Actual Inspection Research Articles

Optimization and Improvement of Three-phase Asynchronous Motor Soft Starter Control Circuit

Abstract The soft starters of ABB company are used in a pumped storage power station. Sometimes, a phase absent fault was reported during the process of the motor starting and then led to starting failure. To solve this problem, this article first explained the working principle of the device, and then analyzed the design of the control circuit and the characteristics of the components. Combined with the actual inspection results to find the crux of the problem, It proposed optimization and improvement methods for the soft starter control circuit, and achieved the desired results.

Comparative analysis of different guidelines for risk classification, safety assessment and monitoring of existing bridges in Italy: A16 viaducts case-studies

Bridges are vital components of any existing infrastructure, enabling essential links for transportation and commerce. In Italy, a nation renowned for its historical bridges, the importance of proper risk classification, safety assessment and monitoring of existing structures is of paramount importance. Risk classification involves categorizing bridges into different classes based on their potential risks and deterioration levels. The fundamental purpose of risk classification is consistent across guidelines: to prioritize bridge inspections and maintenance efforts based on risk levels. Once bridges are classified according to risk, safety assessments and continuous monitoring are performed to evaluate their structural integrity and overall long-term safety. The methodologies for safety assessment can differ significantly among guidelines, highlighting the influence of regional factors, such as construction materials, design practices, and environmental conditions. However, the common goal is to ensure the safe operation of existing bridges by identifying deficiencies and prescribing necessary interventions. Monitoring techniques can range from visual inspections to advanced sensor networks and non-destructive testing methods. In Italy various guidelines and standards have been developed to address these concerns, to ensure the safety and longevity of its bridges. The actual guidelines inspections for existing bridges include those by the Italian Ministry of Infrastructure and Transport (MIT), which plays a central role in setting the national guidelines and regulations for bridge safety in Italy. MIT is responsible for overseeing the safety and maintenance of publicly-owned bridges in Italy, providing a comprehensive framework for risk classification, considering factors such traffic loads, material degradation and level of defectiveness. On the other hand, other manual exist, such those provided by Autostrade per l’Italia S.p.A. (ASPI), a private company responsible for the management and maintenance of Italy’s motorway network. Both MIT and ASPI follow a risk-based approach to prioritize inspections and maintenance efforts on bridges, however, some differences and variations characterised the approaches. While the specifics may differ, the overarching goal of ensuring bridge safety and preventing catastrophic failures remains consistent. This paper aims to provide a comparative analysis of different guidelines for the risk classification, safety assessment, and monitoring of existing bridges, considering several A16 viaducts as case-studies, made of prestressed concrete, highlighting the key similarities and differences among them. The comparative analysis reveals important insight aimed to continue refining and adapting these guidelines and manuals to best serve the safety of existing bridges and the communities they support.

Feature-based domain disentanglement and randomization: A generalized framework for rail surface defect segmentation in unseen scenarios

Deep neural network has demonstrated high-level accuracy in rail surface defect segmentation. However, deploying these deep models in actual inspection situations results in generalizability deficits and accuracy degradation. This phenomenon is mainly caused by the appearance difference between training and test images. To alleviate this issue, we propose a feature-based domain disentanglement and randomization (FDDR) framework to improve the generalization of deep models in unseen datasets. Specifically, two encoders are introduced to decompose the defect image into domain-invariant structural features and domain-specific style features. Only the domain invariant features are used to identify the defects. Additionally, we design a shuffle whitening module to remove the style information from the domain-invariant features. Meanwhile, the extracted style features are used to train a style variational autoencoder to randomly generate novel defect styles. Then, the randomly generated style features are combined with the domain-invariant features to obtain new defect images, thus expanding the training sample. We validate the proposed FDDR framework in six defect segmentation datasets. Extensive experimental results show that FDDR demonstrates robust defect segmentation performance in unseen scenarios and outperforms other state-of-the-art domain generalization methods. The source code will be released at https://github.com/Rail-det/FDDR.

