Manual Detection Method Research Articles

Intelligent Recognition of Transmission Line Inspection Image Based on Deep Learning

Under the background of high-speed development of the power grid, the traditional manual detection method is low efficiency and high cost, which are not able to meet the current requirements. With the development of UAV technology, the method of using UAV to inspect transmission lines has a positive effect on improving detection efficiency and quality. In recent years, deep learning has developed rapidly, which provides a new solution for the analysis and processing of UAV aerial inspection images. This paper focuses on the deep learning algorithm to study the insulator location and fault identification in aerial photographs, which plays an important role in realizing the automation and intelligence of UAV detection.

Automatic Detection of Grammatical Errors in English Verbs Based on RNN Algorithm: Auxiliary Objectives for Neural Error Detection Models.

With the rapid development of neural network technology, we have widely used this technology in various fields. In the field of language translation, the research on automatic detection technology of English verb grammatical errors is in a hot stage. The traditional manual detection cannot be applied to the current environment. Therefore, this paper proposes an automatic detection technology of English verb grammatical errors based on recurrent neural network (RNN) algorithm to solve this problem. Firstly, the accuracy and feedback speed of traditional manual detection and recurrent neural network RNN algorithm are compared. Secondly, a detection model which can be calculated according to grammatical order combined with context is designed. Finally, when the output verb result is inconsistent with the original text, it can automatically mark the error detection effect. The experimental results show that the algorithm model studied in this paper can effectively improve the detection accuracy and feedback efficiency and is more applicable and effective than the traditional manual detection method.

Fabric Defect Detection based on Improved Object as Point

In the field of fabric manufacturing, many factories still utilise the traditional manual detection method. It requires a lot of labour, resulting in high error rates and low efficiency. In this paper, we represent a realtime automated detection method based on object as point. This work makes three attributions. First, we build a fabric defects database and augment the data to training the intelligence model. Second, we provide a real-time fabric defects detection algorithm, which have potential to be applied in manufacturing. Third, we figure out CenterNet with soft NMS will improved the performance in fabric defect detection area, which is considered an NMS-free algorithm. Experiment results indicated that our lightweight network based method can effectively and efficiently detect five different fabric defects.

Severity Recognition of Aloe Vera Diseases Using AI in Tensor Flow Domain

Agriculture plays an important role in the economy of all countries. However, plant diseases may badly affect the quality of food, production, and ultimately the economy. For plant disease detection and management, agriculturalists spend a huge amount of money. However, the manual detection method of plant diseases is complicated and time-consuming. Consequently, automated systems for plant disease detection using machine learning (ML) approaches are proposed. However, most of the existing ML techniques of plants diseases recognition are based on handcrafted features and they rarely deal with huge amount of input data. To address the issue, this article proposes a fully automated method for plant disease detection and recognition using deep neural networks. In the proposed method, AlexNet and VGG19 CNNs are considered as pre-trained architectures. It is capable to obtain the feature extraction of the given data with fine-tuning details. After convolutional neural network feature extraction, it selects the best subset of features through the correlation coefficient and feeds them to the number of classifiers including K-Nearest Neighbor, Support Vector Machine, Probabilistic Neural Network, Fuzzy logic, and Artificial Neural Network. The validation of the proposed method is carried out on a self-collected dataset generated through the augmentation step. The achieved average accuracy of our method is more than 96% and outperforms the recent techniques.

Deriving forest plot inventory parameters using terrestrial laser scanning in the tropical rainforest of Malaysia

ABSTRACT Forest inventory parameters, in particular DBH (diameter at breast height) and tree height, can be used to accurately estimate above-ground carbon, which is an important indicator of forest productivity and sequestration of carbon. This study demonstrates and tests a method to automatically and manually derive forest inventory parameters for estimating above-ground carbon (AGC) using a Terrestrial Laser Scanner (TLS). The study was conducted in the Royal Belum State Park of Peninsular Malaysia by establishing 24 circular inventory sample plots of 0.05 ha. A RIEGL VZ-400 TLS system was used to acquire point cloud data of the sample plots through multiple scanning, and these data were further pre-processed and co-registered using the RiSCAN PRO software. The DBH and tree height for each tree within the plots were manually measured using a distance measurement tool in RiSCAN PRO and automatically derived from Computree software using a tree segmentation approach. The inventory parameters derived from TLS were compared with the field measurements for calculating the AGC using an allometric equation. On average, 89% and 90% of the sampled trees were detected from the point cloud data of the plots using the manual and automatic detection methods, respectively. The mean values of R 2 (coefficient of determination) and RMSE (Root Mean Square Error) for manually measured DBH and tree height across the plots were 0.95 and 2.70 cm, and 0.77 and 2.96 m, respectively. We also obtained an average value of 0.86 and 2.47 cm, and 0.51 and 3.15 m for R 2 and RMSE, respectively between field-measured and automatically derived DBH and tree height from TLS data across all the sample plots of the study area. Our method provides clear evidence that TLS has the potential to derive forest inventory parameters, which can be used to estimate above-ground carbon in the tropical rainforest accurately.

