Search

Featured Application: An improved YOLOv4 algorithm for PCB surface defect detection can achieve higher detection accuracy and faster detection speed with lower memory consumption and fewer multiply–accumulate operations compared with the cutting-edge YOLOv4. Surface defect detection for printed circuit board (PCB) is indispensable for managing PCB production quality. However, automatic detection of PCB surface defects is still a challenging task because, even within the same category of surface defect, defects present great differences in morphology and pattern. Although many computer vision-based detectors have been established to handle these problems, current detectors struggle to achieve high detection accuracy, fast detection speed and low memory consumption simultaneously. To address those issues, we propose a cost-effective deep learning (DL)-based detector based on the cutting-edge YOLOv4 to detect PCB surface defect quickly and efficiently. The YOLOv4 is improved upon with respect to its backbone network and the activation function in its neck/prediction network. The improved YOLOv4 is evaluated with a customized dataset, collected from a PCB factory. The experimental results show that the improved detector achieved a high performance, scoring 98.64% on mean average precision (mAP) at 56.98 frames per second (FPS), outperforming the other compared SOTA detectors. Furthermore, the improved YOLOv4 reduced the parameter space of YOLOv4 from 63.96 M to 39.59 M and the number of multiply-accumulate operations (Madds) from 59.75 G to 26.15 G. [ABSTRACT FROM AUTHOR]

Surface defect detection of an automobile wheel hub is important to the automobile industry because these defects directly affect the safety and appearance of automobiles. At present, surface defect detection networks based on convolutional neural network use many pooling layers when extracting features, reducing the spatial resolution of features and preventing the accurate detection of the boundary of defects. On the basis of DeepLab v3+, we propose a semantic segmentation network for the surface defect detection of an automobile wheel hub. To solve the gridding effect of atrous convolution, the high-resolution network (HRNet) is used as the backbone network to extract high-resolution features, and the multi-scale features extracted by the Atrous Spatial Pyramid Pooling (ASPP) of DeepLab v3+ are superimposed. On the basis of the optical flow, we decouple the body and edge features of the defects to accurately detect the boundary of defects. Furthermore, in the upsampling process, a decoder can accurately obtain detection results by fusing the body, edge, and multi-scale features. We use supervised training to optimize these features. Experimental results on four defect datasets (i.e., wheels, magnetic tiles, fabrics, and welds) show that the proposed network has better F1 score, average precision, and intersection over union than SegNet, Unet, and DeepLab v3+, proving that the proposed network is effective for different defect detection scenarios. [ABSTRACT FROM AUTHOR]

Nowadays there are numerous problems for which use of a multi-objective in image classification would be desirable although, unfortunately, the number of samples is too low. In these situations, higher level classifications could also work (for example, in surface defect detection, it is important to identify the defect, but it could also be useful to detect whether or not the object has a defect). To this end, we present a methodology called BoDoC which allows to improve this classification. To evaluate the methodology, we have created a new dataset, obtained from a foundry, to detect surface errors in casting pieces with 2 different defects: (i) inclusions, (ii) coldlaps and (iii) misruns. We also present a collection of techniques to select featu res from the images. We prove that our methodology improves the direct classification results in real world scenarios, with 91.305% precision. [ABSTRACT FROM AUTHOR]

Copyright of Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji is the property of Gazi University 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.)

As a crucial load-bearing component of the cable-stayed bridge, the cable requires surface defect detection to maintain its safety. The current deep-learning-based stay-cable surface defect detection still faces challenges, including insufficient actual defect samples, cumbersome labeling processes, and an effective method for further evaluation and classification of defects. To address these problems, an unsupervised vision transformer-based lightweight reconstruction network called ViTR-Net is proposed, which only uses normal stay-cable surface images to achieve end-to-end localization of defect areas in the image. ViTR-Net is a generative-based method that uses a hybrid network of convolutional neural networks and visual transformers as a feature extraction network. A feature correction fusion module is designed by integrating differential convolution and attention mechanisms to compensate for potential detail loss caused by feature extraction. Subsequently, a severity index for defects is established to assess the severity of various defects, and the detected defects are adaptively classified using the affinity propagation clustering algorithm. ViTR-Net achieves the highest defect detection accuracy and exquisite defect segmentation results on a realistic cable defect dataset through experimental validation. The proposed model can realize the surface defect assessment and adaptive classification of the stay cable, which has a good application prospect in the actual maintenance process of the stay cable. • Unsupervised vision transformer-based lightweight reconstruction network. • Feature correction fusion method combining differential convolution and attention. • Address the issue of limited defect samples and labor-intensive manual labeling. • Introduce a metric for evaluating different defect severity levels. • Realize surface defect assessment and adaptive classification of the stay cable. [ABSTRACT FROM AUTHOR]

Surface defects in steel manufacturing can compromise product quality and safety. Detecting these diverse and complex defects under industrial conditions is challenging. This paper proposes WSS-YOLO, a YOLOv8-based model, for accurate defect detection on industrial steel. Firstly, the dynamic non-monotonic focusing mechanism based on WIoU loss was employed to focus on anchor boxes with ordinary quality, thereby improving the overall performance of the detector. Secondly, the C2f-DSC module based on dynamic snake convolution is designed to enable the model to adaptively adjust the receptive field. Finally, GSConv and VOV-GSCSP modules are introduced into the neck network to reduce the computational complexity and parameter quantity while ensuring the accuracy of the model. This paper conducts extensive experiments on the public datasets NEU-DET and GC10-DET, achieving mAP of 82.3% and 72.0%, respectively, outperforming other excellent models. Moreover, the effectiveness of the proposed method in industrial defect detection is also validated. • Utilizing the WIoU loss to mitigate the harm of low-quality samples. • Designing the C2f-DSC module, employing dynamic snake convolution (DSC) to enhance the model's feature extraction capability. • Implementing a slim-neck paradigm design to achieve lightweight design while maintaining detection accuracy. • Validating the effectiveness of the proposed methods through extensive experimental results. [ABSTRACT FROM AUTHOR]

Deep convolutional neural networks (CNNs) have been widely used in surface defect detection. However, no CNN architecture is suitable for all detection tasks and designing effective task-specific architectures requires considerable effort. The neural architecture search (NAS) technology makes it possible to automatically generate adaptive data-driven networks. Here, we propose a new method called NAS-ASDet to adaptively design network for surface defect detection. First, a refined and industry-appropriate search space that can adaptively adjust the feature distribution is designed, which consists of repeatedly stacked basic novel cells with searchable attention operations. Then, a progressive search strategy with a deep supervision mechanism is used to explore the search space faster and better. This method can design high-performance and lightweight defect detection networks with data scarcity in industrial scenarios. The experimental results on four datasets demonstrate that the proposed method achieves superior performance and a relatively lighter model size compared to other competitive methods, including both manual and NAS-based approaches. [ABSTRACT FROM AUTHOR]

