DE-CNN to the state-of-the-art manually-designed and neuroevolution-based methods under both unsupervised and supervised learning in terms of Precision, Recall, F1-Score and the number of the CNN model parameters.

Acoustic-Signal-Based Damage Detection of Wind Turbine Blades-A Review.

Monitoring and maintaining the health of wind turbine blades has long been one of the challenges facing the global wind energy industry. Detecting damage to a wind turbine blade is important for planning blade repair, avoiding aggravated blade damage, and extending the sustainability of blade operation. This paper firstly introduces the existing wind turbine blade detection methods and reviews the research progress and trends of monitoring of wind turbine composite blades based on acoustic signals. Compared with other blade damage detection technologies, acoustic emission (AE) signal detection technology has the advantage of time lead. It presents the potential to detect leaf damage by detecting the presence of cracks and growth failures and can also be used to determine the location of leaf damage sources. The detection technology based on the blade aerodynamic noise signal has the potential of blade damage detection, as well as the advantages of convenient sensor installation and real-time and remote signal acquisition. Therefore, this paper focuses on the review and analysis of wind power blade structural integrity detection and damage source location technology based on acoustic signals, as well as the automatic detection and classification method of wind power blade failure mechanisms combined with machine learning algorithm. In addition to providing a reference for understanding wind power health detection methods based on AE signals and aerodynamic noise signals, this paper also points out the development trend and prospects of blade damage detection technology. It has important reference value for the practical application of non-destructive, remote, and real-time monitoring of wind power blades.

A High-Precision Plant Disease Detection Method Based on a Dynamic Pruning Gate Friendly to Low-Computing Platforms.

Timely and accurate detection of plant diseases is a crucial research topic. A dynamic-pruning-based method for automatic detection of plant diseases in low-computing situations is proposed. The main contributions of this research work include the following: (1) the collection of datasets for four crops with a total of 12 diseases over a three-year history; (2) the proposition of a re-parameterization method to improve the boosting accuracy of convolutional neural networks; (3) the introduction of a dynamic pruning gate to dynamically control the network structure, enabling operation on hardware platforms with widely varying computational power; (4) the implementation of the theoretical model based on this paper and the development of the associated application. Experimental results demonstrate that the model can run on various computing platforms, including high-performance GPU platforms and low-power mobile terminal platforms, with an inference speed of 58 FPS, outperforming other mainstream models. In terms of model accuracy, subclasses with a low detection accuracy are enhanced through data augmentation and validated by ablation experiments. The model ultimately achieves an accuracy of 0.94.

Interactive interface for spatio-temporal mapping of epileptic human brain using characteristics of high frequency oscillations (HFOs)

Spontaneous High Frequency Oscillations (HFOs) have been considered as emerging specific biomarkers of the epileptogenic region. As a first issue, a significant difference in the implementation of automatic HFOs detection methods can sometimes occur between researchers. In addition, clinicians are not even particularly familiar with the concept of signal and image processing, and programming skills. To overcome these limitations, we propose a plug-and-play interactive Graphical User Interface (GUI) that incorporates an amalgamation of six validated methods used for detecting and quantifying of HFOs events. As a second issue, the most automated HFOs detection methods to date have a high false detection rate and low specificity, ranging, in some cases up to 80% and below 37% respectively. Therefore, the eventual utilization of HFOs detection algorithms in clinical settings requires a checking step to save clinically relevant HFOs and remove spurious oscillations from the detection results. As a last issue addressed in the present study, the major previous HFOs studies have been limited only to the detection and classification of HFOs, but only a few studies have been conducted to efficiently follow the neural dynamics of epileptic focus by studying HFOs characteristics through different brain regions and clinical stages. Therefore, in our software, the brain mapping of HFOs characteristics is done based on the duration, the inter-duration, the average frequency, and the power of HFOs. The present developed software may be considered helpful for understanding the functional significance of HFOs and also to reduce the interaction gap between fundamental research and applied clinical practice related to HFOs.

Smart Iterative Analysis Tool for the Size Distribution of Spherical Nanoparticles.

The size of nanoparticles is a critical parameter with regard to their performance. Therefore, precise measurement of the size distribution is often required. While electron microscopy (EM) is a useful tool to image large numbers of particles at once, manual analysis of individual particles in EM images is a time-consuming and labor-intensive task. Therefore, reliable automatic detection methods have long been desired. This paper introduces a novel automatic particle analysis software package based on the circular Hough transform (CHT). Our software package includes novel features to enhance precise particle analysis capabilities. We applied the CHT algorithm in an iterative workflow, which ensures optimal detection over wide radius intervals, to deal with overlapping particles. In addition, smart intensity criteria were implemented to resolve common difficult cases that lead to false particle detection. Implementing these criteria enabled an effective and precise analysis by minimizing detection of false particles. Overall, our approach showed reliable particle analysis results by resolving common types of particle overlaps and deformation with only negligible errors.