Electricity theft recognition and time period detection considering based on multi-task learning

Electricity theft is an economic problem faced around the power companies. After the suspicious electricity theft users are recognized, the potentially stolen electricity (PSE) and the loss caused by the electricity theft needs to be accurately evaluated. In the actual inspection work, the key to evaluating PSE is electricity theft time period detection. However, most of the current electricity theft detection algorithms can only recognize electricity theft, which cannot detect the specific time period of electricity theft simultaneously. Therefore, this paper proposes a deep model named multi-task deep residual network (MDRN), which can simultaneously recognize electricity theft users and detect their electricity theft time period. The MDRN is constructed based on one-dimensional convolutional and residual network, which can effectively extract characteristics from power consumption data. To automatically balance the multiple tasks in training, a joint multi-task loss with task uncertainty is proposed. The experimental results based on the Irish dataset show that the proposed multi-task model obtains the highest Accuracy with 93.17% in electricity theft recognition and the highest intersection-over-union (IOU) with 76.58% in time period detection. It should be noted that the proposed method can be directly used to calculate the PSE to maximize economic return of electricity theft inspection.

Data Augmentation of X-ray Images for Automatic Cargo Inspection of Nuclear Items.

As part of establishing a management system to prevent the illegal transfer of nuclear items, automatic nuclear item detection technology is required during customs clearance. However, it is challenging to acquire X-ray images of major nuclear items (e.g., nuclear fuel and gas centrifuges) loaded in cargo with which to train a cargo inspection model. In this work, we propose a new means of data augmentation to alleviate the lack of X-ray training data. The proposed augmentation method generates synthetic X-ray images for the training of semantic segmentation models combining the X-ray images of nuclear items and X-ray cargo background images. To evaluate the effectiveness of the proposed data augmentation technique, we trained representative semantic segmentation models and performed extensive experiments to assess its quantitative and qualitative performance capabilities. Our findings show that multiple item insertions to respond to actual X-ray cargo inspection situations and the resulting occlusion expressions significantly affect the performance of the segmentation models. We believe that this augmentation research will enhance automatic cargo inspections to prevent the illegal transfer of nuclear items at airports and ports.

Detection of prohibited and restricted object X-ray images based on Hi color space segmentation.

X-ray imaging plays an important role in security inspection. However, the objects are complex, which makes it difficult to automatically detect prohibited and restricted objects. This study aims to develop and test a detection method based on a new image segmentation scheme to solve the problem of detecting prohibited and restricted objects from pseudo-color X-ray images with complex backgrounds. The internal mechanism of the influence of different color spaces on image segmentation effect is explored, and the color space component Hi is studied. Furthermore, the mechanism of the new Hi component and the influence law of its adjustable coefficient are revealed. Additionally, a detection method based on Hi color space segmentation for pseudo-color X-ray images is proposed. The segmentation and detection methods are then tested on actual X-ray images. The results show that hue has the greatest influence on image segmentation effect of the pseudo-color X-ray images. For different pseudo-color X-ray images with complex backgrounds, applying the proposed new Hi color space segmentation method achieves overall accuracy of 0.974 and 1.0 in detecting the gun and knife, respectively. The new X-ray image detection method based on the Hi color space segmentation proposed in this paper enables to better solve the complex background problem including object overlap and adhesion and thus more effectively meet the requirements of actual security inspection.

Electromagnetic Ultrasonic Shear-Horizontal Wave to Detect Corrosion Defect of Flat Steel for Grounding Device of Transmission Pole Tower

Detecting defects in grounded flat steel is essential for ensuring the safety and reliability of transmission tower grounding devices. However, traditional inspection methods, such as physical excavation and verification, are costly and time-consuming. This paper proposes a corrosion defect detection method for flat steel transmission tower grounding devices based on electromagnetic ultrasonic SH waves. In addition, using commercial software, a three-dimensional finite element simulation model of grounded flat steel with simulated pitting corrosion defects is constructed. The specified displacements applied to multiple surface sources mimic the horizontal shear vibrations generated by the electromagnetic ultrasonic transducer on the surface of the grounded flat steel during actual inspection. A simulation was used to investigate the propagation and attenuation characteristics of shear-horizontal ultrasonic SH0guided waves for simulated corrosion defects with various geometric configurations in grounded flat steel. The simulation investigated the propagation and attenuation characteristics of the SH0 wave in grounded flat steel and the detection of various defects for linear analysis of the results. The simulation results show that the attenuation of the electromagnetic ultrasonic guided wave is small, at only 0.0016 dB/mm, and the displacement amplitude of the echo signal decreases with the increase of the SH0 wave propagation distance. Increasing the depth and length of corrosion defects increases the echo signal amplitude. At the same time, the width of corrosion defects has little effect on the echo amplitude. Finally, a flat steel defect detection experiment was conducted, and the experimental results fit with the simulation to verify the accuracy of the simulation model. This detection method introduces a new idea for the on-site detection and quantitative identification of corrosion defects in grounded flat steel, which has significant reference value and can provide a more effective and economical method for ensuring the safety and dependability of transmission tower grounding devices.

