Search

Steel-manufacturing sites are extremely harsh and dangerous environments. Visibility is reduced by dust, water vapor, oil, and low light conditions. Additionally, the steel products from hot furnaces have surface temperatures ranging from 1200 to 800 degrees celsius and weigh several to tens of tons. It is extremely dangerous for humans to visually inspect these steel products in such environments. Therefore, the use of automatic inspection equipment for steel surfaces is essential. Initially, image-processing methods were used, but with recent advances in deep learning, deep-learning-based methods are also being applied in this field. However, the method currently widely used involves utilizing existing models through transfer learning, which inevitably causes curvature of the input image data and limits performance. Furthermore, previous studies have focused on 2D sheet metal products, and there has been no research on three-dimensional geometric products. In this study, we propose dataset generation through geometric transformations that parameterize the structure of the steel surface defect detector hardware, along with a performance-based model optimization algorithm. In validation experiments, an average F1 score of 0.932 and an average area under curve of 0.99 were obtained, implying that the proposed algorithm has near-ideal performance. [ABSTRACT FROM AUTHOR]

When detecting surface defects in a complex industrial cutting environment, the defects are easily polluted and covered by interfering factors (chips or coolant residues). The defect of the surface images with interference factors is a novel problem in the existing studies, and it is also a difficulty in the detection field. Hence, this paper proposes a high-precision anti-interference detection method for surface defects under the influence of complex environment. The detection method provides a new research idea, which is divided into three main processes: interference regions location, interference regions repair, defect detection. The regions affected by interference factors are adaptively located through the proposed Efficient Channel Attention Network (ECANet)-DeeplabV3 + network model. The mean Pixel Accuracy (mPA) and mean Intersection over Union (mIoU) of ECANet-DeeplabV3 + network model for interference factor identification are 98.37% and 95.46%, respectively. The Criminisi algorithm is improved from priority, finding the best matching block, and searching regions. Directional repair based on the improved Criminisi algorithm is performed on the identified interfering regions removing the interfering factors in the image, which is the research core. Then, defect detection is performed on the repaired image using the improved superpixel technology. At the same time, the defect detection results provide a variety of surface defect information for the cutting staff, including defect types, the number of pixels in different defect regions, and the area ratio of different defect regions. This information improves predictive maintenance and surface quality control. [ABSTRACT FROM AUTHOR]

The detection of surface defects on metals plays a pivotal role in the evaluation of aluminum profile product quality. To date, the most prevalent approach for detecting metal surface defects is via the use of conventional neural networks (CNNs). However, in recent years, transformer-based methods have achieved notable improvements in computer vision and exhibited superiority over traditional CNNs in most tasks. Regarding aluminum surface defect image characteristics, such as intricate textures, few defect samples, and large-scale differences in various defects, it is difficult to further improve the performance of traditional CNNs. Hence, this study proposes a novel variant transformer-based method for aluminum profile surface defect detection that combines a traditional CNN and transformer architecture with a deformable attention module (DAM). The DAM focuses on a subset of the critical sampling points around the reference point to alleviate the intractable complexity associated with high-resolution feature maps and enhances the recognition effect of small target defects. The proposed image-stitching method uses defect-free samples to generate new samples, which partially mitigates the issue of class imbalance. Moreover, we performed supplementary experiments to confirm the effectiveness of transfer learning in training small-scale datasets. Compared with a baseline, the mean average precision (mAP) improved by 3.32%. Extensive experimental results demonstrate the efficacy of our method in accurately detecting surface defects in aluminum profiles and significantly improving the detection of small target defects. [ABSTRACT FROM AUTHOR]

As part of the evolving Industry 4.0 landscape, machine learning-based visual inspection plays a key role in enhancing production efficiency. Screen printing, a versatile and cost-effective manufacturing technique, is widely applied in industries like electronics, textiles, and automotive. However, the production of complex multilayered designs is error-prone, resulting in a variety of defect appearances and classes. These defects can be characterized as small in relation to large sample areas and weakly pronounced. Sufficient defect visualization and robust defect detection methods are essential to address these challenges, especially considering the permitted design variability. In this work, we present a novel automatic visual inspection system for surface defect detection on decorated foil plates. Customized optical modalities, integrated into a sequential inspection procedure, enable defect visualization of production-related defect classes. The introduced patch-wise defect detection methods, designed to leverage less labeled data, prove effective for industrial defect detection, meeting the given process requirements. In this context, we propose an industry-applicable and scalable data preprocessing workflow that minimizes the overall labeling effort while maintaining high detection performance, as known in supervised settings. Moreover, the presented methods, not relying on any labeled defective training data, outperformed a state-of-the-art unsupervised anomaly detection method in terms of defect detection performance and inference speed. [ABSTRACT FROM AUTHOR]

