• 대한임베디드공학회논문지
• /
• 제17권3호
• /
• pp.129-138
• /
• 2022

Collecting and labeling sufficient training data, which is essential to deep learning-based visual inspection, is difficult for manufacturers to perform because it is very expensive. This paper presents a steel plate surface defect detection system with industrial-grade detection performance by training a small amount of steel plate surface images consisting of labeled and non-labeled data. To overcome the problem of lack of training data, we propose two data augmentation techniques: program-based augmentation, which generates defect images in a geometric way, and generative model-based augmentation, which learns the distribution of labeled data. We also propose a 4-step semi-supervised learning using pseudo labels and consistency training with fixed-size augmentation in order to utilize unlabeled data for training. The proposed technique obtained about 99% defect detection performance for four defect types by using 100 real images including labeled and unlabeled data.

• International Journal of Internet, Broadcasting and Communication
• /
• 제14권2호
• /
• pp.136-146
• /
• 2022

Some surface defects make the weak quality of steel materials. To limit these defects, we advocate a one-stage detector model RetinaNet among diverse detection algorithms in deep learning. There are several backbones in the RetinaNet model. We acknowledged two backbones, which are ResNet50 and VGG19. To validate our model, we compared and analyzed several traditional models, one-stage models like YOLO and SSD models and two-stage models like Faster-RCNN, EDDN, and Xception models, with simulations based on steel individual classes. We also performed the correlation of the time factor between one-stage and two-stage models. Comparative analysis shows that the proposed model achieves excellent results on the dataset of the Northeastern University surface defect detection dataset. We would like to work on different backbones to check the efficiency of the model for real world, increasing the datasets through augmentation and focus on improving our limitation.

• 한국진공학회:학술대회논문집
• /
• 한국진공학회 2015년도 제49회 하계 정기학술대회 초록집
• /
• pp.221.1-221.1
• /
• 2015

The vapor-liquid-solid (VLS) method, where the "liquid" catalytic droplets collecting atoms from vapor precursors build the solid crystal layers via supersaturation, is a ubiquitous technique to synthesize 1-dimensional nanoscale materials. However, the lack of fundamental understanding of chemical information governing the process inhibits the rational route to the structural programming. By combining the in situ or operando IR spectroscopy with post-growth high resolution electron microscopy, we show the strong correlation between the surface chemical species concentration and nanowire structures. More specifically, the critical role of surface adsorbed hydrogen, generated from the decomposition of Si2H6 precursor on the interplay between nanowire / kinking and the defect propagation is demonstrated. Our results show that adsorbed hydrogen atoms are responsible for selecting -oriented growth and indicate that a twin boundary imparts structural coherence. The twin boundary, only continuous at / kinks, reduces the symmetry of the trijunction and limits the number of degenerate directions available to the nanowire. These findings constitute a general approach for rationally engineering kinking superstructures and also provide important insight into the role of surface chemical bonding during VLS synthesis.

• 한국자동차공학회논문집
• /
• 제6권1호
• /
• pp.134-142
• /
• 1998

The effects of small hole defect size and effective case depth(ECD) on the four point bending fatigue limit of induction surface hardened S45C steel were investigated the fatigue limit evaluation of hardened materials is very difficult because of relations of the hardness gradient and residual stress. In this study, it was possible to characterize fatigue limit and fatigue life of induction surface hardened S45C steel in terms of the hole defect size and effective case depth(ECD) and quantitative evaluation of the fatigue limit with hole defects use Murakami's evaluation method and the range of evaluated values is a good accuracy compared with results.

• 전기학회논문지
• /
• 제66권5호
• /
• pp.870-876
• /
• 2017

In this paper, we proposed the automatic vision inspection system using multi-layer perceptron to detect the defects occurred on rail surface. The proposed system consists of image acquisition part and analysis part. Rail surface image is acquired as equal interval using line scan camera and lighting. Mean filter and dynamic threshold is used to reduce noise and segment defect area. Various features to characterize the defects are extracted. And they are used to train and distinguish defects by MLP-classifier. The system is installed on HEMU-430X and applied to analyze the rail surface images acquired from Honam-line at high speed up to 300 km/h. Recognition rate is calculated through comparison with manual inspection results.

