Manual Visual Inspection Research Articles

An Approach to Checking Map Tiles by Annotation Recognition

Abstract. As the capabilities of geographic information services increase, web geographic information services represented by internet mapping have become increasingly popular, widely used in various location-based services. It is precisely because of the surge in demand that higher requirements have been put forward for the updating and quality control of internet map data. The quality inspection of geographic vector data is relatively well-developed, with many inspection rules and automated processes, but there is a clear deficiency in the quality inspection of map tiles, which mainly relies on manual visual inspection, restricting the rapid updating and release of map tiles. To solve this problem, this paper proposes an automated map tile inspection method. First, web crawling technology is used to capture map tiles, and to avoid the problem of incomplete display of annotations across tiles, adjacent tiles are all captured and merged into one image; second, OCR technology is used for image text detection and text recognition, and finally, the newly added recognition results are stored in the database and compared and confirmed with the results in the database. The experiment proves that the method proposed in this paper is feasible and may greatly improve work efficiency, especially as the updating speed of online maps accelerates, an automated tile quality inspection method can greatly improve the updating speed.

Tower and power line segmentation method based on RandLA-Net

Abstract The scale of the power grid of China has gradually expanded, and the number of long-distance and high-level transmission lines has also increased. However, it is very hard to detect the power infrastructure transmission lines because of the complex laying environment and various ground features. Lines and towers in the power system are important components of power transmission, and their safety directly affects the operation of the entire power system. Reasonable management of lines and towers can reduce the occurrence of faults and system safety risks. At present, both manual inspection efficiency and visual inspection processing accuracy are low. An airborne LiDAR point cloud has become a key means for monitoring towers and power lines. In response to the above problems, we propose a tower and power line segmentation algorithm based on RandLA-Net. Experimental results demonstrate that the proposed method gets precise detection of power poles and lines, and enhances the efficiency in identifying hazardous areas along power lines. The highest Mean Intersection over Union (mloU) recorded is 95.34%, establishing a clear correlation between power towers and lines. Power line data holds significant importance for the power sector, facilitating the monitoring and management of transmission lines.

Hierarchical robot learning method for industrial fluorescent penetrant inspection

Fluorescent Penetrant Inspection (FPI) is a Non-Destructive Testing (NDT) method, extensively used to evaluate components for identifying defects across a broad range of industries. FPI process remains a manual visual inspection, where the operator by means of fluorescent dye, that penetrates discontinuities on the component, aims to distinguish between indications that are relevant (i.e., can be associated with surface defects) and non-relevant (i.e., can be associated to insufficient wash-off, dust or other non-relevant factors). The FPI process can be decomposed into the following steps: (a) thorough visual examination of the component, (b) executing manual wiping-off of the fluorescent dye with a brush, of all areas that require interrogation for potential indications, and (c) disposition of the inspected components. The number of those areas on the component that require interrogation and hence to be wiped-off is unknown a priori of the inspection and varies depending on the condition of the part. As a result, replacing this manual wipe-off step by a robot requires tedious manual programming of an excessive number of robot paths to assure reach of the robot to the entire surface of the part as well as safe robot motion. In addition, these robot motions are part specific and thus not transferrable to other geometries of components, making scaling of this technology across manufacturing industry not possible. In this paper, we propose a hierarchical robot learning method to address the challenge of reducing manual robot programming and enable the scaling of this automated NDT technology. The proposed method integrates and fuses Deep Reinforcement Learning (DRL), Screw Linear Interpolation (ScLERP) and Learning from Demonstration (LfD), enabling an autonomous generation of brushing strokes with a six-degrees of freedom (DoF) industrial manipulator, and automating the wipe-off step of the FPI process. Using this approach, a robot learning policy is generated for the wipe-off motion in a simulated industrial robotic cell at first and then the policy is transferred to the real system for validation.

