• Journal of the Korea Society of Computer and Information
• /
• v.25 no.3
• /
• pp.19-26
• /
• 2020

Steel production requires high-quality surfaces with minimal defects. Therefore, the detection algorithms for the surface defects of steel strip should have good generalization performance. To meet the growing demand for high-quality products, the use of intelligent visual inspection systems is becoming essential in production lines. In this paper, we proposed a ShuffleDefectNet defect detection system based on deep learning. The proposed defect detection system exceeds state-of-the-art performance for defect detection on the Northeastern University (NEU) dataset obtaining a mean average accuracy of 99.75%. We train the best performing detection with different amounts of training data and observe the performance of detection. We notice that accuracy and speed improve significantly when use the overall architecture of ShuffleDefectNet.

• Proceedings of the KIEE Conference
• /
• 2000.07d
• /
• pp.2368-2370
• /
• 2000

In this paper, we proposed an inverse hysteresis model to cancel the nonlinear hysteresis phenomenon of a piezoelectric actuator and design a feedback control system based on the inverse hysteresis model. The piezoelectric actuator performs much better in open-loop response. However, the nonlinear hysteresis phenomenon should be linearized and the closed-loop control should be executed to get the required performance in the area, where high-speed and high-accuracy are required. Thus, it is shown by simulation that a good position tracking performance can be obtained for the repetitive desired position trajectory.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.45 no.2
• /
• pp.167-174
• /
• 1996

In this paper, a fault location algorithm is presented, which uses novel signal processing techniques and takes a new paradigm to overcome some drawbacks of the conventional methods. This new method for fault location on electric power transmission lines uses only one-terminal fault signals. The main feature of the method is hat it uses the high frequency components in fault signal and considers the influence of the source network by using a traveling wave propagation characteristics. As a result, we can develop a high speed, good accuracy fault locator.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.12
• /
• pp.45-53
• /
• 2020

The purpose of this paper is to reduce the ratio of the patient accidents that may occur in nursing hospitals. In other words, it determines whether the person approaching the dangerous area is a elderly (patient uniform) group or a practitioner(Casual Clothing) group, based on the clothing displayed by CCTV. We collected the basic learning data from web crawling techniques and nursing hospitals. Then model training data was created with Image Generator and Labeling program. Due to the limited performance of CCTV, it is difficult to create a good model with both high accuracy and speed. Therefore, we implemented the ResNet model with relatively excellent accuracy and the YOLO3 model with relatively excellent speed. Then we wanted to allow nursing hospitals to choose a model that they wanted. As a result of the study, we implemented a model that can distinguish patient and casual clothes with appropriate accuracy. Therefore, it is believed that it will contribute to the reduction of safety accidents in nursing hospitals by preventing the elderly from accessing the danger zone.

• Clinical Endoscopy
• /
• v.55 no.3
• /
• pp.390-400
• /
• 2022

Background/Aims: Previous artificial intelligence (AI) models attempting to segment gastric intestinal metaplasia (GIM) areas have failed to be deployed in real-time endoscopy due to their slow inference speeds. Here, we propose a new GIM segmentation AI model with inference speeds faster than 25 frames per second that maintains a high level of accuracy. Methods: Investigators from Chulalongkorn University obtained 802 histological-proven GIM images for AI model training. Four strategies were proposed to improve the model accuracy. First, transfer learning was employed to the public colon datasets. Second, an image preprocessing technique contrast-limited adaptive histogram equalization was employed to produce clearer GIM areas. Third, data augmentation was applied for a more robust model. Lastly, the bilateral segmentation network model was applied to segment GIM areas in real time. The results were analyzed using different validity values. Results: From the internal test, our AI model achieved an inference speed of 31.53 frames per second. GIM detection showed sensitivity, specificity, positive predictive, negative predictive, accuracy, and mean intersection over union in GIM segmentation values of 93%, 80%, 82%, 92%, 87%, and 57%, respectively. Conclusions: The bilateral segmentation network combined with transfer learning, contrast-limited adaptive histogram equalization, and data augmentation can provide high sensitivity and good accuracy for GIM detection and segmentation.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.22 no.6
• /
• pp.527-532
• /
• 2009