Automatic cloud detection method for optical remote sensing images combining channel attention mechanism and multi-feature fusion

Cloud detection is a crucial step in remote sensing image pre-processing, but there are still issues such as cloud miss detection and false detection. Therefore, this work suggests a multi-feature fusion and channel attention mechanism combined approach for automatically recognizing clouds. Based on DeepLabV3+, to make the feature extraction faster and more accurate, a lightweight network structure MobileNetV2 is used as the backbone network. To effectively distinguish snow and clouds, channel attention is added to the low-level semantic information, which makes the network more able to focus on essential features. Meantime, mid-level semantic features are added to the decoder module to make use of more abundant features, thus reducing the false detection of mist. Experimental findings on the GF1 WHU dataset demonstrate that the method’s detection accuracy is improved by 2.1% when compared to DeepLabV3+, and it performs well in the cloud detection job of optical remote sensing images.

Hyperparameter Optimization for Transfer Learning-based Disease Detection in Cassava Plants

Cassava is quite possibly the most widely recognized staple food crop. It is a nutty-flavored, starchy root vegetable that is a primary energy source and carbs for individuals. During crop cultivation, cassava plant infections can influence the leaf and root, bringing about a tremendous loss to the harvest and financial market esteem. Hence, it is vital to detect diseases in cassava plants. But it requires enormous labor, longer time planning, and thorough plant-specific knowledge. If disease detection is possible at the initial stages, then actions can be taken on time. Hence, there is a need to develop automatic detection methods for monitoring different parts of cassava plants. This study evaluates the efficiency of applying transfer learning to the pre-trained models for identifying diseases in cassava plants. The pre-trained EfficientNet model detects the disorders using data augmentation, fine-tuning the hyperparameters, cross-validation, and transfer learning. The experimentation is done with the cassava dataset provided by Kaggle, which contains cassava plant leaf images belonging to five classes. An experimental investigation shows that EfficientNet with transfer learning attains up to 89% accuracy. The effect of transfer learning is significant; consider getting the results of high accuracy and less dispersion; in very few cases, the model forecasts the wrong class labels. The outcomes give a promising strength to the objective of this work, i.e., a model trained explicitly for agriculture with transfer learning can assist the farmers with highly accurate results during farming to get a high yield.

Investigation on the transition of a propagating planar flame

A novel method to find the burning velocity and stretch rate of a premixed flame out of flame surface area has been developed by an image processing technique. The study is tailored for the control of precarious burning during combustion based on an automatic flame detection method. The flame propagation of premixed LPG-Air mixture is carried out in a tube of 30 mm diameter at the equivalence ratio of 1.2. A visible transition of tulip flame is observed towards the finger-shaped flame during initial propagation, which is envisaged to be triggered by the hydrodynamic instability. However, the flame is driven by stabilizing effect of diffusion and stretch. The flame structure is experimentally captured by an image processing method using Python 3.6 software to analyse the burning behaviour. The laminar burning velocity of an unstretched propagating flame has been found in congruence with the non-propagating steady flame at exit. The laminar burning velocity of steady flame is calculated by the volumetric flame formation rate. Based on the study of flame stretch rates, and flame propagation velocities at stationary burned conditions, the burning velocity of axially propagating laminar planar flame is found to be closer agreement with the higher stretched radially propagating spherical flame. The Markstein number is found to be 20.734, which divulges the presence of significant intrinsic diffusional effects in planar flame than the spherical flame during transition. The laminar and turbulent burning velocities are obtained to be accurate, respectively, at different equivalence ratios and turbulent intensities. The transition is also achieved and compared numerically by the Flamelet Generated Manifold method. The laminar burning velocity of LPG flamelet is observed to be precise at 75% propane composition and ϕ≤ 1.09. A higher laminar burning has been ascertained in lean mixtures due to the longer transition time.