Fatigue life prediction of orthotropic steel deck strengthened with UHPC under stochastic traffic load

In recent years, the ultra-high performance concrete (UHPC) has been increasingly used to strengthen orthotropic steel decks (OSD) to solve the cracking problems at fatigue-prone details and pavement damage. In this paper, the fatigue life of a cable-stayed orthotropic steel decks bridge under stochastic traffic loads is calculated before and after the orthotropic steel decks strengthened with the ultra-high performance concrete layer. The traffic data of the real bridge for 1 week is first obtained based on the weigh-in-motion system. Then, a stochastic traffic load on the bridge is simulated for its service life by the Monte Carlo method. A fatigue life analysis framework, which includes the traffic load simulation, a refined finite element model, the S-N curve and Miner linear cumulative damage criterion, is proposed for fatigue life prediction of orthotropic steel decks. For the bridge before reinforcement, the predicting results for the fatigue life of three fatigue-prone details, including the scallop cutout, rib-to-diagram and rib-to-deck joint are basically consistent with that of the actual bridge inspection results. After strengthening by ultra-high performance concrete, the fatigue life of the three structural details are increased from 15.87, 13.89, and 32.26 years to more than 100 years, respectively, as compared with the original orthotropic steel decks structure.

Integrating the root cause analysis to machine learning interpretation for predicting future failure

The research proposes a new model for evaluating offshore pipelines due to corrosion. The existing inspection method has an inherent limitation in reusing the primary root cause analysis data to forecast the potential loss and corrosion mitigation, particularly in the scope of data utilization. The study implements Artificial Intelligence to transfer the knowledge of failure analysis as a consideration for conducting the inspection and lowering the risk of failure. This work combines experimental and modelling methodologies to assert the actual and feasible inspection method. The elemental composition, hardness, and tensile tests are utilized to unveil the types of corrosion products and metallic properties. Scanning Electronic Microscope and Energy Dispersive X-Ray (SEM-EDX) and X-Ray Diffractometer (XRD) was utilized to assess the corrosion product and their corresponding morphology to reveal the corrosion mechanism. The Gaussian Mixture Model (GMM), aided by the Pearson Multicollinear Matrix, shows the typical risk and predicts the damage mechanism of the spool to suggest the types of mitigation scenarios for the pipeline's longevity. According to the laboratory result, the wide and shallow pit corrosion and channelling are evident. The result of the tensile and hardness test confirms the types of the API 5 L X42 PSL 1 standard material. The SEM-EDX and XRD provide a piece of clear evidence into the corrosion product are primarily due to CO2 corrosion. The silhouette score agrees well with the results of the Bayesian information criterion of GMM to show three different risk levels low, medium, and high-risk profiles. The combination of injection of chemicals such as parasol, biocide and cleaning pigging are a few solutions to address CO2 corrosion. This work can be used as a guideline for assessing and clustering the risk based on the risk-based inspection.

PHM-IRNET: Self-training thermal segmentation approach for thermographic inspection of industrial components

Automating thermographic analysis for industrial and construction inspection is one of the essential aspects of NDT 4.0 in industrial transformation. Thermal segmentation is a preliminary step for characterizing a component’s heat signatures by detecting and extracting thermal patterns. Manual segmentation of areas with distinctive thermal patterns is a time-consuming task that consequently can increase the operation cost. Thus, automatic thermal segmentation can significantly assist companies in avoiding possible misinterpretation and decrease the operation cost. Due to the lack of enough labeled thermal images to train deep neural networks and the fact that, by nature, thermal abnormalities may have non-uniform patterns and form in different shapes and sizes, training a deep neural network can be a challenging task. Moreover, selecting common features can cause possible convergence onto dependency toward specific shapes or patterns. Thus, this paper proposes a self-training thermal segmentation method involving a novel process pipeline for using labeled and unlabeled data for training the presented deep-learning model. The method uses an unsupervised approach for predicting the ground truth in case of unlabeled data before passing the data to the secondary model for segmentation. For this paper, three experiments were conducted to evaluate the method’s performance and provide a comparative analysis with the most recent techniques. The first experiment used a simulated inspection created by COMSOL software. The second experiment employed several plate inspections in several isolated environments. Finally, the third experiment used thermal images acquired during an actual inspection of the piping structure and aerospace components.