Research on gear flank surface defect detection of automotive transmissions based on improved YOLOv8

Abstract In industrial production, the manufacturing processes may introduce defects on the gear flanks of transmission gears, potentially leading to premature failures and diminished performance. The early detection and precise assessment of surface defects on transmission gear flanks are critical for maintaining the safety, reliability, and cost-effectiveness of automobiles. At present, the principal approach for identifying defects on automotive transmission gear flanks predominantly involves manual visual inspections, supplemented by fluorescent magnetic particle testing. However, this approach suffers from low accuracy and efficiency. Consequently, this paper presents a defect detection algorithm that leverages an enhanced YOLOv8 model to facilitate the efficient detection of surface defects on automotive transmission gear flanks. Initially, the collected image data underwent data augmentation and exploratory analysis, which informed targeted enhancements. Subsequently, the YOLOv8 algorithm was thoroughly examined. The SPPE architecture was incorporated into the backbone network, and the DCNv4 module was integrated to boost the model's capability in detecting irregular defects. In the neck network, the BiFormer attention mechanism was implemented to enhance detection performance for small-scale defects. Moreover, the newly developed MASFF_Head structure was adopted as the detection head to augment detection efficacy for multi-scale defects. Additionally, the bounding box-loss function was substituted with the WIoU loss to improve performance on low-quality samples. Experimental results demonstrated that the mAP@0.5 of the refined YOLOv8 network model reached 86.1%, marking a 2.8% increase over the original model and significantly boosting detection accuracy. When compared to other deep learning models, the enhanced YOLOv8 model exhibits considerable superiority in terms of detection precision and efficiency. The P value and R value achieved were 82.9% and 80.8%, respectively, with a detection time of 21.6 milliseconds. This underscores the method's effectiveness and reliability in detecting automotive transmission gear defects, underscoring its pivotal role in facilitating automated detection processes on industrial production lines.

Wearable damaged clothing fabric image recognition system based on image restoration algorithm

With the improvement of people's living standards and the rapid growth of national technological strength, wearable devices have gradually integrated into people's daily lives, making them convenient for people's work and lives. Traditional fabric detection methods often rely on manual visual inspection, which has low efficiency and accuracy that does not meet the requirements. In view of this, the research has innovatively integrated GAN, CNN, field Programmable gate array (FPGA), multi-stage image recovery algorithm based on graph neural network, for high frequency and low frequency fusion, and combined it with wearable devices to develop a new wearable fabric recovery and recognition system. The results showed that in the COCO dataset, after 250 iterations, the WFR2S algorithm achieved a damage identification accuracy of 95.5 %, a mean absolute error of 0.011, a real-time processing speed of 30 fps, and a resource consumption of 2.4 W. These data demonstrate the significant performance advantages of WFR2S in image recovery and damage detection. The newly developed image recovery algorithm improves the accuracy, availability and stability of damage recognition in garment fabrics. This study not only has theoretical value but also has broad application prospects in practical application. The aim is to contribute to the intelligent development of the garment industry.

Utilization of convolutional neural networks to analyze microscopic images for high-throughput screening of mesenchymal stem cells.

This work investigated the high-throughput classification performance of microscopic images of mesenchymal stem cells (MSCs) using a hyperspectral imaging-based separable convolutional neural network (CNN) (H-SCNN) model. Human bone marrow mesenchymal stem cells (hBMSCs) were cultured, and microscopic images were acquired using a fully automated microscope. Flow cytometry (FCT) was employed for functional classification. Subsequently, the H-SCNN model was established. The hyperspectral microscopic (HSM) images were created, and the spatial-spectral combined distance (SSCD) was employed to derive the spatial-spectral neighbors (SSNs) for each pixel in the training set to determine the optimal parameters. Then, a separable CNN (SCNN) was adopted instead of the classic convolutional layer. Additionally, cultured cells were seeded into 96-well plates, and high-functioning hBMSCs were screened using both manual visual inspection (MV group) and the H-SCNN model (H-SCNN group), with each group consisting of 96 samples. FCT served as the benchmark to compare the area under the curve (AUC), F1 score, accuracy (Acc), sensitivity (Sen), specificity (Spe), positive predictive value (PPV), and negative predictive value (NPV) between the manual and model groups. The best classification Acc was 0.862 when using window size of 9 and 12 SSNs. The classification Acc of the SCNN model, ResNet model, and VGGNet model gradually increased with the increase in sample size, reaching 89.56 ± 3.09, 80.61 ± 2.83, and 80.06 ± 3.01%, respectively at the sample size of 100. The corresponding training time for the SCNN model was significantly shorter at 21.32 ± 1.09 min compared to ResNet (36.09 ± 3.11 min) and VGGNet models (34.73 ± 3.72 min) (P < 0.05). Furthermore, the classification AUC, F1 score, Acc, Sen, Spe, PPV, and NPV were all higher in the H-SCNN group, with significantly less time required (P < 0.05). Microscopic images based on the H-SCNN model proved to be effective for the classification assessment of hBMSCs, demonstrating excellent performance in classification Acc and efficiency, enabling its potential to be a powerful tool in future MSCs research.

