Positive-Sample-Based Surface Defect Detection Using Memory-Augmented Adversarial Autoencoders

Neural network, a nonalgorithmic, nondigital, intensely parallel and distributive information processing system, is being used more and more everyday. The main interest in neural networks is rooted in the recognition that the human brain processes information in a different manner than conventional digital computers. Computers are extremely fast and precise at executing sequences of instructions that have been formulated for them. A human information processing system is composed of neurons switching at speeds about a million times slower than computer gates. Yet, humans are more efficient than computers at such computationally complex tasks as speech and other pattern-recognition problems. Artificial neural systems, or neural networks, are physical cellular systems that can acquire, store, and use experiential knowledge. The knowledge is in the form of stable states or mapping embedded in networks that can be recalled in response to the presentation of cues. In a typical neural data processing procedure, the database is divided into two separate portions called training and test sets. The training set is used to develop the desired network. In this process (depending on the paradigm that is being used), the desired output in the training set is used to help the network learn by adjusting the weights between its neurons or processing elements. Once the network has learned the information in the training set and has"converged," the test set is applied to the network for verification. It is important to note that although the user has the desired output of the test set, it has not been seen by the network. This ensures the integrity and robustness of the trained network. A handful of articles on the use of neural networks in the petroleum industry has appeared in SPE conferences, proceedings, and publications in the past 2 years. These articles can be divided into two categories: those that use neural networks to analyze formation lithology from well logs and those that use neural networks to pick a reservoir model to be used in conventional welltest interpretation studies. These tasks are usually done by log analysts and reservoir engineers, and their automation using a fault-tolerant process may prove valuable. Neural networks can help engineers and researchers by addressing some fundamental petroleum engineering problems as well as specific ones that conventional computing has been unable to solve. Petroleum engineers may benefit from neural networks on occasions when engineering data for design and interpretations are less than adequate. This is an especially common occurrence in the Appalachian basin, where some fields are quite old. Lack of adequate engineering data may also be encountered because of the high cost of coring, well testing, and so on. Neural networks have shown great potential for generating accurate analysis and results from large amounts of historical data that otherwise would seem not to be useful or relevant in the analysis. An example of such a problem was encountered by a gas company for a gas storage field in Ohio. In the absence of appropriate data, which normally would make engineering design and evaluation of the fracturing jobs virtually impossible, a carefully designed neural network was able to predict the performance of fracturing jobs with great accuracy. A linear plot of the actual fracturing job results (data never seen by the network during training) and network predictions resulted in a correlation coefficient of 0.98, where 1.00 is a perfect match. Neural networks have proved to be valuable pattern-recognition tools. They are capable of finding highly complex patterns within large amounts of data. A relevant example is well log interpretation. It is generally accepted that there is more information embedded in well logs than meets the eye. Determination, prediction, or estimation of formation permeability without actual laboratory measurement of the cores or interruption in production for well test data collection has been a fundamental problem for petroleum engineers. From geophysical well log data, it was possible to predict and/or estimate permeability of a highly heterogeneous formation in West Virginia, as shown in Fig 1. Reference Mohaghegh, S., Arefi, R., and Ameri, S.: "Design and Development of an Artificial Neural Network for Prediction of Formation Permeability,"paper SPE 28237 presented at the 1994 SPE Petroleum Computer Conference, July31-Aug. 3, Dallas.

In industrial quality control, anomaly detection plays a critical role in identifying defective products. However, because of the rarity and time-consuming nature of defect collection, training models often rely solely on defect-free samples. This necessitates the use of unsupervised anomaly-detection techniques trained exclusively on defect-free data. Alternatively, defect data can be synthesized to augment the dataset with defective samples. In the textile industry, expeditious model training is crucial to ensure a smooth production flow. Unfortunately, most unsupervised methods require extensive training time. This paper proposes a novel transfer learning approach designed to achieve training times in seconds while effectively adapting the model to the target domain of fabric anomaly detection. The key contributions of our method include significantly reduced training times, up to 10 times faster than current state-of-the-art methods, and comparable performance in anomaly detection, achieving results on par with state-of-the-art approaches on benchmark datasets (MVTEC Anomaly Detection, TILDA, AITEX and DAGM). Additionally, our approach improves inference times, ensuring expedited and efficient anomaly detection during production. The proposed method offers a practical and efficient solution for real-time industrial quality control.