To accurately detect defects, we propose an enhanced model based on YOLOv8, named STE-YOLO. To address the aforementioned challenges, this paper adopts YOLOv8 as the improved model. The structure of this paper is as follows: We enhance the model's feature extraction and small detail recognition by integrating GhostConv into partial convolutions. In order to address the attention bias of the model, we introduce a Bottleneck Transformer self-attention convolution layer that effectively improves localization box accuracy. For the problem of defect category mismatches, we exploit the C2f-LSKA attention mechanism in the model head to address this issue. The experimental results indicate that the improved model achieves a mean average precision (mAP) of 79.0%, compared to 65.8% for the original model, marking an improvement of 13.1%. STE-YOLO significantly increases the precision of detecting surface defects on strip steel. [ABSTRACT FROM AUTHOR]

The abstract serves both as a general introduction to the topic and as a brief, non-technical summary of the main results and their implications. Authors are advised to check the author instructions for the journal they are submitting to for word limits and if structural elements like subheadings, citations, or equations are permitted. Engine blades, as key components of the aviation engine, operate under high-speed rotation and high-temperature conditions, making them susceptible to defects such as fatigue, cracks and corrosion. This paper presents an innovative approach to detecting defects in aviation engine blades. To increase the detection accuracy of irregular defects, we design a novel deformable convolutional network (DCN) based feature extraction module. It employs the deformable convolutional structure to extract the features of blades with different shapes. To enhance the accuracy of locating small targets, the Channel Attention Module is adopted to enable the network focus on surface anomalies. Apart from that, the DSConv module is designed to decrease the model parameters and improve the detection speed. Extensive tests have been conducted on the collected dataset of aviation engine blade with surface defects. The algorithm achieves an average detection accuracy of 97.1%. The algorithms inference performance could reach up to 25 fps on the TX2 device, which satisfies the real-time detection requirement. [ABSTRACT FROM AUTHOR]

Copyright of Computer Measurement & Control is the property of Magazine Agency of Computer Measurement & Control and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In order to address challenges such as small target sizes, low contrast, significant intra-class variations, and indistinct inter-class differences in surface defect detection, this paper proposes the Enhanced Context-aware Parallel Fusion Network (EC-PFN). The network employs a Featur Weave Network architecture to enhance contextal awareess and parallel fusion capabilities. It utilizes a Feature Fusion Module (UniFusionLayer) for effective multiscale and multisemantic feature learning, offering new perspectives on feature fusion. Additionally, a Receptive Field Block (RFB) module is introduced to expand the receptive field, enhancing feature extraction in scenarios with low contrast and subtle defects. The Loss Ranking Module (LRM) is incorporated to optimize the target-oriented loss, improving performance by omitting low-confidence bounding boxes. Extensive experiments on a light guide plate defect dataset demonstrate that EC-PFN achieves a detection accuracy (mAP) of 98.9%, a detection speed of 92 FPS, and a computational cost of 14.5 GFLOPs, outperforming mainstream surface defect detection models. [ABSTRACT FROM AUTHOR]

This paper presents an enhanced Faster R-CNN model for detecting surface defects in resistance welding spots, improving both efficiency and accuracy for body-in-white quality monitoring. Key innovations include using high-confidence anchor boxes from the RPN network to locate welding spots, using the SmoothL1 loss function, and applying Fast R-CNN to classify detected defects. Additionally, a new pruning model is introduced, reducing unnecessary layers and parameters in the neural network, leading to faster processing times without sacrificing accuracy. Tests show that the model achieves over 90% accuracy and recall, processing each image in about 15 ms, meeting industrial requirements for welding spot inspection. [ABSTRACT FROM AUTHOR]