Comparing visual inspection methods for parenteral products in hospital pharmacy: between reliability, cost, and operator formation considerations

IntroductionThe COVID-19 pandemic has led to unforeseen and novel manifestations, as illustrated by the management of drug shortages through the development of hospital production of sterile pharmaceutical preparations (P2S). Visual...

UAV imagery-based railroad station building inspection using hybrid learning architecture

The apparent defects of railroad station building such as the broken glass, missing screw, and rust corrosion need to be inspected regularly and timely to ensure the safety of passenger. Unmanned aerial vehicle (UAV) imagery-based inspections have the potential to revolutionize current manual visual inspections by providing a better overhead view and mitigating safety concerns. This paper proposes a hybrid learning architecture called YOLOS (you only look once station scene) to simultaneously detect and segment station building surface defects of UAV images. First, a novel squeeze-and-excitation (SE) attention block is integrated into the detection branch to adaptively learn the weights of the feature channels, thereby promoting the network to pay attention to the critical deep features of the objects. And then, a new semantic segmentation branch parallel to the detection branch is designed and assembled in YOLOS for pixel-level defect recognition. Finally, extensive experiments on railroad station dataset established with drone imagery prove the effectiveness and robustness of the SE-based network on station surface defect detection. This method can quickly convert UAV imagery into useful information with a high detection rate.

UAV vision-based crack quantification and visualization of bridges: system design and engineering application

Accurately measuring visible cracks in bridges is crucial for their structural health diagnosis, damage detection, performance evaluation, and maintenance planning. The primary means of visual crack detection still relies heavily on manual visual inspection, an inefficient process that can pose significant safety risks. This article develops a unmanned aerial vehicle (UAV) vision-based surface crack measurement methodology and visualization scheme for the bridges that can detect and measure cracks automatically with improved efficiency. The surface crack measurement methodology is achieved by designing a three-stage crack sensing system including the You Only Look Once-based crack recognition, U-shaped network-based crack segmentation, and deep-vision-based crack width calculation. This workflow is integrated into a comprehensive UAV inspection system, which is intended for operation at the field. The surface crack visualization scheme is accomplished by taking advantage of time-series image fusion, GPS information migration, and three-dimensional (3D) point cloud technique to reconstruct the 3D geometrical model of the tested bridge, which is convenient for unveiling the crack information in the bridge. The proposed methodology was successfully validated by a case study on an arch bridge. The achievement of this article promotes the UAV vision-based bridge’s surface crack inspection technology to a new status that no preparation for pasting calibration marker is needed, and crack identification, segmentation, and width calculation are realized promptly during the UAV flying on-site, as well as damage evaluation for bridges is visually fulfilled based on the reconstructed digital-graphical 3D model. The working environments and influencing factors to the developed system are sufficiently discussed. Certain limitations in the current application are pointed out for future improvements.

WaveSegNet: An Efficient Method for Scrap Steel Segmentation Utilizing Wavelet Transform and Multiscale Focusing

Scrap steel represents a sustainable and recyclable resource, instrumental in diminishing carbon footprints and facilitating the eco-friendly evolution of the steel sector. However, current scrap steel recycling faces a series of challenges, such as high labor intensity and occupational risks for inspectors, complex and diverse sources of scrap steel, varying types of materials, and difficulties in quantifying and standardizing manual visual inspection and rating. Specifically, we propose WaveSegNet, which is based on wavelet transform and a multiscale focusing structure for scrap steel segmentation. Firstly, we utilize wavelet transform to process images and extract features at different frequencies to capture details and structural information in the images. Secondly, we introduce a mechanism of multiscale focusing to further enhance the accuracy of segmentation by extracting and perceiving features at different scales. Through experiments conducted on the public Cityscapes dataset and scrap steel datasets, we have found that WaveSegNet consistently demonstrates superior performance, achieving the highest scores on the mIoU metric. Particularly notable is its performance on the real-world scrap steel dataset, where it outperforms other segmentation algorithms with an average increase of 3.98% in mIoU(SS), reaching 69.8%, and a significant boost of nearly 5.98% in mIoU(MS), achieving 74.8%. These results underscore WaveSegNet’s exceptional capabilities in processing scrap steel images. Additionally, on the publicly available Cityscapes dataset, WaveSegNet shows notable performance enhancements compared with the next best model, Segformer. Moreover, with its modest parameters and computational demands (34.1 M and 322 GFLOPs), WaveSegNet proves to be an ideal choice for resource-constrained environments, demonstrating high computational efficiency and broad applicability. These experimental results attest to the immense potential of WaveSegNet in intelligent scrap steel rating and provide a new solution for the scrap steel recycling industry. These experimental results attest to the immense potential of WaveSegNet in intelligent scrap steel rating and provide a new solution for the scrap steel recycling industry.