In modern industrial production, quality inspection and control play a crucial role in enhancing production efficiency and ensuring product reliability. Traditional inspection methods primarily rely on manual labor, which not only results in low efficiency but also introduces human error, leading to unstable inspection outcomes. With the advancement of science and technology, automated image processing technology has emerged, revolutionizing industrial testing. By leveraging computer vision and image analysis algorithms, automated image processing technology can rapidly and accurately identify and analyze defects and anomalies in images, significantly improving inspection efficiency and accuracy. This technology not only enables real-time monitoring of product quality on production lines but also provides data support for process optimization, driving industrial production toward intelligent and automated development.

This study proposes a novel sparse encoding detection method that utilizes pre-trained deep neural networks to extract fused multi-scale features, addressing critical challenges in automated surface defect detection in manufacturing. By building sparse dictionaries with minimal redundancy, our technique overcomes issues related to poor feature representation and uneven features, requiring only defect-free samples for the model training. The pre-trained neural network extracts high-dimensional features uniformly, aligning and fusing the intermediate features across different network scales to enhance the image semantic information. We construct a representative sparse subset from the feature pool derived from defect-free samples using minimal redundancy as a constraint. The final step involves querying the sparse dictionary to calculate the image-level anomaly score and region. Extensive experiments on the MVTec Anomaly Detection (MVTec AD) dataset and the corrugated pipe dataset validate the efficacy of our approach. Specifically, we achieve a pixel-level area under the receiver operating characteristic (AUROC) score of 98.3% on the corrugated pipe dataset, demonstrating our method's superior performance compared to existing state-of-the-art defect detection and segmentation techniques. Our research contributes significantly to advancements in Artificial Intelligence for manufacturing applications by enhancing the reliability and accuracy of surface defect detection.

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.

Industry 4.0 is revolutionizing the world of manufacturing by merging modern digital technologies with conventional production techniques. One of the most crucial advancements is the use of robotics and automated systems for quality checking, which improves accuracy, consistency, and real-time sensitivity. In this paper, deep robotics-based quality control systems are presented with an in-depth analysis focusing on the incorporation of artificial intelligence (AI), machine vision, and sensor networks. The transformation towards Industry 4.0 has heightened the demand for smart and automated quality inspection systems that can work with speed, accuracy, and flexibility. Traditional inspection methods tend to fail in meeting the increasing complexity and amount of contemporary manufacturing processes. Robotics and automation, enabled by machine learning, computer vision, and sensor fusion, offer a revolutionary solution to such issues. This paper contribution acknowledges the manner in which these technologies are transforming industrial quality assurance in the context of Industry 4.0.

In practical industrial inspection scenarios, the challenges of limited samples and severe domain shift render achieving reliable few-shot image classification a pressing challenge. In response to this, we propose a lightweight small-sample industrial quality inspection image classification and confidence calibration framework based on Fast AutoAugment. The framework employs EfficientNetV2 as the basic feature extractor, using an advanced policy search mechanism enabled by Fast AutoAugment, which successfully enhances the model’s generalization ability to defect features. In relation to feature fusion, we design a multi-branch feature fusion module that combines spatial attention, channel recalibration and contrast alignment mechanisms to further promote discriminant expression in small sample conditions. Additionally, we design an uncertainty calibration module based on Dirichlet distribution, which integrates task-adaptive priors and temperature scaling strategies to effectively calibrate the model’s confidence and reduce the risk of "overconfidence errors" in industrial settings. Experiments on the NEU-CLS industrial quality inspection dataset show that the proposed method outperforms the existing Few-shot baseline, demonstrating a classification accuracy that is 4.6%-7.1% higher than that of the existing few-shot baselines.

In this paper, we systematically review recent advances in surface inspection using computer vision andimage processing techniques, particularly those based on texture analysis methods. The aim is to reviewthe state-of-the-art techniques for the purposes of visual inspection and decision making schemes that areable to discriminate the features extracted from normal and defective regions. This field is so vast that itis impossible to cover all the aspects of visual inspection. This paper focuses on a particular but importantsubset which generally treats visual surface inspection as texture analysis problems. Other topics related tovisual inspection such as imaging system and data acquisition are out of the scope of this survey.The surface defects are loosely separated into two types. One is local textural irregularities which is themain concern for most visual surface inspection applications. The other is global deviation of colour and/ortexture, where local pattern or texture does not exhibit abnormalities. We refer this type of defects as shadeor tonality problem. The second type of defects have been largely neglected until recently, particularly whencolour imaging system has been widely used in visual inspection and where chromatic consistency plays animportant role in quality control. The emphasis of this survey though is still on detecting local abnormalities,given the fact that majority of the reported works are dealing with the first type of defects.The techniques used to inspect textural abnormalities are discussed in four categories, statistical approaches,structural approaches, filter based methods, and model based approaches, with a comprehensivelist of references to some recent works. Due to rising demand and practice of colour texture analysis inapplication to visual inspection, those works that are dealing with colour texture analysis are discussedseparately. It is also worth noting that processing vector-valued data has its unique challenges, which conventionalsurface inspection methods have often ignored or do not encounter.We also compare classification approaches with novelty detection approaches at the decision makingstage. Classification approaches often require supervised training and usually provide better performancethan novelty detection based approaches where training is only carried out on defect-free samples. However,novelty detection is relatively easier to adapt and is particularly desirable when training samples areincomplete