This study tackles a critical challenge in converting two-dimensional (2D) images into three-dimensional (3D) representations, focusing on the precise detection of flat surfaces. The research utilizes a triangulation method involving laser and camera systems, emphasizing the optimization of laser shooting angles and camera positioning to accurately determine z-coordinates. The methodology employs hue, saturation, and value (HSV) color masking, which has proven superior to traditional red, green, blue (RGB) methods for isolating red line objects. Key findings indicate that the optimal laser angle, ß1=70.65°, significantly minimizes root mean square (RMS) error, thereby enhancing the accuracy of 3D imaging. Additionally, the use of three laser lines at different angles enables a more comprehensive detection of z-coordinates by creating multiple reference points across the surface. This arrangement improves the robustness and precision of the 3D reconstruction process, as the intersecting laser lines generate detailed coordinate data that is critical for accurately mapping surface irregularities. These results not only support existing theories in digital feature extraction but also offer a robust framework for practical applications in manufacturing and quality control, particularly in surface defect detection. The study's innovative approach advances the field of computer vision, providing new insights and methodologies for optimizing image conversion techniques. [ABSTRACT FROM AUTHOR]

Surface defect detection plays a crucial role in ensuring the quality standards of hot-rolled steel strips. To meet the demands for high precision and real-time performance in industrial defect detection, this paper introduces an improved one-stage detector based on YOLOv5s, named GLF-NET, that focuses on a good balance between speed and precision. Firstly, an Attention Augmented Module (AAM) is proposed and used in the backbone, with the aim of minimizing the loss of semantic and location information of defects during the process of feature extraction. Secondly, to enrich the model’s capacity of representing multi-scale features of defects, an innovative Global and Local Dynamic Feature Fusion (GLF) module is designed and plugged into the top-down FPN part of the neck, bridging the semantic gap between different feature layers and enabling the model to adaptively select features for fusion. Additionally, a novel Receptive Field Augmented Module (RFA) is proposed and integrated into the bottom-up PAN structure of the neck, enhancing the detector’s ability of perceiving defects with irregular shapes and large aspect ratios. Extensive experimental results on the NEU-DET steel strip surface defect dataset demonstrate that GLF-NET obtains an impressive mAP value of 79.2%, exceeding YOLOv5s by 4.2%. Furthermore, with an impressive detection speed of 95 Frames Per Second (FPS), GLF-NET not only meets the real-time demands of industrial defect detection but also demonstrates exceptional capabilities in defect detection. Code is available at https://github.com/MYF1124/GLF-NET.

In industrial applications, surface defect segmentation is a critical task. However, facing challenges such as diverse defect scales, low contrast between defects and background, high interclass similarity and real-time detection in defect inspection, we propose an efficient lightweight network, named DMC-Net, for real-time surface defect segmentation. The structural optimization of DMC-Net includes the following components: (1) depthwise separable convolution attention module, a lightweight and efficient feature extraction module for extracting multi-scale defect features. (2) Multi-scale feature enhancement module, providing long-range information capture and local information focusing to enhance defect localization capability. (3) Channel shuffle group convolution, enhancing feature interaction and information propagation while reducing the parameter quantity. Based on the experimental results, DMC-Net achieved an mIoU of 73.74% on the NEU-SEG dataset, while achieving an FPS of 211.7. This indicates that we have successfully reduced the complexity and computational cost of the model while improving performance, providing a feasible solution for industrial applications. The relevant code can be obtained at [ABSTRACT FROM AUTHOR]

Image processing and object recognition technologies for detecting defects on the surfaces of a variety of materials is critical to ensuring the safety and durability of infrastructure. That is why this topic and problems of the field of defect detection were chosen in the study. Therefore, the development of effective image processing methods for defect identification, especially in low-light conditions and hard-to-reach places, is of great relevance. The research is a comparison of classical image processing methods with modern deep learning algorithms such as CNN (Convolutional Neural Networks) and YOLO (You Only Look Once). The study analyzes the effectiveness of these methods under specific defect inspection conditions, including diffuse lighting and device mobility. An important aspect is the use of microcomputers such as Raspberry Pi and Nvidia Jetson Nano, which ensures the mobility and autonomy of the system. The practical value of the research lies in the implementation of effective image processing methods for detecting defects on the surfaces of engineering structures. This makes it possible to significantly improve the accuracy of surface defect identification, which is confirmed by the IoU (Intersection over Union) and Dice metrics. In particular, using CNNs for surface defect identification showed 35% better results compared to existing implementations of similar networks and 12% more efficient compared to YOLO. On the other hand, YOLO proved to be more productive in terms of processing frames per second on microcomputers, which is important for real-time monitoring.