Machine vision model for detection of foreign substances at the bottom of empty large volume parenteral.

Empty large volume parenteral (LVP) bottle has irregular shape and narrow opening, and its detection accuracy of the foreign substances at the bottom is higher than that of ordinary packaging bottles. The current traditional detection method for the bottom of LVP bottles is to directly use manual visual inspection, which involves high labor intensity and is prone to visual fatigue and quality fluctuations, resulting in limited applicability for the detection of the bottom of LVP bottles. A geometric constraint-based detection model (GCBDM) has been proposed, which combines the imaging model and the shape characteristics of the bottle to construct a constraint model of the imaging parameters, according to the detection accuracy and the field of view. Then, the imaging model is designed and optimized for the detection. Further, the generalized GCBDM has been adopted to different bottle bottom detection scenarios, such as cough syrup and capsule medicine bottles by changing the target parameters of the model. The GCBDM, on the one hand, can avoid the information at the bottom being blocked by the narrow opening in the imaging optical path. On the other hand, by calculating the maximum position deviation between the center of visual inspection and the center of the bottom, it can provide the basis for the accuracy design of the transmission mechanism in the inspection, thus further ensuring the stability of the detection.

Automated Measurement of Smartphone Screen Scratches Using Visual Methods

Abstract Quantitative measurement of smartphone screen scratches is crucial for pricing in the used smartphone market. Traditional manual visual inspection methods suffer from inherent limitations, namely being labor-intensive, subjective, and prone to inaccuracy. Hence, this study proposes a vision-based measurement method as a viable solution to overcome these challenges. The algorithm uses the Hessian enhancement to extract scratch features, applies adaptive thresholding to distinguish features from the background, and employs morphological reconstruction to reconstruct complete scratches. The topological analysis splits and mergers intersecting scratches, enabling individual segmentation. Finally, four metrics for measuring screen scratches include length, brightness, contrast, and maximum width to quantitatively characterize the damage of screen scratches. Experiments showed that the proposed algorithm outperforms other vision-based methods, with an accuracy of 99.6% in estimating the scratch length and a running time of 43.7 ms, which fully meets the efficiency and accuracy requirements of industrial application.

Comparative evaluation of SNVs, indels, and structural variations detected with short- and long-read sequencing data

Short- and long-read sequencing technologies are routinely used to detect DNA variants, including SNVs, indels, and structural variations (SVs). However, the differences in the quality and quantity of variants detected between short- and long-read data are not fully understood. In this study, we comprehensively evaluated the variant calling performance of short- and long-read-based SNV, indel, and SV detection algorithms (6 for SNVs, 12 for indels, and 13 for SVs) using a novel evaluation framework incorporating manual visual inspection. The results showed that indel-insertion calls greater than 10 bp were poorly detected by short-read-based detection algorithms compared to long-read-based algorithms; however, the recall and precision of SNV and indel-deletion detection were similar between short- and long-read data. The recall of SV detection with short-read-based algorithms was significantly lower in repetitive regions, especially for small- to intermediate-sized SVs, than that detected with long-read-based algorithms. In contrast, the recall and precision of SV detection in nonrepetitive regions were similar between short- and long-read data. These findings suggest the need for refined strategies, such as incorporating multiple variant detection algorithms, to generate a more complete set of variants using short-read data.

YOLOv5s fabric defect detection model with mixed attention mechanism

In the field of industrial quality inspection, fabric defect detection is regarded as an important research topic. The traditional inspection method mainly relies on manual visual inspection, which is time-consuming and easy to be affected by human factors, and has the problems of low accuracy, slow speed and high cost. Therefore, it is very important to use digital image processing technology and target detection technology to develop high precision, high efficiency and low-cost fabric defect detection model suitable for application deployment. Driven by deep learning-based neural network technology, many models with excellent performance have emerged in the field of object detection. This paper will improve on the basis of YOLOv5s model. ACmix module is introduced in Backbone layer of YOLOv5s model. This module overcomes the problem that convolution operation may cause local information loss after feature extraction, improves the model's limitation on feature extraction of complex background images, and enhances the model's processing ability in local feature fusion. Compared with the original YOLOv5s algorithm, mAP has improved by 0.6%. In the complex background, the detection accuracy of fabric defects targets has been significantly optimized and improved.

