Steel Surface Defect Detection Using an Ensemble of Deep Residual Neural Networks

Steel defect diagnostics is important for industry task as it is tied to the product quality and production efficiency. The aim of this paper is evaluating the application of residual neural networks for recognition of industrial steel defects of three classes. Developed and investigated models based on deep residual neural networks for the recognition and classification of surface defects of rolled steel. Investigated the influence of various loss functions, optimizers and hyperparameters on the obtained result and selected optimal model parameters. Based on an ensemble of two deep residual neural networks ResNet50 and ResNet152, a classifier was constructed to detect defects of three classes on flat metal surfaces. The proposed technique allows classifying images with high accuracy. The average binary accuracy of classifying the test data is 96.7% for all images (including defect-free ones). The fields of neuron activation in the convolutional layers of the model were investigated. Feature maps formed in the process were found to reflect the position, size and shape of the objects of interest very well. The proposed ensemble model has proven to be robust and able to accurately recognize steel surface defects. Erroneous recognition cases of the classifier application are investigated. It was shown that errors most often occur in ambiguous situations, where surface artifacts of different types are similar.

Design of an Optical Transparent Absorber and Defect Diagnostics Analysis Based on Near-Field Measurement

To reduce the electromagnetic wave interference caused by cavity resonance or electromagnetic wave leakage, herein, an optical transparent radar absorbing structure (RAS) was designed using transparent conductive oxides (TCOs) with a high optical transmittance and electrical conductivity, and a procedure was proposed for detecting possible defects in the fabrication and operation and for assessing the influence of the defects on the electromagnetic performance. To detect locally occurring defects in planar and three-dimensional absorbing structures, a measurement system based on an open-ended near-field antenna capable of producing small beam spots at a close distance was constructed. Moreover, the reflection characteristics of the transparent RAS were derived from a simplified multiple reflection equation, and the derived results were compared with the analysis results of an equivalent circuit model to predict the surface resistance of the TCO coating layer, based on which the presence of defects could be confirmed. By using the experimental results, the predicted surface resistance results of the coating layer and the results measured using a non-contact sheet resistance meter were compared and were found to correspond, thereby confirming the effectiveness of the proposed defect detection method.

Congenital Disorders of genital organs with menstrual Blood outflow Defect: Diagnostics, Treatment and Prevention of Complications

There is significant growth in numbers of patients with congenital disorders of various organs and systems in recent decades. The complexity of this problem is defined by difficulties in diagnostics especially in cases of associative disorders which leads to untimely treatment and numerous complications. This article covers features of management of female patients with malformations of genital organs with menstrual blood outflow defect.

Steel Surface Defect Diagnostics Using Deep Convolutional Neural Network and Class Activation Map

Steel defect diagnostics is considerably important for a steel-manufacturing industry as it is strongly related to the product quality and production efficiency. Product quality control suffers from a real-time diagnostic capability since it is less-automatic and is not reliable in detecting steel surface defects. In this study, we propose a relatively new approach for diagnosing steel defects using a deep structured neural network, e.g., convolutional neural network (CNN) with class activation maps. Rather than using a simple deep learning algorithm for the classification task, we extend the CNN diagnostic model for being used to analyze the localized defect regions within the images to support a real-time visual decision-making process. Based on the experimental results, the proposed approach achieves a near-perfect detection performance at 99.44% and 0.99 concerning the accuracy and F-1 score metric, respectively. The results are better than other shallow machine learning algorithms, i.e., support vector machine and logistic regression under the same validation technique.

Магнитометрическая диагностика дефектов высокотемпературных сверхпроводящих лент в градиентном магнитном поле

A new technique for contactless magnetometric determination of the local critical current in high-temperature superconducting tapes is proposed. In contrast to conventional approaches, where currents in a superconductor are induced by a uniform magnetic field of an external source, in our variant the tape is magnetized by a series of bipolar permanent magnets. It is shown that, for solving a number of technical problems of defect diagnostics, the proposed approach is more effective than those used earlier. Two variants of diagnostics are discussed. The first variant is intended for express diagnostics of local defects (first of all, transverse cracks) in a separate tape and in a tape stack and the second variant is developed for the case of a smooth critical current variation in a separate tape. The proposed method can significantly improve the accuracy of determining the local critical current density.