Quantitative Detection of Microplastics in Water through Fluorescence Signal Analysis

Microplastics (MPs) have recently been acknowledged as a new major and ubiquitous environmental pollutant with still unclear, yet potentially high, risks for different ecosystems and human health. Nevertheless, quantitative identification protocols rely on long and subjective visual counting necessarily performed on microscopes by well-trained operators. In this study, an automatic, fast, portable, and inexpensive method for the quantitative detection of MPs in water is proposed. The system is based on the typical optical setup of a fluorescence confocal microscope specifically adapted to automatically count dye-stained MPs in flowing liquids using a low-power laser beam. The fluorescence pulses emitted by flowing MPs are revealed and processed by a specific software using a pattern recognition algorithm to discriminate and count real fluorescence pulses out of noise fluctuations. The system was calibrated with commercial orange fluorescent 10 µm and 1 µm polystyrene microspheres, and remarkable agreement with theoretical predictions was obtained regarding different parameters. Tests were also performed with laboratory-prepared MPs dispersed in different types of real water samples. In this case, the agreement with theory was slightly worse and differences found in the quantitative results require further investigation. However, the present study demonstrated the proof of concept of a method for quick automated MP counting in water.

From face detection to emotion recognition on the framework of Raspberry pi and galvanic skin response sensor for visual and physiological biosignals

The facial and physiological sensor-based emotion recognition methods are two popular methods of emotion recognition. The proposed research is the first of its kind in real-time emotion recognition that combines skin conductance signals with the visual-based facial emotion recognition (FER) method on a Raspberry Pi. This research includes stepwise documentation of method for automatic real-time face detection and FER on portable hardware. Further, the proposed work comprises experimentation related to video induction and habituation methods with FER and the galvanic skin response (GSR) method. The GSR data are recorded as skin conductance and represent the subject's behavioral changes in the form of emotional arousal and face emotion recognition on the portable device. The article provides a stepwise implementation of the following methods: (a) the skin conductance representation from the GSR sensor for arousal; (b) gathering visual inputs for identifying the human face; (c) FER from the camera module; and (d) experimentation on the proposed framework. The key feature of this article is the comprehensive documentation of stepwise implementation and experimentation, including video induction and habituation experimentation. An illuminating aspect of the proposed method is the survey of GSR trademarks and the conduct of psychological experiments. This study is useful for emotional computing systems and potential applications like lie detectors and human–machine interfaces, devices for gathering user experience input, identifying intruders, and providing portable and scalable devices for experimentation. We termed our approaches "sensovisual" (sensors + visual) and "Emosense" (emotion sensing).

Acoustic animal identification using unsupervised learning

Abstract Passive acoustic monitoring is usually presented as a complementary approach to monitoring wildlife communities and assessing ecosystem conditions. Automatic species detection methods support biodiversity monitoring and analysis by providing information on the presence–absence of species, which allows understanding the ecosystem structure. Therefore, different alternatives have been proposed to identify species. However, the algorithms are parameterized to identify specific species. Analysing multiple species would help to monitor and quantify biodiversity, as it includes the different taxonomic groups present in the soundscape. We present an unsupervised methodology for multi‐species call recognition from ecological soundscapes. The proposal is based on a clustering algorithm, specifically the learning algorithm for multivariate data analysis (LAMDA) 3pi algorithm, which automatically suggests the number of clusters associated with the sonotypes. Emphasis was made on improving the segmentation of the audio to analyse the whole soundscape without parameterizing the algorithm according to each taxonomic group. To estimate the performance of our proposal, we used four datasets from different locations, years and habitats. These datasets contain sounds from the four major taxonomic groups that dominate terrestrial soundscapes (birds, amphibians, mammals and insects) in audible and ultrasonic spectra. The methodology presents performances between 75% and 96% in presence–absence species recognition. Using the clusters proposed by our methodology, the whole soundscape biodiversity was measured and compared with the estimate of four acoustic indices (ACI, NP, SO and BI). Our approach performs biodiversity assessments similar to acoustic indices with the advantage of providing information about acoustic communities without the need for prior knowledge of the species present in the audio recordings.

An efficient online peak detection algorithm for synchronized intestinal electrical stimulation and its application for treating diabetes.