Statistical texture features are essential for metal surface defect detection. However, low contrast and cluttered backgrounds exacerbate the intrinsic blurriness and variability of defect textures, hampering accurate identification and localization. To tackle these challenges, we propose a Statistical Texture Feature Enhancement Network (STFE-Net) based on YOLO model, a multi-stage approach based on pooling attention mechanism and cross-scale shallow feature reinforcement. Firstly, we enhance the image contrast and propose a Stacked Pooling Attention Residual (SPA-Res) module to capture better texture features; then, we embed the Shallow Feature Weighted (SFW) branches to achieve cross-layer shallow feature reinforcement and add feature layer (AFL) to extract rich semantic features; and finally, we propose an Efficient Complete Intersection over Union (ECIoU) loss function and employ four detection heads to achieve precise localization of defects. On the NEU-DET dataset, our model achieves the state-of-the-art performance. Especially, the mean Average Precision (mAP@IoU = 0.5) of our model is 82.3 %, which is 6.2 % higher than baseline network. With the input resolution of 640 × 640, STFE-Net achieves 54 FPS inference speed with a GeForce GTX 1080. In addition, the experimental results on the DAGM dataset and GC10-DET dataset demonstrate that the proposed model has excellent generalization ability. [ABSTRACT FROM AUTHOR]

In practical industrial applications, the inference speed of deep learning models directly affects the efficiency of industrial production. Therefore, the lightweight real-time detection method of surface defects is an essential task in the industrial process. We need to achieve a favorable balance between efficiency and accuracy since the rising demand for production efficiency. However, most of the existing pixel-level detection methods 1) often adopt huge computational overhead to learn rich features, resulting in slow inference speed and 2) show a performance degradation when applied to different industrial surface defect scenarios. To this end, we propose an efficient targeted design (ETD) for real-time defect detection of surface defects. It consists of two branches: (i) an efficient feature enhancement branch, with global aggregation module (GAM) and cross-scale guide module (CGM) to gradually enhance defect features, and (ii) an edge posterior branch, with verification module (VM) and scale interaction module (SIM) to implicitly guide the boundary details of defects. Specifically, while inheriting this framework, we reconsider the relationship between precision, parameters, and speed so that our model can be applied to different industrial scenarios. Extensive experimental results on four datasets indicate that ETD outperforms other leading saliency detection methods. Meanwhile, our method ETD-S achieves 347 FPS on ESDIs-SOD dataset, 254 FPS on Crack500 dataset, 227 FPS on NRSD-MN dataset and 273 FPS on DAGM dataset. Additionally, we conduct real-time analysis of ETD on an intelligent paradigm for industrial surface defect detection, further demonstrating its efficacy in practical scenarios. ETD demonstrates effective detection performance while achieving a lightweight architecture, which can be implemented using various deep learning frameworks, showcasing substantial potential for real-time surface defect detection. The source code and dataset are publicly available at https://github.com/VDT-2048/ETD. • Proposed ETD focuses more on the trade-off between efficiency and accuracy. • Proposed two-branch strategy effectively reduces the network inference overhead. • Proposed efficient feature enhancement branch aggregates multi-scale features. • Proposed ETD is verified and analyzed in a real industrial surface inspection scene. • Two publicly available pixel-level annotated steel strip surface defect datasets are constructed. [ABSTRACT FROM AUTHOR]

Industrial product surface defect detection is very important to guarantee high product quality and production efficiency. In this work, we propose a regression and classification based framework for generic industrial defect detection. Specifically, the framework consists of four modules: deep regression based detection model, pixel-level false positive reduction, connected component analysis and deep network for defect type classification. To train the detection model, we propose a high performance deep network structure and an algorithm to generate label data to capture the defect severity information from data annotation. We have tested the method on two public benchmark datasets, AigleRN and DAGM2007, and an in-house capacitor image dataset. The results have shown that our method can achieve the state-of-the-art performance in terms of detection accuracy and efficiency.

Surface defect detection is a critical task in product quality assurance for manufacturing lines. The deep learning-based methods recently developed for defect detection are typically trained using a supervised learning strategy and large defect sample sets. Conventional methods often require additional pixel-level labeling or bounding boxes to predict the location of defects. However, the number of required samples and the time-intensive annotation process limits the practical use of these algorithms. As such, this study proposes a weakly supervised detection framework in which a CNN model is trained to identify surface cracks in motor commutators. The model was trained using small subsets of defect samples (~5-30) and does not require a pre-trained network. This approach consists of localization and decision networks that simultaneously predict both the location and probability of defects. A new loss function was also developed to identify abnormal regions in a sample with accessible image-level labels. A collaboration learning strategy was then applied to utilize the loss function and compensate for imbalances at the pixel level. Experimental results using a small number of image-level training labels from a real industrial dataset exhibited a 99.5% recognition accuracy, which is comparable to relevant methods using pixel-level labels.

Surface defect detection aims to classify and locate a certain defect that exists in the target surface area. It is an important part of industrial quality inspection. Most of the research on surface defect detection are currently based on convolutional neural networks (CNNs), which are more concerned with local information and lack global perception. Thus, CNNs are unable to effectively extract the defect features. In this paper, a defect detection method based on the Swin transformer is proposed. The structure of the Swin transformer has been fine-tuned so that it has five scales of output, making it more suitable for defect detection tasks with large variations in target size. A bi-directional feature pyramid network is used as the feature fusion part to efficiently fuse to the extracted features. The focal loss is used as a loss function to weight the hard- and easy-to-distinguish samples, potentially making the model fit the surface defect data better. To reduce the number of parameters in the model, a shared detection head was chosen for result prediction. Experiments were conducted on the flange surface defect dataset and the steel surface defect dataset, respectively. Compared with the classical CNNs target detection algorithm, our method improves the mean average precision (mAP) by about 15.4%, while the model volume and detection speed are essentially the same as those of the CNNs-based method. The experimental results show that our proposed method is more competitive compared with CNNs-based methods and has some generality for different types of defects.