In this paper simulations for the impact into ceramics and/or metal materials have been discussed. To model discrete nature for fracture and damage of brittle materials, we implemented cohesive-law fracture model with a node separation algorithm for the tensile failure and Mohr-Coulomb model for the compressive loading. The drawback of this scheme is that it requires a heavy computational time. This is because new nodes are generated continuously whenever a new crack surface is created. In order to reduce the amount of calculation, parallelization with MPI library has been implemented. For the high-speed impact problems, the mesh configuration and contact calculation changes continuously as time step advances and it causes unbalance of computational load of each processor. Dynamic load balancing technique which re-allocates the loading dynamically is used to achieve good parallel performance. Some impact problems have been simulated and the parallel performance and accuracy of the solutions are discussed.

• Journal of Multimedia Information System
• /
• v.8 no.4
• /
• pp.211-220
• /
• 2021

The process of inspecting SMDs on the PCB boards improves the product quality, performance and reduces frequent issues in this field. However, undesirable scenarios such as assembly failure and device breakdown can occur sometime during the assembly process and result in costly losses and time-consuming. The detection of these components with a model based on deep learning may be effective to reduce some errors during the inspection in the manufacturing process. In this paper, YOLO models were used due to their high speed and good accuracy in classification and target detection. A SMD detection and classification method using YOLO networks based on robust data preprocessing and augmentation techniques to deal with various types of variation such as illumination and geometric changes is proposed. For 9 different components of data provided from a PCB manufacturer company, the experiment results show that YOLOv4 is better with fast detection and classification than YOLOv3.

• Wind and Structures
• /
• v.10 no.2
• /
• pp.99-119
• /
• 2007

Estimations of wind flow over terrain are often needed for applications such as pollutant dispersion, transport safety or wind farm location. Whilst field studies offer very detailed information regarding the wind potential over a small region, the cost of instrumenting a natural fetch alone is prohibitive. Wind tunnels offer one alternative although wind tunnel simulations can suffer from scale effects and high costs as well. Computational Fluid Dynamics (CFD) offers a second alternative which is increasingly seen as a viable one by wind engineers. There are two issues associated with CFD however, that of accuracy of the predictions and set-up and simulation times. This paper aims to address the two issues by demonstrating, by way of an investigation of wind potential for the Askervein Hill, that a good level of accuracy can be obtained with CFD (10% for the speed up ratio) and that it is possible to automate the simulations in order to compute a full wind rose efficiently. The paper shows how a combination of script and session files can be written to drive and automate CFD simulations based on commercial software. It proposes a general methodology for the automation of CFD applied to the computation of wind flow over a region of interest.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.40 no.8
• /
• pp.816-824
• /
• 1991

It is well known that dynamic control is needed for fast and accurate control. Neural networks are ideal for representing the strongly nonlinear relationship in the dynamic equations including complex unmodeled effects. It thus creates many advantages over conventional methods such as simple, fast and accurate control through neural network's inherent learning and massive parallelism. In this paper, dynamic control of the full six degrees of freedom of an industrial robot arm will be presented using neural networks. Moreover, through application to a real robot the usefulness of neurocontrol is demonstrated. The back propagation and feedback-error learning is used to train the neurocontroller. Simulated control of a PUMA 560 arm demonstrates that it moves at high speed with good accuracy and generalizes over untrained trajectories as well as adapt to unforseen load changes and sensor noise.

• Journal of ILASS-Korea
• /
• v.22 no.4
• /
• pp.169-174
• /
• 2017

The objective of this study was to verify the experimental and numerical results of spray evolution injected from different types of the nozzle-hole geometries. Spray visualization was taken by high speed camera under the different conditions. For the simulations of spray tip penetration, turbulence, evaporation and break-up model were applied K-zeta-f, Dukowicz and Wave model, respectively. Also, the prediction accuracy of spray tip penetration was increased by varying the spray cone angle. At the same time, the results of this work were compared in terms of spray tip penetration, and SMD characteristics. The numerical results of spray evolution process and spray tip penetration showed good agreement with experimental one.