Obesity is one of leading risk factors for type 2 diabetes and other types of chronic diseases. Synchronized intestinal electrical stimulation (SIES) has been explored for treating obesity and diabetes. In SIES, electrical stimulation is delivered to the small intestine in synchronization with the intrinsic intestinal myoelectrical activity (its basic rhythm is called slow wave) and therefore, the accurate detection of intestinal slow waves is critically important for SIES. The aim of this study is to detect the peaks in intestinal slow waves in real-time based on the automatic multiscale peak detection (AMPD) method. In this paper, we introduce an efficient technique for real-time detection of peaks in intestinal slow waves. The presented method is based on peak estimation of a given quasi-periodic signal using the AMPD method. This method uses a multi-scale approach to identify the peaks of the intestinal slow waves with high detection accuracy and a minimal delay. Throughout the experiments, the multi-scale technique is used to estimate the quasi-periodic signals using different signal-to-noise ratio, λ (optimal scale), and the "lag" β (number of datapoints for right hand estimation) as important performance factors. The performance of the presented method is also calculated and utilized in the comparison process for 10 datasets of the intestinal slow waves from rats at λ = 150ms and two values of β = 100ms and 150ms. The experimental results show that the presented method has good overall accuracy for online peak detection while maintaining low memory and computational complexity. Numerically, the overall accuracy is above 90%, and 98% for the rodent intestinal slow waves at a time-lag of 150ms. The developed SIES system has been applied to successfully reduce postprandial blood glucose in a rodent model of hyperglycemia. In conclusion, the developed algorithm is adequate for on-line peak detection of the intestinal slow waves; the SIES method used the developed peak detection algorithm which is effective in reducing postprandial blood glucose in a rodent model of hyperglycemia.

Corporate Sustainability Communication as ‘Fake News’: Firms’ Greenwashing on Twitter

Fake news on social media has engulfed the world of politics in recent years and is now posing the same threat in other areas, such as corporate social responsibility communications. This study examines this phenomenon in the context of firms’ deceptive communications concerning environmental sustainability, usually referred to as greenwashing. We first develop and validate a new method for automatically identifying greenwashing, using linguistic cues in a sample of tweets from a diverse set of firms in two highly polluting industries. We then examine the relationship between greenwashing and financial market performance for the firms in our sample. Prior research has identified these issues as some of the most important gaps in the extant literature. By addressing them, we make several important contributions to corporate sustainability research and practice, as well as introducing notable improvements to automatic greenwashing detection methods.

SE_SPnet: Rice leaf disease prediction using stacked parallel convolutional neural network with squeeze‐and‐excitation

AbstractRice is one of the significant crops, and the early identification and prevention of its diseases are essential to ensure adequate and healthy availability to the world's growing population. The use of image processing is an encouraging method for automatic rice leaf disease identification and detection. In particular, the recent advancements indicate the effectiveness of convolutional neural network (CNN) based deep learning approaches. In this direction, the present work proposes a novel stacked parallel convolution layers‐based network (SPnet) with the squeeze‐and‐excitation (SE) architecture, named (SE_SPnet), for classifying diseased rice leaf images. The stacked parallel network block comprises four parallel convolution layers with different kernel sizes for abstractions of the global and local features. The SE block extracts feature information automatically while removing invalid ones. We compare the SE_SPnet model with state‐of‐the‐art CNN models such as VGG16, DenseNet121, and InceptionV3 based on computational effort, accuracy, sensitivity, specificity, precision, recall, and F1‐score. The experimental results show that the SE_SPnet outperforms standard CNN models for the considered rice leaf disease image datasets. In particular, the SE_SPnet achieves the highest accuracy (99.2%), sensitivity (98.2%), specificity (98.5%), precision (98.4%), recall (98.2%), and F1‐score (98.5%) while using stochastic gradient descent (with momentum) optimizer with a 0.01 learning rate. Furthermore, the SE_SPnet also exhibits to outperform when compared with some of the most recent and relevant existing works.

Automated leakage detection method of pipeline networks under complicated backgrounds by combining infrared thermography and Faster R-CNN technique

Leakage detection is essential to process safety and loss prevention of pipeline networks. As one of the attractive methods for detecting leakages in single pipelines, infrared thermography (IRT) faces new challenges for leakage detection of pipeline networks. Compared with the single pipeline, the pipeline network has a complicated structure with higher topology features and diverse material composition. In addition, the environment of the pipeline network is complex and changeable. These factors increase the complexity of the infrared (IR) thermal images of pipeline networks, which makes the leakage features too weak to be detected via naked-eye observation in an efficient and precise manner. To address this issue, this study develops an automated detection method for pipeline network leakages under complicated background, which is achieved by combining IRT with the Faster Region-based Convolutional Neural Network (Faster R-CNN) technique. First, the pipeline network system in the complicated background is designed for IR thermal image acquisition and data analysis. Then, an automated leakage detection model is built based on Faster R-CNN, which uses a modified VGG16 network for feature extraction for its outstanding performance in the feature extraction of small objects under complicated backgrounds. Finally, the high efficiency and precision of the proposed method for automated detection of pipeline network leakages in complicated backgrounds are verified by extensive experiments. Meanwhile, the generalization ability of the proposed method is verified by conducting experiments under various experimental conditions (different moments all-day, different viewing angles of IR camera, pedestrian disturbing). In general, the proposed method provides a promising way to detect pipeline network leakages in complicated backgrounds.