Object detection is widely used in the industry for visual inspection of defects. But it fails to fully reflect the defects on the smooth surface due to the difficulty in illuminating, and the high performance of defect detection tends to be a challenge. In this paper, we proposed a novel defect detection method for laptop panels. On the one hand, we utilized the wrapped phase map obtained by phase-shift reflection fringes for defect detection and used the DenseFuse network to fuse the wrapped phase map with the fringe modulation map to optimize the dataset. On the other hand, we proposed an improved deep learning-based network Vovecwnet. The experiments indicated that our image-based multi-information fusion system could achieve excellent performance in defect detection on the smooth surface (e.g., laptop panels surface), with high accuracy of 97.5% and a few reference time 0.0762s on GTX 2080ti GPU. Thus, our proposed method outperformed and tackled the challenges encountered in the defect detection of smooth surfaces. [ABSTRACT FROM AUTHOR]

In the integrated circuit (IC) packaging, the surface defect detection of flexible printed circuit boards (FPCBs) is important to control the quality of IC. Although various computer vision (CV)-based object detection frameworks have been widely used in industrial surface defect detection scenarios, FPCB surface defect detection is still challenging due to non-salient defects and the similarities between diverse defects on FPCBs. To solve this problem, a decoupled two-stage object detection framework based on convolutional neural networks (CNNs) is proposed, wherein the localization task and the classification task are decoupled through two specific modules. Specifically, to effectively locate non-salient defects, a multi-hierarchical aggregation (MHA) block is proposed as a location feature (LF) enhancement module in the defect localization task. Meanwhile, to accurately classify similar defects, a locally non-local (LNL) block is presented as a SEF enhancement module in the defect classification task. What is more, an FPCB surface defect detection dataset (FPCB-DET) is built with corresponding defect category and defect location annotations. Evaluated on the FPCB-DET, the proposed framework achieves state-of-the-art (SOTA) accuracy to 94.15% mean average precision (mAP) compared with the existing surface defect detection networks. Soon, source code and dataset will be available at https://github.com/SCUTyzy/decoupled-two-stage-framework. [ABSTRACT FROM AUTHOR]

This article proposes a fast and accurate network for surface defect detection, termed SDDNet. SDDNet mainly addresses two challenging issues—large texture variation and small size of defects—by introducing two modules: feature retaining block (FRB) and skip densely connected module (SDCM). FRB fuses multiple pyramidal feature maps with different resolutions and is plugged on the top of pooling layers, aiming to preserve the texture information, which may be lost because of downsampling. SDCM is designed to propagate the fine-grained details from low- to high-level feature maps for better prediction of defects, especially small defects. Extensive experiments conducted on the publicly available data sets NEU-DET (88.8% mAP), DAGM (99.1% mAP), and Magnetic-Tile (93.4% mAP) have demonstrated the effectiveness of the proposed SDDNet and its feasibility for real-time industrial applications. [ABSTRACT FROM AUTHOR]

Surface defects have an adverse effect on the quality of industrial products, and vision-based defect detection is widely researched due to its objective and stable performance. However, the task is still challenging due to diversified defect types and complex background texture. The robust principal component analysis (RPCA) has proven applicable in defect inspection by regarding nondefective background as the low-rank part and defective area as the sparse part. However, such methods cannot sufficiently detect defects due to complex cluttered background, noise interference, and limited features available. To address these issues, in this article, we proposed an unsupervised surface defect detection method based on nonconvex total variation (TV) regularized RPCA with kernelization, named KRPCA-NTV. Specifically, the kernel method is integrated into RPCA to better handle complex cluttered background lying in a nonstrict low-rank subspace. Furthermore, nonconvex TV regularization is introduced to prevent the noise pixel from being separated into the defect region; meanwhile, nonconvex optimization promotes higher solution accuracy. In addition, the kernel canonical correlation analysis (KCCA) is utilized to fuse complementary features for boosting feature representation ability. To demonstrate the superiority and robustness of the proposed method, we compare it with the state of the art on five defect data sets; the results show that the proposed method outperforms competing methods in terms of accuracy and generalizability. [ABSTRACT FROM AUTHOR]

Surface defect detection is very important for the quality control of product and routine maintenance of facilities, but it is still a big challenge due to the diversity and complexity of defects and environmental factors. To improve the accuracy of defect detection, we proposed a pixel-level segmentation network based on deep feature fusion for surface defect detection. The network adopts encoder–decoder structure, and it extracts low-, middle-, and high-level features via ResNet50 first. Second, by fusing adjacent feature maps at all levels and integrating the highest level feature map, multilevel feature aggregation module makes all feature maps contain context information and more details of defects. Then, multibranch decoder adopts attention modules and a multibranch structure to recover the details of defects gradually and improve the accuracy of defects segmentation. Finally, the segmentation result is produced by fusion of all branches outputs. We have evaluated the proposed network on three public data sets: MT, RSDD, and CFD. The results indicate that our proposed method outperforms the other compared methods in terms of F-measure and intersection of union (MT:73.7%, RSDD:85%, CFD:60.1%). [ABSTRACT FROM AUTHOR]

A fresh-cut cauliflower surface defect detection and classification model based on a convolutional neural network with transfer learning is proposed to address the low efficiency of the traditional manual detection of fresh-cut cauliflower surface defects. Four thousand, seven hundred and ninety images of fresh-cut cauliflower were collected in four categories including healthy, diseased, browning, and mildewed. In this study, the pre-trained MobileNet model was fine-tuned to improve training speed and accuracy. The model optimization was achieved by selecting the optimal combination of training hyper-parameters and adjusting the different number of frozen layers; the parameters downloaded from ImageNet were optimally integrated with the parameters trained on our own model. A comparison of test results was presented by combining VGG19, InceptionV3, and NASNetMobile. Experimental results showed that the MobileNet model’s loss value was 0.033, its accuracy was 99.27%, and the F1 score was 99.24% on the test set when the learning rate was set as 0.001, dropout was set as 0.5, and the frozen layer was set as 80. This model had better capability and stronger robustness and was more suitable for the surface defect detection of fresh-cut cauliflower when compared with other models, and the experiment’s results demonstrated the method’s feasibility.

Metallic surface defect detection is critical to ensure the quality of industrial products. Recently, human-advanced surface defect detection algorithms have been proposed. Most of these algorithms rely on convolutional neural networks (CNN) and an anchoring scheme. However, a convolution unit only samples the input feature maps at fixed shapes and locations. Similarly, a set of anchors are uniformly predefined with fixed scales and shapes, which increases the difficulties of bounding box regression. Therefore, we propose an adaptive convolution and anchor network for metallic surface defect detection, named ACA-Net. Specifically, an adaptive convolution and anchor (ACA) module is proposed, which mainly consists of adaptive convolution and an adaptive anchor. Firstly, an adaptive convolution module (ACM) is designed, which adaptively determines the location and shape of each convolution unit. In addition, a multi-scale feature adaptive fusion (MFAF) is proposed, which is used in ACM to extract and integrate multi-scale features. Then, an adaptive anchor module (AAM) is proposed to yield more suitable anchor boxes by adaptively adjusting shapes. Extensive experiments on NEU-DET dataset and GC10 dataset validate the performance of the proposed approach. ACA-Net achieves 1.8% on NEU-DET dataset higher Average Precision (AP) than GA-RetinaNet. Furthermore, the proposed ACA module is also adopted in GA-Faster R-CNN, improving the AP by 1.2% on NEU-DET dataset.