Estimation of Forest Variable and Aboveground Biomass using Terrestrial Laser Scanning in the Tropical Rainforest

Estimation of accurate forest variable is crucial for forest monitoring and quantifying forest aboveground biomass (AGB). The objective of this study was to estimate forest variables and AGB using the TLS data acquired over the tropical rainforest of Malaysia. Individual trees were detected using the manual detection method in the RiSCAN PRO software. An average tree detection rate of 99.55% was achieved from the 10 sample plots. Forest variables, including DBH and height, were measured from 10 circular sample plots in the field. The accuracy of diameter at breast height (DBH) of trees measured from TLS was validated using field DBH as reference. A root means square error (RMSE) of 1.37 cm (6.60%) was obtained for manually measured TLS DBH. Similarly, TLS-based tree height was validated using airborne laser scanner height as a reference and resulted in RMSE of 1.74 m (9.30%). Finally, AGB was calculated using the variables derived from the TLS data. The result has shown an R2 value of 0.98 and RMSE of 0.077 Mg for AGB calculated from TLS data. The result of this study confirmed that TLS can provide a very good estimation of forest variables and AGB in the tropical dense rainforest.

Identification and Detection of Surface Defects of Outer Package Printed Matter Based on Machine Vision

In the printing process, the surface of the outer package printed matter is prone to defects such as chromatic aberration and misting, leading to a defective product. It is necessary to take an effective method to identify and detect surface defects in printed matter. In this paper, the application of machine vision technology in the detection of defects in outer package was studied. Firstly, a machine vision image acquisition device based on charge coupled device (CCD) camera was designed; secondly, the obtained image was processed by graying the average method, then OTSU binarization was performed after denoising by median filtering, and finally an improved differential matching method was designed for defect detection. The example analysis showed that the clear defect image of the outer package could be obtained by using the method proposed in this paper. The detection rate of the method in detecting 2,000 images reached 99.4%, and the average detection time was 103 ms. In the detection of 5,000 images, the detection rate of the manual detection method was 81%, but the detection rate of the method proposed in this paper was 99.1%. The experimental results proved the reliability of the method and provided some theoretical basis for the further application of defect detection technology based on machine vision in the print industry, which was conducive to the good development of the print industry.

Modified DenseNet for Automatic Fabric Defect Detection With Edge Computing for Minimizing Latency

As an essential step in quality control, fabric defect detection plays an important role in the textile manufacturing industry. The traditional manual detection method is inaccurate and incurs a high cost; as a result, it is gradually being replaced by deep learning algorithms based on cloud computing. However, a high data transmission latency between end devices and the cloud has a significant impact on textile production efficiency. In contrast, edge computing, which provides services near end devices by deploying network, computing and storage facilities at the edge of the Internet, can effectively solve the above-mentioned problem. In this article, we propose a deep-learning-based fabric defect detection method for edge computing scenarios. First, this article modifies the structure of DenseNet to better suit a resource-constrained edge computing scenario. To better assess the proposed model, an optimized cross-entropy loss function is also formulated. Afterward, six feasible expansion schemes are utilized to enhance the data set according to the characteristics of various defects in fabric samples. To balance the distribution of samples, proportions of various defect types are used to determine the number of enhancements. Finally, a fabric defect detection system is established to test the performance of the optimized model used on edge devices in a real-world textile industry scenario. Experimental results demonstrate that compared with the conventional convolutional neural network (CNN), the proposed optimized model attains an average improvement of 18% in the area under the curve (AUC) metric for 11 defects. Data transmission is reduced by approximately 50% and latency is reduced by 32% in the Cambricon 1H8 platform compared with a cloud platform.