Instance segmentation algorithm for sorting dismantling components of end-of-life vehicles

In the recycling and dismantling procedures of End-of-Life Vehicles, the diverse thicknesses of disassembled parts necessitate specialized recycling processes, underscoring the imperative for efficient classification and recycling methods. Presently, the classification of disassembled components from scrapped vehicles relies heavily on manual visual inspection and caliper measurements, resulting in inefficiency and low accuracy. To overcome the challenges associated with thickness classification for disassembled components of End-of-Life Vehicles, propose a paradigm shift by introducing an intelligent grading strategy. This approach aims to significantly improve efficiency and accuracy in the classification process, offering a more effective solution for the recycling of dismantled parts. In response to the inefficiencies, low accuracy, and safety concerns associated with manual sorting, propose a deep learning model, incorporating an attention mechanism, for accomplishing instance segmentation tasks of End-of-Life Vehicles components. The model introduces a global attention mechanism into the backbone network for feature extraction, which is further integrated into the instance segmentation branch, resulting in a significant enhancement of instance segmentation performance. The model is trained and optimized using a dataset collected from an End-of-Life Vehicles dismantling facility in China. Experimental results show that when the intersection over union is 0.5, the mean Average Precision of the model in target detection and instance segmentation reaches 93.2% and 92.1% respectively., and 77.2% and 72.4% in the intersection over union is 0.5–0.95. In comparison to the unmodified baseline model, the model exhibits improvements of 1.7%, 2%, 5.8%, and 3.1% across evaluation metrics. When contrasted with traditional manual sorting methods, the proposed model exhibits pronounced advantages in terms of accuracy and fairness. In conclusion, the model effectively addresses the challenge of intelligent sorting of dismantled components from End-of-Life Vehicles during the dismantling process.

Autonomous damage segmentation of post-fire reinforced concrete structural components

The existing methods for detecting fire damage in reinforced concrete (RC) structures rely heavily on manual visual inspection, which is time-consuming and challenging due to poor visibility and safety concerns. To address this, an automatic damage segmentation network at the pixel level is proposed. Using a dataset of 403 high-resolution images of fire-damaged RC components, the study employed MobileNetv3 as the encoder in a UNet model, enhanced with three average pooling layers. Among four submodules tested, SPPF demonstrated the best performance in the MB-SPPF-UNet network, achieving an MIoU of 82%, a 7% improvement over the Baseline-UNet. While OCRNet showed the best performance among ten state-of-the-art networks, its MIoU was still 8% lower than MB-SPPF-UNet. The generalization capabilities of the networks were evaluated using four crack datasets, with MB-SPPF-UNet showing a performance decrease of 5%-10%. However, this is acceptable given its primary application in segmenting post-fire concrete spalling and exposed reinforcement. The study aims to establish an efficient damage segmentation network using deep learning for quick assessment and reference in firefighting, rescue, and structural repair efforts.

High-Performance Defect Detection Methods for Real-Time Monitoring of Ceramic Additive Manufacturing Process Based on Small-Scale Datasets

Vat photopolymerization is renowned for its high flexibility, efficiency, and precision in ceramic additive manufacturing. However, due to the impact of random defects during the recoating process, ensuring the yield of finished products is challenging. At present, the industry mainly relies on manual visual inspection to detect defects; this is an inefficient method. To address this limitation, this paper presents a method for ceramic vat photopolymerization defect detection based on a deep learning framework. The framework innovatively adopts a dual-branch object detection approach, where one branch utilizes a fully convolution network to extract the features from fused images and the other branch employs a differential Siamese network to extract the differential information between two consecutive layer images. Through the design of the dual branches, the decoupling of image feature layers and image spatial attention weights is achieved, thereby alleviating the impact of a few abnormal points on training results and playing a crucial role in stabilizing the training process, which is suitable for training on small-scale datasets. Comparative experiments are implemented and the results show that using a Resnet50 backbone for feature extraction and a HED network for the differential Siamese network module yields the best detection performance, with an obtained F1 score of 0.89. Additionally, as a single-stage defect object detector, the model achieves a detection frame rate of 54.01 frames per second, which meets the real-time detection requirements. By monitoring the recoating process in real-time, the manufacturing fluency of industrial equipment can be effectively enhanced, contributing to the improvement of the yield of ceramic additive manufacturing products.