Labor Standards Enforcement Program and Code of Ethics of Labor Inspectors in Region XII

The study focused on the labor standards enforcement program and code of ethics practices among the labor inspectors of the Department of Labor and Employment in Region 12. The study utilized primary data through one-on-one interviews coming from the Labor Inspectors of the Department of Labor and Employment Region 12, specifically in areas of South Cotabato, Sultan Kudarat, North Cotabato, Sarangani, General Santos City, Regional Office, and TSI. A total of twenty-two participants, which are Region 12 Labor Inspectors, answered the qualitative questionnaire, with the data gathered to be analyzed through thematic analysis. The results show that the Labor Standard Enforcement Program under the Department of Labor and Employment is an effective program and tool in the enforcement of Labor Standards laws, statutes, and orders through Proper Implementation and actual Inspection; however, they have an insufficient workforce. In terms of Labor Inspectors’ performance, they adhere to norms and conduct, yet they also commit violations. Lastly, the norms of conduct that are commonly ignored by the Labor Inspectors are they become biased, receive gifts in exchange for a favor, and they are more into “sarili muna” than “bayan muna”.

Overlapped tobacco shred image segmentation and area computation using an improved Mask RCNN network and COT algorithm.

The classification of the four tobacco shred varieties, tobacco silk, cut stem, expanded tobacco silk, and reconstituted tobacco shred, and the subsequent determination of tobacco shred components, are the primary tasks involved in calculating the tobacco shred blending ratio. The identification accuracy and subsequent component area calculation error directly affect the composition determination and quality of the tobacco shred. However, tiny tobacco shreds have complex physical and morphological characteristics; in particular, there is substantial similarity between the expanded tobacco silk and tobacco silk varieties, and this complicates their classification. There must be a certain amount of overlap and stacking in the distribution of tobacco shreds on the actual tobacco quality inspection line. There are 24 types of overlap alone, not to mention the stacking phenomenon. Self-winding does not make it easier to distinguish such varieties from the overlapped types, posing significant difficulties for machine vision-based tobacco shred classification and component area calculation tasks. This study focuses on two significant challenges associated with identifying various types of overlapping tobacco shreds and acquiring overlapping regions to calculate overlapping areas. It develops a new segmentation model for tobacco shred images based on an improved Mask region-based convolutional neural network (RCNN). Mask RCNN is used as the segmentation network's mainframe. Convolutional network and feature pyramid network (FPN) in the backbone are replaced with Densenet121 and U-FPN, respectively. The size and aspect ratios of anchors parameters in region proposal network (RPN) are optimized. An algorithm for the area calculation of the overlapped tobacco shred region (COT) is also proposed, which is applied to overlapped tobacco shred mask images to obtain overlapped regions and calculate the overlapped area. The experimental results showed that the final segmentation accuracy and recall rates are 89.1% and 73.2%, respectively. The average area detection rate of 24 overlapped tobacco shred samples increases from 81.2% to 90%, achieving high segmentation accuracy and overlapped area calculation accuracy. This study provides a new implementation method for the type identification and component area calculation of overlapped tobacco shreds and a new approach for other similar overlapped image segmentation tasks.

Handling Class Imbalance and Class Overlap in Machine Learning Applications for Undeclared Work Prediction

Undeclared work is a composite socioeconomic matter severely affecting the welfare of workers, legitimate companies, and the state by issuing unfair competition in the labour market and causing considerable state revenue losses by tax evasion. Labour inspectorates are tasked to deal effectively with this issue but usually lack adequate resources and proper tools, yet they own large volumes of past inspection data that, if aptly processed through innovative machine learning techniques, may produce understandable insights into the extent and prevailing patterns of undeclared work and efficient tools to address it. Such datasets are typically imbalanced regarding undeclared work, and contain overlapping inspection discoveries, two issues that impede the learning process. This research points to the problems of class imbalance and class overlap in this domain and applies combinations of data engineering techniques to address them using a dataset of 16.7 K actual labour inspections. Three associative classification algorithms are employed, and multiple classifiers are built and assessed for their predictability and interpretability. The study indicates the overall benefits for the inspection authorities when integrating machine learning methods in targeting undeclared work and proves considerable prediction performance improvement when following data engineering approaches to address the class imbalance and class overlap issues.