Abstract Leather has played a very important role in human life for thousands of years. Ensuring the quality of leather and addressing surface defects poses significant challenges. Traditionally, human inspectors are responsible for detecting surface defects in tanneries, but this approach is labor-intensive and susceptible to human error. As a result, there is a growing demand for automated systems to detect the defects. Herein, artificial intelligence (AI) was developed to detect the defects on leather surfaces. Six targeted defect types, denoted as insect bites, scratches, holes, stitch marks, diseased and ruptures, were specifically addressed to enhance the overall quality assessment process. AI-based vision techniques were used to detect flaws on the leather on photographs taken with a high-resolution camera. Deep learning algorithms Mask R-CNN, YOLOv8 and within the framework of Detectron2, RetinaNet R101 3x, Faster R-CNN R101-FPN 3x models were performed and a comparison was made between these algorithms. By using the slicing aided hyper-inference (SAHI) algorithm in coordination with these algorithms, the detection rates of small defects on the images were increased. The highest accuracy rate was achieved when the YOLOv8 algorithm had 75 epoch values for training, and the SAHI algorithm had slice height-width values of 256 × 256 pixels. Graphical Abstract

Abstract In order to address challenges such as small target sizes, low contrast, significant intra-class variations, and indistinct inter-class differences in surface defect detection, this paper proposes the Enhanced Context-aware Parallel Fusion Network (EC-PFN). The network employs a Featur Weave Network architecture to enhance contextal awareess and parallel fusion capabilities. It utilizes a Feature Fusion Module (UniFusionLayer) for effective multiscale and multisemantic feature learning, offering new perspectives on feature fusion. Additionally, a Receptive Field Block (RFB) module is introduced to expand the receptive field, enhancing feature extraction in scenarios with low contrast and subtle defects. The Loss Ranking Module (LRM) is incorporated to optimize the target-oriented loss, improving performance by omitting low-confidence bounding boxes. Extensive experiments on a light guide plate defect dataset demonstrate that EC-PFN achieves a detection accuracy (mAP) of 98.9%, a detection speed of 92 FPS, and a computational cost of 14.5 GFLOPs, outperforming mainstream surface defect detection models.

Abstract Compared to the surface defect detection of industrial products produced according to specified processes, the detection of surface defects in naturally grown red jujubes poses unique and significant challenges for researchers. The high diversity of surface defects, subtle distinctions from the background, low contrast, varying scales, and the presence of high levels of noise in images are among the factors that greatly amplify the complexity of defect detection tasks. Existing methods show some deficiencies in addressing these issues, mainly due to insufficient feature extraction capabilities and overly complex network structures, leading to limitations in model efficiency and practical application performance. To tackle the challenges associated with red jujube surface defect detection, this study proposes an optimized Tiny Vision Transformer (TinyViT) network structure, named RJ-TinyViT. This method refines the TinyViT-5 m network structure to reduce network burden and introduces an improved Multi-Kernel Block (MK Block) and an improved Mobile Inverted Bottleneck Convolution Block (MBConv Block) to enhance feature extraction capabilities. Additionally, we have integrated the Coordinate Attention (CA) module to enhance the model’s capacity for recognizing and focusing on features of surface defects on red jujubes. Experimental results show that RJ-TinyViT achieved a classification accuracy of 93.38%, marking an improvement of 1.84% over the original TinyViT network. At the same time, its Floating-point Operations (FLOPs) and Parameters (Params) were reduced to 58.97% and 39.84% of the original TinyViT network, respectively. These results not only demonstrate that RJ-TinyViT achieves model lightweighting while maintaining high accuracy but also highlight its value in practical industrial applications.