Automatic Detection Method of Sewer Pipe Defects Using Deep Learning Techniques

Regular inspection of sewer pipes can detect serious defects in time, which is significant to ensure the healthy operation of sewer systems and urban safety. Currently, the widely used closed-circuit television (CCTV) inspection system relies mainly on manual assessment, which is labor intensive and inefficient. Therefore, it is urgent to develop an efficient and accurate automatic defect detection method. In this paper, an improved method based on YOLOv4 is proposed for the detection of sewer defects. A significant improvement of this method is using the spatial pyramid pooling (SPP) module to expand the receptive field and improve the ability of the model to fuse context features in different receptive fields. Meanwhile, the influence of three bounding box loss functions on model performance are compared based on their processing speed and detection accuracy, and the effectiveness of the combination of DIoU loss function and SPP module is verified. In addition, to address the lack of datasets for sewer defect detection, a dataset that contains 2700 images and 4 types of defects was created, which provides useful help for the application of computer vision techniques in this field. Experimental results show that, compared with the YOLOv4 model, the mean average precision (mAP) of the improved model for sewer defect detection are improved by 4.6%, the mAP can reach 92.3% and the recall can reach 89.0%. The improved model can effectively improve the detection and classification accuracy of sewer defects, and has significant advantages compared with other methods.

Outlier galaxy images in the Dark Energy Survey and their identification with unsupervised machine learning

The Dark Energy Survey is able to collect image data of an extremely large number of extragalactic objects, and it can be reasonably assumed that many unusual objects of high scientific interest are hidden inside these data. Due to the extreme size of DES data, identifying these objects among many millions of other celestial objects is a challenging task. The problem of outlier detection is further magnified by the presence of noisy or saturated images. When the number of tested objects is extremely high, even a small rate of noise or false positives leads to a very large number of false detections, making an automatic system impractical. This study applies an automatic method for automatic detection of outlier objects in the first data release of the Dark Energy Survey. By using machine learning-based outlier detection, the algorithm is able to identify objects that are visually different from the majority of the other objects in the database. An important feature of the algorithm is that it allows to control the false-positive rate, and therefore can be used for practical outlier detection. The algorithm does not provide perfect accuracy in the detection of outlier objects, but it reduces the data substantially to allow practical outlier detection. For instance, the selection of the top 250 objects after applying the algorithm to more than 2⋅106 DES images provides a collection of uncommon galaxies. Such collection would have been extremely time-consuming to compile by using manual inspection of the data.

Automatic deep learning method for detection and classification of breast lesions in dynamic contrast-enhanced magnetic resonance imaging.

The purpose of this study was to develop a deep learning-based system with a cascade feature pyramid network for the detection and classification of breast lesions in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). This retrospective study enrolled 191 consecutive patients with pathological confirmed breast lesions who underwent preoperative magnetic resonance imaging (MRI) at the Second Affiliated Hospital of Xi'an Jiaotong University. Patients were randomly divided into a training set comprising 153 patients with 126 malignant and 27 benign lesions and a validation set containing 38 patients with 31 malignant and 7 benign lesions under 5-fold cross-validation. Two radiologists annotated the location and classification of all lesions. After augmentation with pseudo-color image fusion, MRI images were fed into the developed cascade feature pyramid network system, feature pyramid network, and faster region-based convolutional neural network (CNN) for breast lesion detection and classification, respectively. The performance on large (>2 cm) and small (≤2 cm) breast lesions was further evaluated. Average precision (AP), mean AP, F1-score, sensitivity, and false positives were used to evaluate different systems. Cohen's kappa scores were calculated to assess agreement between different systems, and Student's t-test and the Holm-Bonferroni method were used to compare multiple groups. The cascade feature pyramid network system outperformed the other systems with a mean AP and highest sensitivity of 0.826±0.051 and 0.970±0.014 (at 0.375 false positives), respectively. The F1-score of the cascade feature pyramid network in real detection was also superior to that of the other systems at both the slice and patient levels. The mean AP values of the cascade feature pyramid network reached 0.779±0.152 and 0.790±0.080 in detecting large and small lesions, respectively. Especially for small lesions, the cascade feature pyramid network achieved the best performance in detecting benign and malignant breast lesions at both the slice and patient levels. The deep learning-based system with the developed cascade feature pyramid network has the potential to detect and classify breast lesions on DCE-MRI, especially small lesions.