Forest Inventory and Aboveground Biomass Estimation with Terrestrial LiDAR in the Tropical Forest of Malaysia

An accurate forest inventory is crucial for forest monitoring and quantifying forest aboveground biomass (AGB). This study aimed to investigate the feasibility of Terrestrial Laser Scanning (TLS) in forest inventory and AGB estimation in the tropical forest of Malaysia. Individual trees were detected using manual and automatic detection methods. An average tree detection rate of 99.55% and 93.75% were achieved using the manual and automatic detection method respectively. The accuracy of the diameter at breast height (DBH) of trees measured from TLS was validated using field DBH as reference. A root means square error (RMSE) of 1.37 cm (6.60%) and 2.36 cm (11.47%), respectively, were obtained for manually and automatically measured TLS DBH. Similarly, TLS based tree height was validated using Airborne Laser Scanner (ALS) height as a reference and resulted in RMSE of 1.74 m (9.30%) and 3.17 m (17.40%) with manual and automatic method respectively. Finally, AGB was calculated using the variables derived from the TLS data. Results show an R2 value of 0.98 and RMSE of 0.08 Mg. The results of this study confirmed that TLS as a nondestructive approach can provide a very good estimation of forest attributes and AGB in the dense tropical forest conditions.

Three-dimensional thermohaline anomaly structures of rings in the Kuroshio Extension region

Using AVISO satellite altimeter observations during 1993–2015 and a manual eddy detection method, a total of 276 anticyclonic rings and 242 cyclonic rings shed from the Kuroshio Extension (KE) were identified, and their three-dimensional (3D) anomaly structures were further reconstructd based on the Argo float data and the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) cruise and buoy data through an interpolation method. It is found that the cyclonic (anticyclonic) rings presented consistent negative (positive) anomalies of potential temperature; meanwhile the relevant maximum anomaly center became increasingly shallow for the cyclonic rings whereas it went deeper for the anticyclonic rings as the potential temperature anomaly decreased from the west to the east. The above deepening or shoaling trend is associated with the zonal change of the depth of the main thermocline. Moreover, the composite cold ring between 140° and 150°E was found to exhibit a double-core vertical structure due to the existence of mode water with low potential vorticity. Specifically, a relatively large negative (positive) salinity anomaly and a small positive (negative) one appeared for the composite cyclonic (anticyclonic) ring at the depth above and below 600 m, respectively. The underlying driving force for the temperature and salinity anomaly of the composite rings was also attempted, which varies depending on the intensity of the background current and the temperature and salinity fields in different areas of the KE region, and the rings’ influences on the temperature and salinity could reach deeper than 1 000 m on average.

Road Crack Detection Using Deep Neural Network with Receptive Field Block

Cracks are common pavement diseases that affect pavement performance. To maintain the road in good condition, localizing and fixing the cracks is a vital responsibility for transportation maintenance department. However, traditional manual detection methods are considerably tedious and require domain expertise. Therefore, the research on automatic detection and identification of pavement crack is of great significance for ensuring traffic safety and pavement maintenance decisions. In this paper, we propose an automatic pavement crack detection network based on the Single Shot MultiBox Detector(SSD) deep learning framework, and introduce the receptive field module to enhance the feature extraction capability of the network, which ensures real-time crack detection and also improves the performance of accuracy in pavement crack detection.

Automatic Tunnel Crack Detection Based on U-net and a Convolutional Neural Network with Alternately Updated Clique.

Regular crack inspection of tunnels is essential to guarantee their safe operation. At present, the manual detection method is time-consuming, subjective and even dangerous, while the automatic detection method is relatively inaccurate. Detecting tunnel cracks is a challenging task since cracks are tiny, and there are many noise patterns in the tunnel images. This study proposes a deep learning algorithm based on U-Net and a convolutional neural network with alternately updated clique (CliqueNet), called U-CliqueNet, to separate cracks from background in the tunnel images. A consumer-grade DSC-WX700 camera (SONY, Wuxi, China) was used to collect 200 original images, then cracks are manually marked and divided into sub-images with a resolution of pixels. A total of 60,000 sub-images were obtained in the dataset of tunnel cracks, among which 50,000 were used for training and 10,000 were used for testing. The proposed framework conducted training and testing on this dataset, the mean pixel accuracy (MPA), mean intersection over union (MIoU), precision and F1-score are 92.25%, 86.96%, 86.32% and 83.40%, respectively. We compared the U-CliqueNet with fully convolutional networks (FCN), U-net, Encoder–decoder network (SegNet) and the multi-scale fusion crack detection (MFCD) algorithm using hypothesis testing, and it’s proved that the MIoU predicted by U-CliqueNet was significantly higher than that of the other four algorithms. The area, length and mean width of cracks can be calculated, and the relative error between the detected mean crack width and the actual mean crack width ranges from −11.20% to 18.57%. The results show that this framework can be used for fast and accurate crack semantic segmentation of tunnel images.