Aiming at the problems of low efficiency and poor accuracy in conventional surface defect detection methods for aero-engine components, a surface defect detection model based on an improved YOLOv5 object detection algorithm is proposed in this paper. First, a k-means clustering algorithm was used to recalculate the parameters of the preset anchors to make them match the samples better. Then, an ECA-Net attention mechanism was added at the end of the backbone network to make the model pay more attention to feature extraction from defect areas. Finally, the PANet structure of the neck network was improved through its replacement with BiFPN modules to fully integrate the features of all scales. The results showed that the mAP of the YOLOv5s-KEB model was 98.3%, which was 1.0% higher than the original YOLOv5s model, and the average inference time for a single image was 2.6 ms, which was 10.3% lower than the original model. Moreover, compared with the Faster R-CNN, YOLOv3, YOLOv4 and YOLOv4-tiny object detection algorithms, the YOLOv5s-KEB model has the highest accuracy and the smallest size, which make it very efficient and convenient for practical applications.

The existing seal ring surface defect detection methods for aerospace applications have the problems of low detection efficiency, strong specificity, large fine-grained classification errors, and unstable detection results. Considering these problems, a fine-grained seal ring surface defect detection algorithm for aerospace applications is proposed. Based on analysis of the stacking process of standard convolution, heat maps of original pixels in the receptive field participating in the convolution operation are quantified and generated. According to the generated heat map, the feature extraction optimization method of convolution combinations with different dilation rates is proposed, and an efficient convolution feature extraction network containing three kinds of dilated convolutions is designed. Combined with the O-ring surface defect features, a multiscale defect detection network is designed. Before the head of multiscale classification and position regression, feature fusion tree modules are added to ensure the reuse and compression of the responsive features of different receptive fields on the same scale feature maps. Experimental results show that on the O-rings-3000 testing dataset, the mean condition accuracy of the proposed algorithm reaches 95.10% for 5 types of surface defects of aerospace O-rings. Compared with RefineDet, the mean condition accuracy of the proposed algorithm is only reduced by 1.79%, while the parameters and FLOPs are reduced by 35.29% and 64.90%, respectively. Moreover, the proposed algorithm has good adaptability to image blur and light changes caused by the cutting of imaging hardware, thus saving the cost. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer-Aided Design & Computer Graphics / Jisuanji Fuzhu Sheji Yu Tuxingxue Xuebao is the property of Gai Kan Bian Wei Hui 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 quality, wear and safety of metal structures can be controlled effectively, provided that surface defects, which occur on metal structures, are detected at the right time. Over the past 10 years, researchers have proposed a number of neural network architectures that have shown high efficiency in various areas, including image classification, segmentation and recognition. However, choosing the best architecture for this particular task is often problematic. In order to compare various techniques for detecting defects such as “scratch abrasion”, we created and investigated U-Net-like architectures with encoders such as ResNet, SEResNet, SEResNeXt, DenseNet, InceptionV3, Inception-ResNetV2, MobileNet and EfficientNet. The relationship between training validation metrics and final segmentation test metrics was investigated. The correlation between the loss function, the DSC, IoU, Recall, Precision and F1 validation metrics and DSC test metrics was calculated. Recognition accuracy was analyzed as affected by the optimizer during neural network training. In the context of this problem, neural networks trained using the stochastic gradient descent optimizer with Nesterov momentum were found to have the best generalizing properties. To select the best model during its training on the basis of the validation metrics, the main test metrics of recognition quality (Dice similarity coefficient) were analyzed depending on the validation metrics. The ResNet and DenseNet models were found to achieve the best generalizing properties for our task. The highest recognition accuracy was attained using the U-Net model with a ResNet152 backbone. The results obtained on the test dataset were DSC=0.9304 and IoU=0.9122.

The surface defects' region of strip steel is small, and has various defect types and, complex gray structures. There tend to be a large number of false defects and edge light interference, which lead traditional machine vision algorithms to be unable to detect defects for various types of strip steel. Image detection techniques based on deep learning require a large number of images to train a network. However, for a dataset with few samples with category imbalanced defects, common deep learning neural network training tasks cannot be carried out. Based on rapid image preprocessing algorithms (improved gray projection algorithm, ROI image augmentation algorithm) and transfer learning theory, this paper proposes a set of processes for complete strip steel defect detection. These methods achieved surface rapid screening, defect feature extraction, sample dataset's category balance, data augmentation, defect detection, and classification. Through verification of the mixed dataset, composed of the NEU surface dataset and dataset in this paper, the recognition accuracy of the improved VGG19 network in this paper reached 97.8%. The improved VGG19 network performs slightly better than the baseline VGG19 in six types of defects, but the improved VGG19 performs significantly better in the surface seams defects. The convergence speed and accuracy of the improved VGG19 network were taken into account, and the detection rate was greatly improved with few samples and imbalanced datasets. This paper also has practical value in terms of extending its method of strip steel defect detection to other products. [ABSTRACT FROM AUTHOR]

Aiming at the problems of low accuracy and poor generalization of industrial aluminum surface defect detection, this paper proposes a multi task feature fusion detection method based on improved segmentation decision network. Firstly, aiming at the common pit defects in industry, it constructs the defect detection data set of this paper; secondly, on the basis of the original segmentation detection model,it also adds the aluminum area segmentation network to reduce the demand for samples in the network structure; in addition, the Squeeze-and-Excitation module is introduced to make the model focus more on the dimensional features related to aluminum defects and improve the detection accuracy of the network. Finally, a comparative experiment is carried out on the data set of dirty spots on the aluminum surface. The structure shows that the segmentation accuracy of the proposed method is improved by 1.6% and the classification accuracy is improved by 4.01% compared with the original model; compared with FCN and u-net network, the detection accuracy of image segmentation is 0. 76% higher than that of FCN and U-Net network, and 1. 12% higher than that of VGG and Resnet50 network. The experimental results have proved that this method has sufficient detection accuracy in the detection of dirty spots on the aluminum surface, and is stable and efficient. [ABSTRACT FROM AUTHOR]

Surface defect detection is a critical task in industrial production process. Nowadays, there are lots of detection methods based on computer vision and have been successfully applied in industry, they also achieved good results. However, achieving full automation of surface defect detection remains a challenge, due to the complexity of surface defect, in intraclass. While the defects between interclass contain similar parts, there are large differences in appearance of the defects. To address these issues, this article proposes a pyramid feature fusion and global context attention network for pixel-wise detection of surface defect, called PGA-Net. In the framework, the multiscale features are extracted at first from backbone network. Then the pyramid feature fusion module is used to fuse these features into five resolutions through some efficient dense skip connections. Finally, the global context attention module is applied to the fusion feature maps of adjacent resolution, which allows effective information propagate from low-resolution fusion feature maps to high-resolution fusion ones. In addition, the boundary refinement block is added to the framework to refine the boundary of defect and improve the result of the prediction. The final prediction is the fusion of the five resolutions fusion feature maps. The results of evaluation on four real-world defect datasets demonstrate that the proposed method outperforms the state-of-the-art methods on mean intersection of union and mean pixel accuracy (NEU-Seg: 82.15%, DAGM 2007: 74.78%, MT_defect: 71.31%, Road_defect: 79.54%). [ABSTRACT FROM AUTHOR]