Leather has played a very important role in human life for thousands of years. Ensuring the quality of leather and addressing surface defects poses significant challenges. Traditionally, human inspectors are responsible for detecting surface defects in tanneries, but this approach is labor-intensive and susceptible to human error. As a result, there is a growing demand for automated systems to detect the defects. Herein, artificial intelligence (AI) was developed to detect the defects on leather surfaces. Six targeted defect types, denoted as insect bites, scratches, holes, stitch marks, diseased and ruptures, were specifically addressed to enhance the overall quality assessment process. AI-based vision techniques were used to detect flaws on the leather on photographs taken with a high-resolution camera. Deep learning algorithms Mask R-CNN, YOLOv8 and within the framework of Detectron2, RetinaNet R101 3x, Faster R-CNN R101-FPN 3x models were performed and a comparison was made between these algorithms. By using the slicing aided hyper-inference (SAHI) algorithm in coordination with these algorithms, the detection rates of small defects on the images were increased. The highest accuracy rate was achieved when the YOLOv8 algorithm had 75 epoch values for training, and the SAHI algorithm had slice height-width values of 256 × 256 pixels. [ABSTRACT FROM AUTHOR]

To address the limitations of existing deep learning-based algorithms in detecting surface defects on brake pipe ends, a novel lightweight detection algorithm, FP-YOLOv8, is proposed. This algorithm is developed based on the YOLOv8n framework with the aim of improving accuracy and model lightweight design. First, the C2f_GhostV2 module has been designed to replace the original C2f module. It reduces the model's parameter count through its unique design. It achieves improved feature representation by adopting specific technique within its structure. Additionally, it incorporates the decoupled fully connected (DFC) attention mechanism, which minimizes information loss during long-range feature transmission by separately capturing pixel information along horizontal and vertical axes via convolution. Second, the Dynamic ATSS label allocation strategy is applied, which dynamically adjusts label assignments by integrating Anchor IoUs and predicted IoUs, effectively reducing the misclassification of high-quality prediction samples as negative samples. Thus, it improves the detection accuracy of the model. Lastly, an asymmetric small-target detection head, FADH, is proposed to utilize depth-separable convolution to accomplish classification and regression tasks, enabling more precise capture of detailed information across scales and improving the detection of small-target defects. The experimental results show that FP-YOLOv8 achieves a mAP50 of 89.5% and an F1-score of 87% on the ends surface defects dataset, representing improvements of 3.3% and 6.0%, respectively, over the YOLOv8n algorithm, Meanwhile, it reduces model parameters and computational costs by 14.3% and 21.0%. Additionally, compared to the baseline model, the AP50 values for cracks, scratches, and flash defects rise by 5.5%, 5.6%, and 2.3%, respectively. These results validate the efficacy of FP-YOLOv8 in enhancing defect detection accuracy, reducing missed detection rates, and decreasing model parameter counts and computational demands, thus meeting the requirements of online defect detection for brake pipe ends surfaces. [ABSTRACT FROM AUTHOR]

Although approaches for the online surface detection of automotive pipelines exist, low defect area rates, small-sample and long-tailed data, and the difficulty of detection due to the variable morphology of defects are three major problems faced when using such methods. In order to solve these problems, this study combines traditional visual detection methods and deep neural network technology to propose a transfer learning multi-channel fusion decision network without significantly increasing the number of network layers or the structural complexity. Each channel of the network is designed according to the characteristics of different types of defects. Dynamic weights are assigned to achieve decision-level fusion through the use of a matrix of indicators to evaluate the performance of each channel's recognition ability. In order to improve the detection efficiency and reduce the amount of data transmission and processing, an improved ROI detection algorithm for surface defects is proposed. It can enable the rapid screening of target surfaces for the high-quality and rapid acquisition of surface defect images. On an automotive pipeline surface defect dataset, the detection accuracy of the multi-channel fusion decision network with transfer learning was 97.78% and its detection speed was 153.8 FPS. The experimental results indicate that the multi-channel fusion decision network could simultaneously take into account the needs for real-time detection and accuracy, synthesize the advantages of different network structures, and avoid the limitations of single-channel networks. [ABSTRACT FROM AUTHOR]