Magnetic Ring Multi-Defect Stereo Detection System Based on Multi-Camera Vision Technology.

Magnetic rings are the most widely used magnetic material product in industry. The existing manual defect detection method for magnetic rings has high cost, low efficiency and low precision. To address this issue, a magnetic ring multi-defect stereo detection system based on multi-camera vision technology is developed to complete the automatic inspection of magnetic rings. The system can detect surface defects and measure ring height simultaneously. Two image processing algorithms are proposed, namely, the image edge removal algorithm (IERA) and magnetic ring location algorithm (MRLA), separately. On the basis of these two algorithms, connected domain filtering methods for crack, fiber and large-area defects are established to complete defect inspection. This system achieves a recognition rate of 100% for defects such as crack, adhesion, hanger adhesion and pitting. Furthermore, the recognition rate for fiber and foreign matter defects attains 92.5% and 91.5%, respectively. The detection speed exceeds 120 magnetic rings per minutes, and the precision is within 0.05 mm. Both precision and speed meet the requirements of real-time quality inspection in actual production.

Abnormal railway fastener detection using minimal significant regions and local binary patterns

Railway fasteners are an important part of the railway system. Keeping the fasteners effective is essential to ensuring the safe operation of railways, so abnormal railway fastener detection is an important task in railway maintenance. With the development of the railway system, the traditional manual fastener detection method has been unable to meet the application requirements because it is very slow, costly, and dangerous. In this paper, we propose what we believe to be a novel method for abnormal fastener detection based on computer vision, which can detect missing and ectopic fasteners automatically. In this method, the minimal significant region is extracted in order to improve the fastener localization accuracy. Then, fastener recognition is operated using local binary features and a support vector machine classifier based on the fastener sub-images that are obtained by fastener localization. The proposed method is evaluated in our own database, which is obtained by a railway inspection system in different environments. The experimental results have shown improved performance against the state-of-the-art algorithm.

AUTOMATIC DAMAGE DETECTION OF STONE CULTURAL PROPERTY BASED ON DEEP LEARNING ALGORITHM

Abstract. Outdoor stone cultural properties are continuously affected by the external environment such as wind, rain, and earthquakes. These cause damage to the cultural properties by not only threatening structural stability but also damaging the aesthetic value. Quick detection of these damages is important to enable appropriate preservation treatment in terms of cultural property conservation management. Even though conventional manual damage detection methods are widely used, they are limited by manpower, cost, and other external conditions. In this paper, we propose a system that automatically detects and classifies damage occurring in cultural properties using deep-learning technique to settle these drawbacks. In detail, the damages are classified into four types (i.e., crack, loss, detachment, biological colonization) based on Faster region-based convolutional neural network (R-CNN) algorithm. In addition, we construct an image dataset of stone damage, which is collected by the regular report of the National Designated Cultural Property in 2017 conducted by the Cultural Heritage Administration of S. Korea, and augment its dataset to enhance damage detection performance. From the experiment conducted, we achieved an average confidence score of 94.6 % or more on the 20 test images.

Linear Seam Elimination of Tunnel Crack Images Based on Statistical Specific Pixels Ratio and Adaptive Fragmented Segmentation

Image processing basis crack detection methods can overcome the deficiency of traditional manual and instrument basis crack detection methods, and can provide an important guiding for damage evaluation. However, for tunnel lining surface, cracks and linear seams have great similarities in both intensity value and texture features, it is difficult for existing crack extraction methods to obtain accurate crack segmentation results. In this paper, we proposed a novel algorithm to adaptively eliminate linear seams in tunnel lining crack images. By analyzing characteristics of linear seams and cracks, the idea of binning is used to classify those detected line edges with disordered directions into multiple angle subintervals. Then, by calculating the ratio of statistical specific pixels (SSP ratio) on the expanded line edge, the length and binning information are used to select linear seam edges with the use of adaptive expansion algorithm. Finally, the fragmented segmentations of linear seams are adopted so that cracks and linear seams can be clearly separated, and linear seams can be removed. The experimental results demonstrate the superior precision and efficiency of our method compared with existing methods.