The quality of cigarette outer packaging is a critical determinant influencing consumers' purchasing decisions; consequently, efficient and accurate detection of packaging defects on the production line is of paramount importance. Deep learning methods have demonstrated considerable promise in this domain; however, their reliance on large quantities of manually annotated data leads to substantial labor costs and limited processing efficiency. To address this limitation, this paper proposes an automatic annotation framework based on an enhanced YOLOv8n (You Only Look Once v8 Nano) architecture. The core innovation centers on the integration of the ASC module, which significantly strengthens the model s capacity to capture discriminative features of complex defects through the synergistic combination of multi-scale dilated convolutions and an attention mechanism. Experimental evaluations conducted on a self-constructed dataset of cigarette small-pack defects reveal that the improved model achieves a mean average precision (mAP) of 95.9%, representing a 1.5 percentage point improvement over the baseline YOLOv8n. Moreover, the proposed method enables direct generation of high-quality bounding boxes and corresponding label files for unannotated images, thereby facilitating end-to-end automatic annotation. This study demonstrates that the proposed approach not only enhances recognition accuracy but also substantially improves annotation efficiency, offering a robust and scalable solution for industrial surface quality inspection systems.

Sparse random neural networks for online anomaly detection on sensor nodes

To address the challenges of high noise levels in multi-source data, easy masking of key features, and difficulty in tracing anomalies in industrial quality inspection, a multi-level feature attention fusion quality anomaly detection and tracing algorithm (MFA-QADT) is proposed. This algorithm utilizes a three-module architecture: preprocessing, feature extraction, and detection and tracing. In the preprocessing stage, density peak clustering and dynamic normalization are used to filter noise and unify dimensions. In the feature extraction stage, atrous convolution and Transformer self-attention are combined with a quality weighting mechanism to enhance the representation of key features. In the detection and tracing stage, an improved VAE model is used to identify and locate anomalies. Experimental data is collected from 58,000 quality inspection data samples (50,000 normal and 8,000 abnormal samples) from an automotive chassis parts manufacturer. Four algorithms, including control charts, support vector machines (SVMs), CNN-VAEs, and LSTM-Attention, are compared. Results show that the MFA-QADT anomaly detection accuracy reached 98.2% (a 4.3% improvement over CNN-VAE), a recall rate of 97.5% (a 5.1% improvement over LSTM-Attention), and a traceability accuracy of 95.3%. The detection latency was 0.32 seconds, and even under 30% noise, it maintained an accuracy rate of 92.1%. Its robustness and real-time performance meet the requirements of industrial quality inspection.