Surface defects affect the quality and safety of oil seals. It is a challenge to detect such defects in a vision system because of the unequal reflection property of oil seal surfaces and low contrast between the defect and the background. This article proposes a visual detection method (VDM) for oil seal surface defects and outlines two key issues of VDMs. First, we present a superpixel segmentation algorithm based on the significant gray level variation in the radial direction of an oil seal surface image. This image is then divided into several ring belts. Subsequently, considering the reflection inequality and low contrast, we propose a new circumferential background difference algorithm based on the small variation along the circumferential direction of the image. This algorithm eliminates the influence of the reflection inequality and improves the contrast distinction between the defects and the background. The experimental results verify the effectiveness of the proposed method with a recall and precision as high as 95.2% and 86.8%, respectively. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNN) have achieved promising performance in surface defect detection recently. Although many CNN-based methods have been proposed, most of them are limited by the few samples available for training, and the imbalance of positive and negative samples. Hence, their detection performance needs to be further improved. To this end, we propose a multi-scale cascade CNN called MobileNet-v2-dense to detect defects more efficiently. Specifically, the multi-scale cascade structure used in our network can help capture the weak defect semantics that may be lost in the deep network. Then, we propose a novel asymmetric loss function to further improve detection performance. Lastly, a two-stage augmentation method effectively enlarges the training dataset. Experimental results show that, compared to the state-of-the-art, the area under the receiver-operating characteristic curve (AUC-ROC) score of our method increased by 0.16. [ABSTRACT FROM AUTHOR]

The appearance quality assessment based on defect inspection for painted car-body surfaces is an essential work to monitor and analyze the level of paint appearance quality. In the industrial application, there are some challenges, such as the huge and stereo skeleton of car bodies, a variety of irregular local surface areas, low visibility of defects due to tiny real size, and specular car-body surface. To overcome these problems, a lightweight online appearance quality assessment system (OAQAS) based on parallel deep learning is proposed, it includes two parts: 1) a vision inspection subsystem with distributed multi-camera image acquisition module and 2) an appearance quality evaluation subsystem (AQES) based on parallel TinyDefectRNet for evaluating the proposed painted surface grinding difficulty criteria. TinyDefectRNet is able to inspect relatively accurate defect size, although it is trained on a coarsely annotated data set. The OAQAS is implemented in an actual painting production line of a car factory, and the application results show that our OAQAS is far superior to the manual inspection in evaluation accuracy and time consumption. Moreover, our system is lightweight so that it is easy to be plugged into existing painting production lines without rebuilding or changing the inspection room. [ABSTRACT FROM AUTHOR]

A phased electromagnetic acoustic transducer (EMAT) array system has been developed for the detection and characterization of surface-breaking defects. An array of four linear coils that are individually controlled is used to generate a Rayleigh wave. The high-current electronics combined with the coil designs enables the array to generate either narrowband or broadband signals, and controlling the phase delay between the channels makes it possible to change the ultrasound wavelength without requiring the physical separation of the coils to be changed. The experimental results show that the four-coil phased array can generate a wavelength range from 3.0 to 11.7 mm. Surface-breaking defects were characterized using a transmit–receive setup with a broadband EMAT detector being used to detect the Rayleigh wave. Machined surface slots with different depths were used for technique validation. The results show that the array is sensitive to surface defects and a wide depth sensitivity range for defect sizing can be easily achieved by applying phasing to tune the wavelength of operation. A large increase in detection flexibility is immediately shown. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer-Aided Design & Computer Graphics / Jisuanji Fuzhu Sheji Yu Tuxingxue Xuebao is the property of Gai Kan Bian Wei Hui 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.)

A novel object rotation hyperspectral imaging system with the wavelength range of 468–950 nm for investigating round-shaped fruits was developed. This system was used to obtain the reflection spectra of jujubes for the application of surface defect detection. Compared to the traditional linear scan system, which can scan about 49% of jujube surface in one scan pass, this novel object rotation scan system can scan 95% of jujube surface in one scan pass. Six types of jujube skin condition, including rusty spots, decay, white fungus, black fungus, cracks, and glare, were classified by using hyperspectral data. Support vector machine (SVM) and artificial neural network (ANN) models were used to differentiate the six jujube skin conditions. Classification effectiveness of models was evaluated based on confusion matrices. The percentage of classification accuracy of SVM and ANN models were 97.3% and 97.4%, respectively. The object rotation scan method developed for this study could be used for other round-shaped fruits and integrated into online hyperspectral investigation systems. [ABSTRACT FROM AUTHOR]

Intelligent defect detection systems based on deep learning have great potential for industrial applications to ensure product quality. However, the success of such data-driven approaches relies heavily on well-established datasets that require considerable time and labor for fine-grained annotation. Actual industrial datasets are often roughly labeled with many inaccurate annotations, thus seriously hindering the performance of defect detection. In particular, the boundaries of surface defects are not only difficult to label reliably, but also to detect accurately. In this paper, a novel approach termed BTDNet (Boundary-guided Task Decomposition Network) is proposed to tackle the challenge posed by inaccurate annotations. BTDNet decomposes the inaccurate original annotation into two accurate annotations, introducing two reliable auxiliary tasks. These auxiliary tasks are supervised separately using the new annotations, enabling BTDNet to mitigate the negative impact of inaccurate annotations during training and learn robust features. An efficient statistics recalibration module is constructed in BTDNet to customize input features for different tasks, while an adaptive feature summation module is employed to merge the outputs of all tasks. As an essential part of BTDNet, both modules are designed with a strong focus on lightweight. Extensive experiments on three publicly available industrial datasets confirm the effectiveness of our method, showing that BTDNet can consistently achieve performance gains across different baselines and input scales. For example, BTDNet clearly improves the F1-score of DeepLabv3 and GCNet methods on the Magnetic-tile dataset by 5.77% and 5.40%, respectively. [Display omitted] • Unreliable annotations lead to poor defect detection, especially on boundaries. • Decomposing unreliable tasks on boundaries helps to learn robust features. • Simple but effective statistics recalibration module and adaptive summation module. • High focus on lightweight, introducing only negligible parameters and calculations. • Consistent performance improvements across different datasets and baselines. [ABSTRACT FROM AUTHOR]