Data-driven intelligent method for detection of electricity theft

Anti-electricity leakage or stealing plays a crucial role in the energy market as accurate detection is beneficial for electricity safety, unit commitment, reduction of corporate losses, and so on. In a contemporary competitive energy market, it can significantly improve the efficiency of electric power enterprises and reduce the operational cost of the power system. In this study, through collecting historical data and analyzing the characteristics of electricity consumption, we established the index system for the detection of electricity leakage or stealing. Based on this, a focal loss-based 1d densely connected convolutional network, which we term the FLB-DCNNs, is proposed for the detection of electricity leakage or stealing. The electricity theft samples obtained in the actual on-site inspection are researched in the empirical analysis, and the results prove the effectiveness of the proposed approach. It attains an average precision of 98.51% for detecting electricity leakage or stealing users, and the average recall rate also reaches 98.17%. The recommended procedure provides a new idea for electricity theft detection and is easily transplanted to other related fields. Our data and code are available at https://github.com/xtu502/electricity-theft-detection.

A Crack Detection Method for Pipelines Using Wavelet-Based Decision-Level Data Fusion

A novel crack detection method using wavelet-based decision-level data fusion is proposed and verified by simulations and experiments. In this work, we established a crack signal model using wavelet functions that fits all crack signal scales without need of windowing, and showed that the signal-to-noise-ratio (SNR) difference between the independent signals to be fused has a significant effect on the overall detection performance. Based on this observation, we designed four wavelet-based decision-level data fusion rules. We then presented a detection method where wavelet processing results of individual NDT input are accepted or rejected based on said rules to produce optimized estimation accuracy. To evaluate the proposed method, the first group of simulations were implemented to show the proposed method identifies the inner and outer-surface cracks with good estimation accuracy; the second and third groups of experiments showed the proposed method does improve upon individual detection methods alone, and has better detection performance than three state-of-the-art methods; finally, the pull-rig experiments verified on our own pipeline inspection gauge (PIG) that the proposed method does improve the detection probability beyond that of individual detection methods, in an actual high-speed pipeline inline inspection (ILI).

Internal Defect Detection of Overhead Aluminum Conductor Composite Core Transmission Lines With an Inspection Robot and Computer Vision

Overhead aluminum conductor composite core (ACCC) transmission lines are extensively used. However, internal defects of ACCC wires are difficult to detect, threatening the stability and security of the grid. Thus, a novel automatic detection system using an X-ray inspection robot and an anchor-free object detection model is proposed to solve the problem of detecting internal defects in ACCC wires. First, a new inspection robot with a nondestructive testing (NDT) system consisting of a digital radiography (DR) detection panel and a portable X-ray generator is developed to acquire X-ray images of ACCC wires. Then, the IN-ACCC dataset is created by collecting the X-ray images of artificial defective ACCC wires and then processing, classifying, and labeling the images. Finally, an anchor-free object detection model named CenterNet-NDT is proposed based on CenterNet for high-performance identification of internal defects. CenterNet-NDT has a specially designed feature fusion module composed of SPPCSPC, polarized self-attention (PSA), and a newly weighted bidirectional feature pyramid network named SOFPN. Compared with some state-of-the-art methods and CenterNet with different modules, the proposed CenterNet-NDT achieves the highest mAP of 90.60% on the IN-ACCC dataset. The proposed automatic internal defect detection system is verified to be effective and robust by lab experiments and has been repeatedly applied in actual ACCC transmission line inspection tasks to reduce the safety hazards of wire breakage.