Wind energy is a renewable resource with abundant reserves, and its sustainable development and utilization are crucial. The components of wind turbines, particularly the blades and various surfaces, require meticulous defect detection and maintenance due to their significance. The operational status of wind turbine generators directly impacts the efficiency and safe operation of wind farms. Traditional surface defect detection methods for wind turbines often involve manual operations, which suffer from issues such as high subjectivity, elevated risks, low accuracy, and inefficiency. The emergence of computer vision technologies based on deep learning has provided a novel approach to surface defect detection in wind turbines. However, existing datasets designed for wind turbine surface defects exhibit overall category scarcity and an imbalance in samples between categories. The algorithms designed face challenges, with low detection rates for small samples. Hence, this study first constructs a benchmark dataset for wind turbine surface defects comprising seven categories that encompass all common surface defects. Simultaneously, a wind turbine surface defect detection algorithm based on improved YOLOv5 is designed. Initially, a multi-scale copy-paste data augmentation method is proposed, introducing scale factors to randomly resize the bounding boxes before copy-pasting. This alleviates sample imbalances and significantly enhances the algorithm’s detection capabilities for targets of different sizes. Subsequently, a dynamic label assignment strategy based on the Hungarian algorithm is introduced that calculates the matching costs by weighing different losses, enhancing the network’s ability to learn positive and negative samples. To address overfitting and misrecognition resulting from strong data augmentation, a two-stage progressive training method is proposed, aiding the model’s natural convergence and improving generalization performance. Furthermore, a multi-scenario negative-sample-guided learning method is introduced that involves incorporating unlabeled background images from various scenarios into training, guiding the model to learn negative samples and reducing misrecognition. Finally, slicing-aided hyper inference is introduced, facilitating large-scale inference for wind turbine surface defects in actual industrial scenarios. The improved algorithm demonstrates a 3.1% increase in the mean average precision (mAP) on the custom dataset, achieving 95.7% accuracy in mAP_50 (the IoU threshold is half of the mAP). Notably, the mAPs for small, medium, and large targets increase by 18.6%, 16.4%, and 6.8%, respectively. The experimental results indicate that the enhanced algorithm exhibits high detection accuracy, providing a new and more efficient solution for the field of wind turbine surface defect detection.

In machine learning-based fractional cover estimation, the fractional cover distribution in training samples critically influences model construction and, consequently the accuracy of the estimations. While some studies have descriptively compared the accuracies of machine learning-based estimations across training sets derived from different sampling methods, a significant gap remains in quantitatively analyzing how the fractional cover distribution in training samples affects accuracy. This study aims to bridge this gap by introducing descriptors for fractional cover distribution in the training set and establishing mathematical relationships between these descriptors and the accuracy of fractional cover estimation. We employed the Dirichlet distribution to characterize the joint fractional cover of multiple land classes and the Beta distribution for single-class cover. Subsequently, two descriptors were developed: the Kullback-Leibler (KL) divergence, measuring the similarity of fractional cover distributions for the target class between the training and test sets, and the geometric angle, representing the fractional cover distributions of the target class in the training set at the same KL divergence. Fractional cover estimation was performed using random forest regression, with accuracy assessed on an independent test set. The relationships between the KL divergence and accuracy, and between the geometric angle and accuracy at the same KL divergence, were modeled using univariate linear models and harmonic models, respectively. The combined effects of these descriptors on accuracy were further analyzed using coupled harmonic analysis and generalized additive models. Our experimental results, using both simulated and real data, demonstrated the effectiveness of these models. Given the strong explanatory power of the KL divergence in the accuracy of fractional cover estimation, we encourage researchers to report detailed statistical information of both training and test sets, enriching the understanding of model performance in fractional cover estimation.

Due to large amount of image data, the accuracy and real-time performance of classification are difficult problems in image classification. On the one hand, the broad learning system (BLS) has achieved good results in the timeliness of classification. On the other hand, deep learning feature extraction effect is good, but its structure is complex and the training time is long. Therefore, based on the above advantages of both methods, this paper proposed BLS based on deep features for image classification. Firstly, a feature extractor is constructed based on the ResNet101 to obtain the deep features of the classification image. Then, the feature nodes and enhancement nodes of BLS are constructed based on the deep features. Through the experiment, our method has two performance on benchmark datasets: high classification accuracy, good real-time.

Given activity training data from high-throughput screening (HTS) experiments, virtual high-throughput screening (vHTS) methods aim to predict in silico the activity of untested chemicals. We present a novel method, the Influence Relevance Voter (IRV), specifically tailored for the vHTS task. The IRV is a low-parameter neural network which refines a k-nearest neighbor classifier by nonlinearly combining the influences of a chemical's neighbors in the training set. Influences are decomposed, also nonlinearly, into a relevance component and a vote component. The IRV is benchmarked using the data and rules of two large, open, competitions, and its performance compared to the performance of other participating methods, as well as of an in-house support vector machine (SVM) method. On these benchmark data sets, IRV achieves state-of-the-art results, comparable to the SVM in one case, and significantly better than the SVM in the other, retrieving three times as many actives in the top 1% of its prediction-sorted list. The IRV presents several other important advantages over SVMs and other methods: (1) the output predictions have a probabilistic semantic; (2) the underlying inferences are interpretable; (3) the training time is very short, on the order of minutes even for very large data sets; (4) the risk of overfitting is minimal, due to the small number of free parameters; and (5) additional information can easily be incorporated into the IRV architecture. Combined with its performance, these qualities make the IRV particularly well suited for vHTS.