Surface defect detection is a crucial inspection phase in ensuring industrial product quality and reliability, particularly in the context of metallic manufacturing components. While existing deep learning-based approaches have demonstrated some effectiveness, the design of certain network structures still lacks consideration for industrial scenarios. This paper proposed a complex metallic surface defect detection neural network which incorporates two innovatively designed modules along with other targeted enhancements. Firstly, the spatial pyramid pooling (SPP) structure is redesigned for richer global-local information fusion. A global feature fusion and redistribution module (GFAR) is designed to fully leverage feature information across different scales. With the combination of a novel global positional attention mechanism (GPAM), GFAR is capable of distinguishing defects from complex backgrounds. Moreover, the asymmetric convolution blocks (ACB) are employed to process the final features, enhancing the representational capacity for intricate defect shapes. Subsequently, the detection head is decoupled in both spatial and task domains to addresses the problems posed by inconsistent focus between classification and regression tasks. Finally, the reassignment of loss functions is undertaken for each respective task to enhance their alignment with the specific requirements of the defect detection task. Extensive ablation experiments are conducted on two steel surface defect datasets, NEU-DET and GC10-DET, to demonstrate the effectiveness of each module. The defects detection accuracy of our method respectively reaches 80.9 % and 84.8 % for the two datasets, which outperforms the majority of state-of-the-art detection neural networks. • A global feature fusion and redistribution (GFAR) module is proposed to strengthen cross-scale feature interactions. • A global positional attention mechanism (GPAM) is integrated into the GFAR to suppress background regions. • A dual context decoupled (DCD) head was constructed to achieve spatial disentanglement for different tasks. [ABSTRACT FROM AUTHOR]

Surface defect detection is crucial in the industrial domain, but deploying deep learning models on edge devices faces challenges of memory constraints and real-time requirements. Therefore, model pruning is necessary to reduce the model's parameter count and improve inference speed. Due to the high detection accuracy demanded in industrial defect detection, it is essential to prune the network while ensuring minimal impact on network accuracy. Different industrial defect scenarios may exhibit defects with varying importance of features and scale representation. Hence, to preserve the network's capability to recognize various defect features, pruning methods need to be adaptive to the dataset and effectively balance the remaining feature information across different scale layers in various scene contexts. To tackle these challenges, we propose a structured pruning method tailored for industrial detection scenarios, which significantly reduces model computation and parameter count while preserving network accuracy. Specifically, we design a scoring network to guide the allocation of pruning amounts for each layer. This network evaluates the importance of different scale layers globally based on the performance of various defect features within the dataset. Additionally, we employ the L1 norm criterion to prune filters with low weights that minimally affect accuracy. We also consider filter correlation and introduce a comprehensive evaluation method that incorporates both filter contribution and substitutability. This enables pruning of filters with low impact on detection accuracy. Finally, we obtain the most lightweight network through an iterative "prune-retraining" process. Our method is evaluated on four datasets using segmentation networks, achieving over 80% reduction in floating-point operations and 90% reduction in parameters while maintaining accuracy, thus significantly improving model memory usage and inference speed. • Pruning for lightweight surface defect detection network. • Reasonable allocation of pruning amounts for each layer based on layer importance. • Integrated assessment of filter redundancy based on filter collaborative contribution. • Pruning achieves minimal model size without compromising network accuracy. [ABSTRACT FROM AUTHOR]

This work proposes efficient multi-scale object detection model with space-to-depth convolution and BiFPN combined with FasterNet (ES-BiCF-YOLOv8), a deep learning method, to address the problems associated with detecting steel surface defects in contemporary industrial production. The method makes innovative improvements based on the YOLOv8 algorithm and enhances the performance of the novel model mainly through the following aspects. First, the space-to-depth layer followed by a non-strided convolution layer (SPD-Conv) and the efficient multi-scale attention mechanism is introduced into the feature extraction network to enhance the model's ability to capture fine-grained information and the fusion of multi-scale features. Second, the feature fusion network is optimized by utilizing a weighted bi-directional feature pyramid network and a lightweight network, FasterNet, to improve computational efficiency. Finally, it is shown that ES-BiCF-YOLOv8 reduces the complexity and computational requirements of the model while increasing the detection accuracy utilizing the NEU-DET dataset and deepPCB dataset with substantial experimental validation. The ES-BiCF-YOLOv8 model achieves a 5% improvement of the mean average precision value on the NEU-DET dataset, with the number of parameters and the computational amount only being the baseline 89% and 27%, and also demonstrates good generalization performance on the deepPCB dataset. Furthermore, the experiments demonstrate that ES-BiCF-YOLOv8 can be used for steel surface defect detection in industrial production because it uses less computational resources and can detect in real-time while maintaining high accuracy, in comparison to other popular object detection algorithms. The results of this work not only improve the efficiency and accuracy of steel surface defect detection but also provide ideas for the application of deep learning in the field of industrial detection. [ABSTRACT FROM AUTHOR]

Anomaly detection (AD) in an unsupervised manner has become the go-to approach in applications where data labeling proves problematic. However, these approaches are not completely unsupervised, since they rely on the weak knowledge of the dataset distribution into anomalous and anomaly-free subsets and typically require post-training threshold calibration in order to perform AD. Yet, they do not take advantage of available positive samples during training. In contrast, fully supervised approaches have proven to be more accurate and more efficient; however, they require a sufficient number of anomalous images to be labeled on a per-pixel level, which represents a labor-intensive task. In this article, we propose a hybrid approach that utilizes the best of both worlds. We use an unsupervised approach to build a model for generating pseudo labels, followed by a supervised approach to increase the robustness of AD. Moreover, we extend this approach with an active learning schema that results in learning with mixed supervision. We achieve several improvements, i.e., the utilization of available positive image samples, improved AD performance, and the retention of real-time performance. The proposed approach yields results that are comparable to the fully supervised approach, and at the very least, reduces the number of required labeled anomalous samples. [ABSTRACT FROM AUTHOR]