High-speed Railway Fastener Detection Using Minima Significant Region and Local Binary Patterns

Railway fastener is an important part of the railway system. Keeping the fasteners effective is essential to ensure the safe operation of the railway, so abnormal railway fastener detection is a main task of railway maintenance. With the development of railway system, the traditional manual fastener detection method has been unable to meet the application requirements, because it is very slow, costly, and dangerous. In this paper, we propose a novel method for abnormal fastener detection based on computer vision, which can detect missing fasteners automatically. In this method, Minima Significant Region is extracted in order to improve the fastener localization accuracy. Then, fastener recognition is operated using local binary features and Support Vector Machine classifier based on the fastener sub-images which are obtained by fastener localization. The proposed method is evaluated in our own database which is obtained by railway inspection system in different environments. The experimental results have shown improved performance against the state-of-the-art algorithm.

MJaya-ELM: A Jaya algorithm with mutation and extreme learning machine based approach for sensorineural hearing loss detection

Sensorineural hearing loss (SNHL) is a common hearing disorder or deafness which accounts for about 90% of the reported hearing loss. Magnetic resonance imaging (MRI) has been found to be an effective neuroimaging technique for detecting SNHL. However, manual detection methods, mainly based on the visual inspection of MRI, are cumbersome, time-consuming and need skilled supervision. Hence, there is a great need to design a computer-aided detection system for fast, accurate and automated detection of SNHL. This paper presents a new method for automated diagnosis of SNHL through brain MR images. Fast discrete curvelet transform is employed for image decomposition. The features are extracted from various decomposed subbands at different scales and orientations. A set of discriminant features is then derived using PCA+LDA algorithm. A hybrid classifier is suggested by integrating extreme learning machine and Jaya optimization with mutation (MJaya-ELM) to distinguish hearing loss images from healthy MR images. The proposed hybrid method overcomes the drawbacks of traditional ELM and other learning algorithms for single layer feedforward neural network. The concept of mutation is introduced to conventional Jaya optimization (MJaya) for improving the global search ability of the solutions by providing additional diversity. The proposed system is evaluated on a well-studied database. The comparison results demonstrate that the proposed scheme outperforms the existing schemes in terms of overall accuracy and sensitivity over different classes. The effectiveness of the proposed MJaya-ELM algorithm is also compared with its counterparts such as PSO-ELM, DE-ELM, and Jaya-ELM, and the results indicate the superiority of MJaya-ELM.

Design Method of Fabric Density Sensor Based on the Virtual Grating With Gradual Constant

In order to overcome the shortcomings of time-consuming and inefficient manual detection methods of fabric density, a fast and efficient automatic detection method of fabric density based on the digital grating is proposed. In order to realize automatic fabric density detection, the design method of the digital grating with gradual constant is described in detail on the basis of the principle of moiré pattern. The moiré pattern containing yarn density information of the fabric can be obtained by superimposing the captured fabric image and the digital grating image. After the moiré pattern is projected, the fabric density can be accurately determined by analyzing the distribution characteristics of projection curve. In the detection process, a digital grating image generated by computer is used to replace the physical grating, which not only makes the detection faster, but also make the system applicable for the fabric density detection of all kinds of specifications. Testing experiments were carried out on various fabric samples including fabrics in a plain weave and in a twill weave, the detection accuracy is more than 99.2% and the detection speed is much faster than the existing methods, which can meet the requirements of on-line detection in fabric production.

Quality inspection method of micro-nano parts based on deep learning

There are some disadvantages such as low efficiency, high work intensity by using the manual methods to detect the quality of micro-nano parts because of the characteristics such as small size and fragile structure. Considering about the disadvantages, computer microscopic vision is introduced into the detection system in this paper, which can collects the image information of the parts into the computer system efficiently. The parts to be detected are transmitted by the spin material platform driven by the stepping motor. It is CNN based on deep learning that used to detect the surface quality and classify the defects of the parts according to the image information in this paper, which can improve the accuracy of the detection and reduce the work intensity of human compared with not only the traditional manual detection methods but also some edge detection methods that former researchers used.