A Systematic Review of Automated Construction Inspection and Progress Monitoring (ACIPM): Applications, Challenges, and Future Directions

Despite the subjective and error-prone nature of manual visual inspection procedures, this type of inspection is still a common process in most construction projects. However, Automated Construction Inspection and Progress Monitoring (ACIPM) has the potential to improve inspection processes. The objective of this paper is to examine the applications, challenges, and future directions of ACIPM in a systematic review. It explores various application areas of ACIPM in two domains of (a) transportation construction inspection, and (b) building construction inspection. The review identifies key ACIPM tools and techniques including Laser Scanning (LS), Uncrewed Aerial Systems (UAS), Robots, Radio Frequency Identification (RFID), Augmented Reality (AR), Virtual Reality (VR), Computer Vision (CV), Deep Learning, and Building Information Modeling (BIM). It also explores the challenges in implementing ACIPM, including limited generalization, data quality and validity, data integration, and real-time considerations. Studying legal implications and ethical and social impacts are among the future directions in ACIPM that are pinpointed in this paper. As the main contribution, this paper provides a comprehensive understanding of ACIPM for academic researchers and industry professionals.

Deep Learning Techniques for Identification of Different Malvaceae Plant Leaf Diseases

INTRODUCTION: The precise and timely detection of plant diseases plays a crucial role in ensuring efficient crop management and disease control. Nevertheless, conventional methods of disease identification, which heavily rely on manual visual inspection, are often time-consuming and susceptible to human error. The knowledge acquired from this research paper enhances the overall comprehension of the discipline and offers valuable direction for future progressions in the application of deep learning for the identification of plant diseases.[1][2] AIM: to investigate the utilization of deep learning techniques in identifying various Malvaceae plant diseases. METHODS: AlexNet, VGG, Inception, REsNet and other CNN architectures are analyzed on Malvaceae plant diseases specially on Cotton, Ocra and Hibiscus, different data collection methods ,Data augmentation and Normalization techniques. RESULTS: Inception V4 have Training Accuracy 98.58%, VGG-16 have Training Accuracy 84.27%, ResNet-50 have Training Accuracy 98.72%, DenseNet have Training Accuracy 98.87%, Inception V4 have Training Loss 0.01%, VGG-16 have Training Loss 0.52%, ResNet-50 have Training Loss 6.12%, DenseNet have Training Loss 0.016%, Inception V4 have Test Accuracy 97.59%, VGG-16 have Test accuracy 82.75%, ResNet-50 have Test Accuracy 98.73%, DenseNet have Test Accuracy 99.81%, Inception V4 have Test Loss 0.0586%, VGG-16 have Test Loss 0.64%, ResNet-50 have Test Loss 0.027%, DenseNet have Test Loss 0.0154% . CONCLUSION: conclusion summarizes the key findings and highlights the potential of deep learning as a valuable tool for accurate and efficient identification of Malvaceae plant diseases.

DSCU-Net: MEMS Defect Detection Using Dense Skip-Connection U-Net

With the rapid development of intelligent manufacturing and electronic information technology, integrated circuits play a vital role in high-end chips. The semiconductor chip manufacturing process requires precise operation and strict control to ensure chip quality. The traditional manual visual inspection method has a high workforce cost and intense subjectivity and is accompanied by a high level of misdetection and leakage. Computer vision-based wafer defect detection technology is gaining popularity in the industry. However, previous methods still find it challenging to meet the production requirements regarding accuracy. To solve the problem, we propose a defect detection network based on a coding and decoding structure, Dense Skip-Connection U-Net (DSCU-Net), which optimizes the skip connection between the encoder and decoder and enhances the profound fusion of high-level semantics and low-level semantics to improve accuracy. To verify the effectiveness of DSCU-Net, we validate it in actual microelectromechanical systems (MEMS) data, and the results show that DSCU-Net reaches an optimal level. Therefore, the DSCU-Net proposed in this paper effectively solves the defect detection problem in semiconductor chip manufacturing. This method reduces workforce cost and subjectivity interference and improves inspection efficiency and accuracy. It will help to promote further development in the field of intelligent manufacturing and electronic information technology.