Visual inspection plays a pivotal role in numerous industrial production processes, and the pursuit of automation has surged with the rise of deep learning and convolutional neural networks (CNNs). Therein, the deployment of visual inspection CNNs on resource-constrained edge devices stands as a critical problem as these devices are the most affordable and well-suited for many industrial applications, e.g., production chains. Nonetheless, it faces challenges in meeting the computational demands of deep CNN models. Consequently, optimizing these models for efficient operation in such settings is imperative. Visual inspection tasks are often highly specialized, differing significantly from general computer vision tasks. As a result, state-of-the-art CNNs can be excessively large for achieving high accuracy on these specific datasets. To address this challenge, this paper introduces a novel approach utilizing neural architecture search (NAS) and hyperparameter optimization. We present the generic toolkit for NAS (GT-NAS), an open-source toolkit available for public use on GitLab (https://gitlab.com/pmf5/open-source/generic-toolkit-for-neural-architecture-search). We showcase the results of applying our methodology to two established state-of-the-art CNN models designed for surface defect detection, a problem that encompasses binary classification and segmentation of images.Our approach yields significantly smaller models relative to baselines, but with accuracy in line with the current state-of-the-art results, demonstrating the potential for enhanced efficiency in industrial visual inspection systems. In one experimental setting (optimizing the Mixed Supervision model on the KolektorSDD2 dataset), GT-NAS produced an architecture that is 6.2 times faster than the baseline while sacrificing only 0.25% of its average precision for binary classification. In another batch of experiments (optimizing the TriNet model on the SensumSODF dataset), GT-NAS also achieved remarkable results. It found a TriNet architecture five times smaller than the baseline, at a small cost of a 0.25% drop in the ROC-AUC classification score on the capsule subset of the SensumSODF dataset. Furthermore, on the softgel subset of the same dataset, GT-NAS produced a model that was 2.7 times smaller than the baseline, yet 0.19% more precise. [ABSTRACT FROM AUTHOR]

The inspection of visual quality represents a crucial step in the production process of bottled products. Numerous machine vision methodologies have demonstrated proficient identification of significant defects on bottle surfaces in well-controlled imaging environments. However, the actual production scenario introduces a myriad of surface defect types on bottled products, exhibiting diverse shapes, with the majority of defect instances characterized by relatively small individual areas. Faced with such diverse and small-sized defects, traditional convolutional neural networks (CNNs) encounter limitations due to the utilization of fixed-size convolutional kernels. This choice results in a constrained receptive field, hindering the capture of sufficiently effective contextual information. Additionally, pooling operations within traditional CNNs lead to a substantial reduction in feature map dimensions, causing excessive blurring or outright neglect of small-sized defects. Consequently, this detrimentally impacts the accuracy of surface defect detection and recognition. This study proposes a novel multi-scale defect detection model incorporating a variable receptive field and Gather–Distribute feature fusion mechanism to overcome limitations of traditional CNNs. Nine different scale defect images of bottled products, including worn, wrinkles, joint markings and etc, were collected and enhanced to construct a surface defect detection dataset for the actual production of bottled products. Enhancements to convolutional layers and the C2f module improve feature extraction for different defect shapes and sizes. Integration of the Gather–Distribute feature fusion module (GDFFM) reduces feature information loss and enhances shallow-layer feature utilization. An efficient detection head (E-Detect) with "parameter sharing" is proposed to reduce computational complexity and improve detection speed without experiencing significant accuracy loss. Experimental results demonstrate that the model's superiority over advanced defect detection algorithms across various categories. Notably, it achieves accuracies of 89.9%, 55.6%, 89.4%, and 75% for challenging defects like small worn, subtle wrinkles, inclined external cover, and misaligned labels, with improvements of 3.2%, 7%, 2%, and 11.8% over the baseline model, respectively. The model's mean Average Precision (mAP) also increases by 2.7%. • Promoting deformable convolution and Gather–Distribute features enhances backbone flexibility, improving small defect extraction in shallow layers. • Leveraging "parameter sharing", we designed E-Detect module, reducing model complexity, computational load, memory consumption, and enhancing detection speed. • This model excels in industrial environments, yielding superior application results, offering a new solution for surface defect detection in manufacturing. [ABSTRACT FROM AUTHOR]

Detecting surface defects plays a crucial role in ensuring the quality, functionality, and security of the production process. Traditional image processing techniques and machine learning models rely on manual analysis and feature extraction for specific vision inspection tasks. Deep learning approaches, which can automatically extract features from images, have demonstrated outstanding performance in computer vision tasks, including detecting surface defects. Motivated by this consideration, a Systematic Literature Review (SLR) method is employed for the comprehensive analysis of studies published between 2020 and 2023 in the field of deep learning-based surface defect detection applications in industrial products. The study provides a technical taxonomy for deep learning models according to the content of current studies through the SLR process, including Convolutional Neural Networks (CNN), encoder–decoder models, pyramid network models, Generative Adversarial Networks (GAN), attention-based models, and other models for surface defect detection. Then, the commonly used datasets for surface defect detection are discussed, and a comparative analysis of deep learning models' performance is provided. Our comparative analysis reveals that pyramid network models and CNN models are the most frequently used deep learning models for surface defect detection. These models yield reasonable results in surface defect detection due to their exceptional feature extraction capabilities. Finally, some hints for addressing future research directions and identifying open issues in surface defect detection applications are presented. [ABSTRACT FROM AUTHOR]

Automated surface detection has gradually emerged as a promising and crucial inspection method in the industrial sector, greatly enhancing production quality and efficiency. However, current semantic network models based on Vision Transformers are primarily trained on natural images, which exhibit complex object textures and backgrounds. Additionally, pure Vision Transformers lack the ability to capture local representations, making it challenging to directly apply existing semantic segmentation models to industrial production scenarios. In this paper, we propose a novel transformer segmentation model specifically designed for surface defect detection in industrial settings. Firstly, we employ a Dual-Attention Transformer (DAT) as the backbone of our model. This backbone replaces the generic 2D convolution block with a new self-attention block in the Spatial Reduction Attention module (SRA), enabling the establishment of a global view for each layer. Secondly, we enhance the collection of local information during decoding by initializing the relative position between q u e r y and k e y pixels. Finally, to strengthen the salient defect structure, we utilize Pixel Shuffle to rearrange the Ground Truth (GT) in order to guide the feature maps at each scale. Extensive experiments are conducted on three publicly industrial datasets, and evaluation results describe the outstanding performance of our network in surface defect detection. • Proposing a novel Transformer-based network for surface defect detections. • Developing a Dual-Attention module within the self-attention layer. • Establishing short-distance relationships between query and key tokens. • Incorporating a ground truth guidance to enhance the global structure of defects. • Our new network outperforms the previous state-of-the-art models. [ABSTRACT FROM AUTHOR]

In industrial production environments, the performance of surface defect detection methods for products is affected by disturbances in the equipment, manufacturing processes, and the environment, resulting in poor robustness of the training model. In this paper, an improved defect detection method based on feature extraction fusion and dynamic label assignment is proposed. Three improved network structures, i.e., the adaptive feature recognition convolution (AFRC), path aggregation balanced feature pyramid, and dynamic label assignment, are proposed for the feature extraction, feature fusion, and detection phases of the existing Faster RCNN model, respectively. The AFRC is employed to improve the feature extraction ability of the model for irregular defects, the path balance feature pyramid is utilized to enhance the recognition ability of the model for different levels of feature maps, and the dynamic label assignment is employed to improve the quality of the candidate frames. Under disturbed environments, the results of the proposed improvements outperform the benchmark faster RCNN model in terms of average precision (AP) by 4.1%, 3.7%, and 4.2% on an aluminum surface defect dataset and by 1.8%, 2.2%, and 3.7% AP on a magnetic tile surface defect dataset. The experimental results demonstrate that the proposed scheme effectively improves the robustness of the baseline model in environments with various disturbances. • A new adaptive feature recognition convolution is introduced. • Path aggregation balance feature pyramid is proposed to enhance recognition ability. • High-quality proposals are obtained based on a dynamic-based assigning module. • The new model achieves promising results on COCO Average Precision metrics and robustness. [ABSTRACT FROM AUTHOR]