Defect detection is essential in the steel production process. Recent years have seen significant advancements in steel surface defect detection based on deep learning methods, notably exemplified by the YOLO series models capable of precise and rapid detection. However, challenges arise due to the high complexity of surface textures on steel and the low recognition rates for minor defects, making real-time and accurate detection difficult. This study introduces Mobile-YOLO-SDD (Steel Defect Detection), a lightweight YOLO-based model designed with high accuracy for real-time steel defect detection. Firstly, based on the effective YOLOv5 algorithm for steel defect detection, the backbone network is replaced with MobileNetV2 to reduce the model size and computational complexity. Then, the ECA (Efficient Channel Attention) module was integrated into the C3 module to reduce the number of parameters further while maintaining the defect detection rate in complex backgrounds. Finally, the K-Means++ algorithm regenerates anchor boxes and determines optimal sizes, enhancing their adaptability to actual targets. Experimental results on NEU-DET data demonstrate that the improved algorithm achieves a 60.6% reduction in model size, a 60.8% reduction in FLOPs, and a 1.8% improvement in mAP compared to YOLOv5s. These results confirm the effectiveness of Mobile-YOLO-SDD and lay the foundation for subsequent lightweight deployment of steel defect detection models. [ABSTRACT FROM AUTHOR]

In this study, we propose a reparameterization You-Only-Look-Once v5 (YOLOv5) algorithm model for strip-steel surface defect detection to address low precision and poor timeliness in traditional methods. The proposed model introduces a re-parameterized VGG Light module, an enhanced bidirectional feature pyramid network feature structure, and a bounding box regression loss function fused with a normalized Gaussian-Wasserstein distance metric to improve small-target-defect detection accuracy. The experimental findings reveal a mean average precision (mAP) of 82.1% on the NEU-DET dataset, representing a notable improvement of 4.1% over the baseline YOLOv5s algorithm. Furthermore, the proposed algorithm model demonstrates superior detection accuracy compared with other prevalent object detection models and effectively mitigates challenges such as false detections and missed detections of small targets. Notably, it achieves an impressive detection speed of 68 FPS, affirming its efficacy in real-time applications. [ABSTRACT FROM AUTHOR]

Traditional industrial surface defect detection method often employs CCD/CMOS cameras, but they are unable to detect the minute defects on aluminum substrates in highly dynamic industrial scenes due to their nature. Event camera is a novel high-resolution vision sensor that measures per-pixel brightness changes in an asynchronous manner and outputs as event information flow (EIF). Small and weak defects on aluminum substrate can be captured by event camera effectively, but the EIF contains a large amount of noise, making it difficult to perform accurate and high-precision defect detection. To address this problem, we propose a frame aggregation method to realize good event information flow processing, and then use an improved circle detection method to locate the aluminum substrate in each frame, removing abundant event information outside the aluminum substrate. Subsequently, we enhance the event signals under different frames based on optical flow tracking using multiple features, and construct a semi-supervised detector based on pseudo-labels to achieve high-precision defect localization. Finally, considering the small inter-class differences in defects on the surface of aluminum substrates, we construct a defect class corrector based on ensemble learning to enhance the ability to determine defect classes, achieving high-precision automatic quality inspection of aluminum substrate surfaces. The performance of our method is compared with other advanced methods based on event camera data of aluminum substrates in real industrial scenarios. The experimental results show that our method has improved the detection accuracy by ∼10% and the classification accuracy by ∼25% compared to the original state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Donghua University (Natural Science Edition) is the property of Journal of Donghua University (Natural Science) Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of Test & Measurement Technology is the property of Publishing Center of North University of China and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Machine Tool & Hydraulics is the property of Guangzhou Mechanical Engineering Research Institute (GMERI) and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The segmentation of surface defects in lithium batteries is crucial for enhancing the overall quality of the production process. However, the severe foreground–background imbalance in surface images of lithium batteries, along with the irregular shapes and random distribution of foreground regions, poses significant challenges for defect segmentation. Based on these observations, this paper focuses on the separation of foreground and background in surface defect images of lithium batteries and proposes a novel Mask Space Optimization Transformer (MSOFormer) for semantic segmentation of these images. Specifically, the Mask Boundary Loss (MBL) module in our model provides more efficient supervision during training to enhance the accuracy of the mask computation within the mask attention mechanism, thereby improving the model's performance in separating foreground and background. Additionally, the Dynamic Spatial Query (DSQ) module allocates spatial information of the image to each query, enhancing the model's sensitivity to the positions of small foreground targets in various scenes. The Efficient Pixel Decoder (EPD) ensures deformable receptive fields for irregularly shaped foregrounds while further improving the model's performance and efficiency. Experimental results demonstrate that our method outperforms other state-of-the-art methods in terms of mean Intersection over Union (mIoU). Specifically, our approach achieves an mIoU of 84.18% on the lithium battery surface defect test set and 85.53% and 87.05% mIoUs on two publicly available defect test sets with similar defect characteristics to lithium batteries. [ABSTRACT FROM AUTHOR]

