A surface defect detection method via fusing multi-level features

Surface defect detection is very important to ensure product quality, but most of the surface defects of industrial products are characterized by low contrast, big size difference and category similarity, which brings challenges to the automatic detection of defects. To solve these problems, we propose a defect detection method based on convolutional neural network. In this method, a backbone network with semantic supervision is applied to extract the features of different levels. While a multi-level feature fusion module is proposed to fuse adjacent feature maps into high-resolution feature maps successively, which significantly improves the prediction accuracy of the network. Finally, an Encoding module is used to obtain the global context information of the high-resolution feature map, which further improves the pixel classification accuracy. Experiments show that the proposed method is superior to other methods in NEU_SEG (mIoU of 85.27) and MT (mIoU of 77.82) datasets, and has the potential of real-time detection.

Investigation of Defect InP(001) Surface by Low Energy Ion Scаttering Spectroscopy

In this paper presents the computer simulation results on the investigations of the ion scattering processe on the defect InP(001)<110>,<ī10> surface under low-energy grazing ion bombardment have been presented. The peculiarities trajectories of the scattered ions from surface defect, atomic chain and semichannel have been investigated by computer simulation. It was found some trajectories nearby surface atomic chain which have loop shape and a line form. At grazing ion incidence, from a correlation of the experimental and calculated energy distributions of the scattered particles, one may determine a spatial extension of the missing atom on the monocrystal surface damaged by the ion bombardment.

Defect Detection of Mechanical Design Products Using Deep Neural Network

In industrial production, defect detection is one of the key methods to control the quality of mechanical design products. Although defect detection algorithms based on traditional machine learning can greatly improve detection efficiency, manual feature extraction is required and the design process is complicated. With the rapid development of CNN, major breakthroughs have been made in computer vision. Therefore, building a surface defect detection algorithm for mechanical design products based on DCNNs plays a very important role in improving industrial production efficiency. This paper studies the surface defect detection algorithm of mechanical products based on deep convolutional neural network, focusing on solving two types of problems: defect recognition and defect segmentation. Aiming at the problem of defect recognition, this paper studies a defect recognition algorithm based on fully convolutional block detection. This algorithm introduces the idea of block detection into the ResNet fully convolutional neural network. While realizing the local discrimination mechanism, it overcomes the shortcomings of the traditional block detection receptive field. Compared with the original ResNet image classification algorithm, this algorithm has stronger generalization ability and detection ability of small defects. Aiming at the problem of defect segmentation, this paper studies a defect segmentation algorithm based on improved Deeplabv3+.

LaTiO2N nanopowders (NPs) with low surface defect density via nitridation of flame made NPs retaining simple perovskite structure

LaTiO2N NP synthesized from flame made LaTiO3 NP exhibits less absorption background above the optical absorption edge than that synthesized from flame made La2Ti2O7 NP, suggesting a low surface defect density.

Surface defect engineered CeO2-x by ultrasound treatment for superior photocatalytic H2 production and water treatment

Surface defects on semiconductor photocatalyst display incredible light absorption bandwidth and function as highly active sites for oxidation processes by interacting with surface band structure. Accordingly, engineering the photocatalyst with...