Surface defect detection plays a crucial role in improving the overall quality of industrial production processes and ensuring that the resulting products meet the required quality standards. However, detecting these defects can be challenging due to various issues, including low image contrast, significant background noise, variable defect scales, and blurred boundaries. Although several segmentation networks have been developed to address these challenges, they still face difficulties in preserving fine-grained details during the encoding process and maintaining global features during the decoding process. Moreover, the simple skip connection that combines global and local information for feature fusion fails to consider their discrepancies and varying levels of significance. To overcome these limitations, this paper proposes a Pyramid Cross Attention Network (PCANet) for pixel-level surface defect detection. The encoder extracts multiresolution features, and the pyramid adaptive selection module (PASM) is introduced to supplement lost information and adaptively select information based on its importance. Furthermore, the cross attention fusion module (CAFM) is designed to address the incompatibility between the features extracted by the encoder and decoder in the raw skip connection, while strengthening defect areas and suppressing irrelevant noise using cross attention mechanisms. Finally, extensive experimental results demonstrate that the proposed PCANet outperforms other state-of-the-art methods in terms of mean Intersection over Union (mIoU). Specifically, it achieves an mIoU of 84.57 % in NEU-Seg and 79.37 % in MT_defect, respectively. [ABSTRACT FROM AUTHOR]

Aiming at the problems of low efficiency and poor accuracy in the product surface defect detection. In this paper, an online surface defects detection method based on YOLOV3 is proposed. Firstly, using lightweight network MobileNetV2 to replace the original backbone as the feature extractor to improve network speed. Then, we propose an extended feature pyramid network (EFPN) to extend the detection layer for multi-size object detection and design a novel feature fusing module (FFM) embedded in the extend layer to super-resolve features and capture more regional details. In addition, we add an IoU loss function to solve the mismatch between classification and bounding box regression. The proposed method is used to train and test on the hot rolled steel open dataset NEU-DET, which contains six typical defects of a steel surface, namely rolled-in scale, patches, crazing, pitted surface, inclusion and scratches. The experimental results show that our method achieves a satisfactory balance between performance and consumption and reaches 86.96% mAP with a speed of 80.96 FPS, which is more accurate and faster than many other algorithms and can realize real-time and high-precision inspection of product surface defects.

Surface defect detection for printed circuit board (PCB) is indispensable for managing PCB production quality. However, automatic detection of PCB surface defects is still a challenging task because, even within the same category of surface defect, defects present great differences in morphology and pattern. Although many computer vision-based detectors have been established to handle these problems, current detectors struggle to achieve high detection accuracy, fast detection speed and low memory consumption simultaneously. To address those issues, we propose a cost-effective deep learning (DL)-based detector based on the cutting-edge YOLOv4 to detect PCB surface defect quickly and efficiently. The YOLOv4 is improved upon with respect to its backbone network and the activation function in its neck/prediction network. The improved YOLOv4 is evaluated with a customized dataset, collected from a PCB factory. The experimental results show that the improved detector achieved a high performance, scoring 98.64% on mean average precision (mAP) at 56.98 frames per second (FPS), outperforming the other compared SOTA detectors. Furthermore, the improved YOLOv4 reduced the parameter space of YOLOv4 from 63.96 M to 39.59 M and the number of multiply-accumulate operations (Madds) from 59.75 G to 26.15 G.

Surface defect detection of an automobile wheel hub is important to the automobile industry because these defects directly affect the safety and appearance of automobiles. At present, surface defect detection networks based on convolutional neural network use many pooling layers when extracting features, reducing the spatial resolution of features and preventing the accurate detection of the boundary of defects. On the basis of DeepLab v3+, we propose a semantic segmentation network for the surface defect detection of an automobile wheel hub. To solve the gridding effect of atrous convolution, the high-resolution network (HRNet) is used as the backbone network to extract high-resolution features, and the multi-scale features extracted by the Atrous Spatial Pyramid Pooling (ASPP) of DeepLab v3+ are superimposed. On the basis of the optical flow, we decouple the body and edge features of the defects to accurately detect the boundary of defects. Furthermore, in the upsampling process, a decoder can accurately obtain detection results by fusing the body, edge, and multi-scale features. We use supervised training to optimize these features. Experimental results on four defect datasets (i.e., wheels, magnetic tiles, fabrics, and welds) show that the proposed network has better F1 score, average precision, and intersection over union than SegNet, Unet, and DeepLab v3+, proving that the proposed network is effective for different defect detection scenarios.

In this work, the creation of a dataset labeled in a pixel-wise manner for the uncommon domain of stain detection on patterned laundry is described. The unique properties of images in this dataset—stains are small and sometimes occur in large amounts—led to the creation of noisy labels. Indeed, the training of a fully convolutional neural network for salient object detection with this dataset revealed that the model predicts stains missed by human labelers. Thus, the reduction in label noise by adding overlooked regions with the help of the model’s predictions is examined in two different experiments. In the model-assisted labeling experiment, a simulation is ran where a human selects correct regions from the predictions. In the self-training experiment, regions of high certainty are automatically selected from the predictions. Re-training the model with the revised labels shows that model-assisted labeling leads to an average improvement in performance by 8.52%. In contrast, with self-training, the performance increase is generally lower (2.58% on average) and a decrease is even possible since regions of high certainty are often false positives.

As is well-known, defects precisely affect the lives and functions of the machines in which they occur, and even cause potentially catastrophic casualties. Therefore, quality assessment before mounting is an indispensable requirement for factories. Apart from the recognition accuracy, current networks suffer from excessive computing complexity, making it of great difficulty to deploy in the manufacturing process. To address these issues, this paper introduces binary networks into the area of surface defect detection for the first time, for the reason that binary networks prohibitively constrain weight and activation to +1 and −1. The proposed Bi-ShuffleNet and U-BiNet utilize binary convolution layers and activations in low bitwidth, in order to reach comparable performances while incurring much less computational cost. Extensive experiments are conducted on real-life NEU and Magnetic Tile datasets, revealing the least OPs required and little accuracy decline. When classifying the defects, Bi-ShuffleNet yields comparable results to counterpart networks, with at least 2× inference complexity reduction. Defect segmentation results indicate similar observations. Some network design rules in defect detection and binary networks are also summarized in this paper.