The Application of Augmented Reality Technology in Urban Greening Plant Growth State Detection

The current target detection network in deep learning has been widely used in plant growth state detection. However, with the development of deep learning, within the field of plant growth state detection, the performance of the detection network is no longer the primary factor limiting the detection accuracy and model generalization ability. The construction of high-quality and large-scale plant datasets is more significant for the improvement of model detection accuracy and generalization ability. However, traditional methods for building deep learning datasets for plants have a large time span and low efficiency. And it is difficult to construct and expand the dataset for plants with complex growth environments and difficult image acquisition by existing methods. To address this problem, this paper proposes a method for constructing plant datasets based on augmented reality techniques. The method proposed in this paper allows for the rapid and efficient construction of large-scale field datasets that match the actual inspection environment in the lack of data. Meanwhile, this paper proposes an automatic annotation method for datasets in conjunction with the imaging environment in virtual space. In this paper, we experimentally compare the proposed method with the method of expanding the dataset using GAN networks. Using the virtual dataset constructed by the method proposed in this paper as the training set, the trained YOLOv5 model achieves an average accuracy (@0.5:0.95) of 0.71 for the three detection categories on the test set. The detection accuracy of the six mixed datasets constructed using the two data expansion methods on the test set was experimentally tested. The proposed method in this paper improved the accuracy by 2.2%, 3.1%, and 7.0%, respectively. The smaller the percentage of real images, the greater the accuracy improvement. Experiments show that the method proposed in this paper can well solve the problems faced in the field of plant growth state detection, such as the lack of data, and provides a new idea for the production and expansion of datasets in plant detection tasks.

Comparative Analysis of the Models of Constitutional Review in Portugal and in Poland

The article critically analyses the models of the constitutionality inspection adopted in Portugal and in Poland. For this purpose the comparative method has been applied, comparing Portuguese and Polish approaches. Theoretical aspects of the constitutionality inspection are discussed and the notions of the centralised and the diffuse inspection model are introduced. Then the paper analysis the actual inspection practices in both countries discussing the diffuse and concentrated inspection separately. Constitutional rules pertaining to the direct effect of the constitutions, constitutional jurisdiction of judges and operation of the constitutional courts are considered. The article concludes that in both countries different models of constitutionality inspection operate. In Portugal operates the mixed constitutionality inspection model while in Poland, despite views advocating for permissibility of the diffuse control, the centralised model functions.

Swedish dairy farmers’ perceptions of animal welfare inspections

Farmers today must comply with animal welfare legislation and often one or more private standards. This makes the number of regulations quite comprehensive and the control arena more complex, with several different animal welfare inspections and audits taking place at farm level. This study investigated perceptions among dairy farmers in Sweden of the official animal welfare inspections and the private Arla and KRAV audits, both separately and in relation to each other, and examined associated factors that could potentially influence their perceptions. An electronic questionnaire was developed and answered by 216 Swedish dairy farmers during spring 2021. The respondents in general acknowledged the need for animal welfare inspections, but had rather diverse perceptions of the actual inspections and inspectors, reporting both positive and negative experiences. They reported more negative experiences of official and Arla inspections than of KRAV (organic farming) inspections and most did not believe that inspections had improved animal welfare on their farm. Most of the respondents called for better coordination between the different inspection types. Most farmers reported being very worried before an official or Arla inspection, which was related to their more negative perceptions of these inspections. Other factors associated with farmers’ perception of the inspections were e.g., acceptance of a regulation, the perceived necessity of an inspection, satisfaction with the inspector’s competence, manner, and behavior, perceived fairness of treatment, and whether non-compliances were recorded. The farmers perceived official inspections as more negative if the inspector was a young woman, if there was more than one inspector present, and if the inspection was not pre-announced. These findings indicate a need for objective and accurate communication regarding different animal welfare regulations and inspections. To achieve greater trust and lower concerns among farmers about animal welfare inspections, all stakeholders need to be involved in communication.

Real‐time automatic crack detection method based on drone

AbstractReal‐time automated drone‐based crack detection can be used for efficient building damage assessment. This paper proposes an automated real‐time crack detection method based on a drone. Using a lightweight classification algorithm, a lightweight segmentation algorithm, a high‐precision segmentation algorithm, and a crack width measurement algorithm, the cracks are classified, roughly segmented, finely segmented, and the maximum width is extracted. A crack information‐assisted drone flight automatic control algorithm for automatic crack detection guides the drone toward the crack position. The effectiveness of the crack detection algorithm and the crack information‐assisted drone flight automatic control algorithm was tested on two different datasets, a two‐story building, and a 16‐m‐high shaking table test building. The results show that crack detection can be finished in real‐time during the flight. Using the proposed method can significantly improve the MIoU of crack edge detection and the accuracy of maximum crack width measurement under the non‐ideal shooting conditions of the actual inspection situation by automatically approaching the vicinity of the crack.