• 동력기계공학회지
• /
• 제7권2호
• /
• pp.66-72
• /
• 2003

Recently, the steel parts used at the aerospace and automobile industries are required to be used light weight parts. Therefore, used material, steel have to be a high stress, which is an indispensable condition in this field. At the consideration of parts design, high hardness of the lightweight parts have an benefit of saving fuel and material. A high stress of metal has a point of difference according to the shape of design, external cyclic load and condition of vibration. A crack generates on the surface of metal or under yield stress by defect of inner metal defect or surface defect and slowly, this crack grow stable growth. Finally, rapidity failure phenomena is happen. Fatigue failure_phenomena, which happen in metal, bring on danger in human life and property therefor, anti-fatigue failure technology take an important part of current industries Currently, the shot peening is used for removing the defect from the surface of steel and improving the fatigue strength on surface. Therefore, this paper investigated the effect on frcature toughness using shot peening which is improve the resistance of crack growth and crack expansion rate by fatigue that make a compressive residual stress on surface.

• Archives of Reconstructive Microsurgery
• /
• 제7권2호
• /
• pp.157-164
• /
• 1998

It has been very difficult to managing partial joint defect in any etiologies, especially in children. Unicondylar defect of the tibial condyle in children reconstructed with proximal fibular head with articular cartilage from 1995. Two kinds of transfering methods were used, peroneal artery pedicled ipsilateral fibula head transposition to defective lateral tibial condyle defect that revealed poor prognosis with gradual absorption of transposed fibular epiphysis. Free vascularized fibular head transplantation with microvascular anastomosis underwent in the case with medial condyle defect of tibia which revealed very satisfactory results. Author can conclude with these clinical experiences: 1. Tranposition without epiphyseal vesssels intact is not sufficient in fibular head osteochondral transplantation in reconstruction of tibial condyle defect. That means peroneal arterial vascular pedicle is not enough for transplanted proximal epiphysis maintains its function on articular surface and growth activity in children. 2. The anterior recurrent tibial artery is one of the most important and easy to utilizing vessel in proximal fibular epiphyseal transplantation. 3. Free vascularized fibular head transplantation is hopeful method in reconstruction of the knee joint in the patient with partial joint defect which has no effective solution in conventional methods.

• 제어로봇시스템학회논문지
• /
• 제12권11호
• /
• pp.1156-1161
• /
• 2006

This paper describes a fast image labeling algorithm for the feature extraction of connected components. Labeling the connected regions of a digitized image is a fundamental computation in image analysis and machine vision, with a large number of application that can be found in various literature. This algorithm is designed for the surface defect inspection of Cold Mill Strip. The labeling algorithm permits to separate all of the connected components appearing on the Cold Mill Strip.

• Journal of Electrical Engineering and Technology
• /
• 제9권4호
• /
• pp.1360-1364
• /
• 2014

In this paper, RCS characteristics on defect pattern of crossed dipole slot FSS having a finite size have been analyzed. To analyze RCS, we applied the electric field integral equation analysis which applies BiCGSTab algorithm with iterative method and uses RWG basis function. To verify the validity of this paper, RCS of PEC sphere has been compared to the theoretical results and FSSs with defect patterns are fabricated and measured. As defect patterns in FSS, missing one column, missing some elements, and discontinuity in surfaces are simulated and compared with the measurement results. Resonant frequency shifts in pass band and changes in bandwidth are observed. From the results, precisely predicting and designing frequency characteristics over defect patterns are essential when applying FSS structures such as FSS radomes.

• 한국재료학회지
• /
• 제32권9호
• /
• pp.396-402
• /
• 2022

In this study, surface roughness and interfacial defect characteristics were analyzed after forming a high-k oxide film on the surface of a prime wafer and a test wafer, to study the possibility of improving the quality of the test wafer. As a result of checking the roughness, the deviation in the test after raising the oxide film was 0.1 nm, which was twice as large as that of the Prime. As a result of current-voltage analysis, Prime after PMA was 1.07 × 10 A/cm² and Test was 5.61 × 10 A/cm², which was about 5 times lower than Prime. As a result of analyzing the defects inside the oxide film using the capacitance-voltage characteristic, before PMA Prime showed a higher electrical defect of 0.85 × 10¹² cm^-2 in slow state density and 0.41 × 10¹³ cm^-2 in fixed oxide charge. However, after PMA, it was confirmed that Prime had a lower defect of 4.79 × 10¹¹ cm^-2 in slow state density and 1.33 × 10¹² cm^-2 in fixed oxide charge. The above results confirm the difference in surface roughness and defects between the Test and Prime wafer.