Compared to the surface defect detection of industrial products produced according to specified processes, the detection of surface defects in naturally grown red jujubes poses unique and significant challenges for researchers. The high diversity of surface defects, subtle distinctions from the background, low contrast, varying scales, and the presence of high levels of noise in images are among the factors that greatly amplify the complexity of defect detection tasks. Existing methods show some deficiencies in addressing these issues, mainly due to insufficient feature extraction capabilities and overly complex network structures, leading to limitations in model efficiency and practical application performance. To tackle the challenges associated with red jujube surface defect detection, this study proposes an optimized Tiny Vision Transformer (TinyViT) network structure, named RJ-TinyViT. This method refines the TinyViT-5 m network structure to reduce network burden and introduces an improved Multi-Kernel Block (MK Block) and an improved Mobile Inverted Bottleneck Convolution Block (MBConv Block) to enhance feature extraction capabilities. Additionally, we have integrated the Coordinate Attention (CA) module to enhance the model's capacity for recognizing and focusing on features of surface defects on red jujubes. Experimental results show that RJ-TinyViT achieved a classification accuracy of 93.38%, marking an improvement of 1.84% over the original TinyViT network. At the same time, its Floating-point Operations (FLOPs) and Parameters (Params) were reduced to 58.97% and 39.84% of the original TinyViT network, respectively. These results not only demonstrate that RJ-TinyViT achieves model lightweighting while maintaining high accuracy but also highlight its value in practical industrial applications. [ABSTRACT FROM AUTHOR]

Strip steel is extensively utilized in industries such as automotive manufacturing and aerospace due to its superior machinability, economic benefits, and adaptability. However, defects on the surface of steel strips, such as inclusions, patches, and scratches, significantly affect the performance and service life of the product. Therefore, the salient object detection of surface defects on strip steel is crucial to ensure the quality of the final product. Many factors, such as the low contrast of surface defects on strip steel, the diversity of defect types, complex texture structures, and irregular defect distribution, hinder existing detection technologies from accurately identifying and segmenting defect areas against complex backgrounds. To address the above problems, we propose a novel detector called S3D-SOD for the salient object detection of strip steel surface defects. For the encoding stage, a residual self-attention block is proposed to explore semantic information cues of high-level features to locate and guide low-level feature information. In addition, we apply a general residual channel and spatial attention to low-level features, enabling the model to adaptively focus on the key channels and spatial areas of feature maps with high resolutions, thereby enhancing the encoder features and accelerating the convergence of the model. For the decoding stage, a simple residual decoder block with an upsampling operation is proposed to realize the integration and interaction of feature information between different layers. Here, the simple residual decoder block is used for feature integration due to the following observation: backbone networks like ResNet and the Swin Transformer, after being pretrained on the large dataset ImageNet and then fine-tuned on a smaller dataset for strip steel surface defects, are capable of extracting feature maps that contain both general image features and the specific characteristics required for the salient object detection of strip steel surface defects. The experimental results on the SD-saliency-900 dataset show that S3D-SOD is better than advanced methods, and it has strong generalization ability and robustness. [ABSTRACT FROM AUTHOR]

In the age of intelligent manufacturing, surface defect detection plays a pivotal role in the automated quality control of industrial products, constituting a fundamental aspect of smart factory evolution. Considering the diverse sizes and feature scales of surface defects on industrial products and the difficulty in procuring high-quality training samples, the achievement of real-time and high-quality surface defect detection through artificial intelligence technologies remains a formidable challenge. To address this, we introduce a defect detection approach grounded in the Fast Denoising Probabilistic Implicit Models. Firstly, we propose a noise predictor influenced by the spectral radius feature tensor of images. This enhancement augments the ability of generative model to capture nuanced details in non-defective areas, thus overcoming limitations in model versatility and detail portrayal. Furthermore, we present a loss function constraint based on the Perron-root. This is designed to incorporate the constraint within the representational space, ensuring the denoising model consistently produces high-quality samples. Lastly, comprehensive experiments on both the Magnetic Tile and Market-PCB datasets, benchmarked against nine most representative models, underscore the exemplary detection efficacy of our proposed approach. [ABSTRACT FROM AUTHOR]

The steel manufacturing process is inherently continuous, meaning that if defects are not effectively detected in the initial stages, they may propagate through subsequent stages, resulting in high costs for corrections in the final product. Therefore, detecting surface defects and obtaining segmentation information is critical in the steel manufacturing industry to ensure product quality and enhance production efficiency. Specifically, segmentation information is essential for accurately understanding the shape and extent of defects, providing the necessary details for subsequent processes to address these defects. However, the time-consuming and costly process of generating segmentation annotations poses a significant barrier to practical industrial applications. This paper proposes a cost-efficient segmentation labeling framework that combines deep learning-based anomaly detection and label enhancement to address these challenges in the steel manufacturing process. Using ResNet-50, defects are classified, and faster region convolutional neural networks (faster R-CNNs) are employed to identify defect types and generate bounding boxes indicating the defect locations. Subsequently, recursive learning is performed using the GrabCut algorithm and the DeepLabv3+ model based on the generated bounding boxes, significantly reducing annotation costs by generating segmentation labels. The proposed framework effectively detects defects and accurately defines them, even in randomly collected images from the steel manufacturing process, contributing to both quality control and cost reduction. This study presents a novel approach for improving the quality of the steel manufacturing process and is expected to enhance overall efficiency in the steel manufacturing industry. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Chongqing University of Technology (Natural Science) is the property of Chongqing University of Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Steel, as a crucial material extensively used in various fields, has a critical impact on the determination of the stability and reliability of engineering structures. Nevertheless, because of inevitable factors in manufacturing, transportation, and other processes, steel may exhibit various surface defects during production and handling. To address these defects, the investigation puts forward a resilient machine-learning method for steel surface defect detection based on lightweight convolution. First, to reduce redundant features, complexity, and computational cost, the Spatial and Channel Reconstruction Convolution (ScConv) module is added before the Spatial Pyramid Pooling-Fast (SPPF) within the YOLOv8n's backbone network. Second, in the Neck layer, lightweight convolution GSConv is used to replace the convolutional modules, and the efficient cross-stage partial network (CSP) module, VoV-GSCSP is substituted for the C2f module to alleviate the model burden while maintaining accuracy. Then, to focus on important information related to the current task, the Coordinate Attention module is added to the Neck layer. Finally, the activation function of YOLOv8n has been swapped for the Leaky Rectified Linear Unit (LeakyReLU) to effectively address issues such as gradient vanishing and overfitting. The method achieved a mean Average Precision (mAP) of 77.7% on the NEU-DET dataset, which is an improvement of 4.7% over the original YOLOv8n. Additionally, the frames per second (FPS) reached 17.36 f/s, representing a 5.79 f/s increase compared to the original YOLOv8n. On the GC10-DET dataset, mAP improves by 5.5%, with a FPS of 15.63 f/s. A plethora of experimentation on both datasets illustrates the method's robustness, meeting the precision criteria for detecting metal defects. [ABSTRACT FROM AUTHOR]

Surface defect detection of hot-rolled steel strip is a crucial process in its production. Aiming at the problem of poor detection accuracy caused by complex shape of defect features and imbalance between object scale and feature scale, this paper proposes a multi-scale sensing and multi-dimensional feature enhancement algorithm. Firstly, a multi-branch residual aggregation module (MRAM) is proposed, which uses the interactive features of different receptive fields between layers to capture global and local texture features, improving the feature extraction capability for defects of different scales. Then, the multi-dimensional attention enhancement head (MAEH) is proposed, which integrates feature information from different prediction branches and realises feature information enhancement in scale, space and channel dimensions for the aggregated information to improve the accuracy and robustness of steel surface defect detection. The results show that the proposed algorithm achieves 82.7% mAP and 89 FPS on the NEU-DET dataset, and 73.3% mAP and 94 FPS on the GC10-DET dataset. Compared with state-of-the-art algorithms, the proposed algorithm has the highest mAP, strong generalisation ability, and good real-time performance, which proves its effectiveness in detecting hot-rolled steel strip surface defects. [ABSTRACT FROM AUTHOR]