• Journal of Internet Computing and Services
• /
• v.23 no.3
• /
• pp.35-45
• /
• 2022

Even in recent years, treatment of first-aid patients is still often delayed due to a shortage of medical resources in marginalized areas. Research on automating the analysis of medical data to solve the problems of inaccessibility for medical services and shortage of medical personnel is ongoing. Computer vision-based medical inspection automation requires a lot of cost in data collection and labeling for training purposes. These problems stand out in the works of classifying lesion that are rare, or pathological features and pathogenesis that are difficult to clearly define visually. Anomaly detection is attracting as a method that can significantly reduce the cost of data collection by adopting an unsupervised learning strategy. In this paper, we propose methods for detecting abnormal images on chest X-RAY images as follows based on existing anomaly detection techniques. (1) Normalize the brightness range of medical images resampled as optimal resolution. (2) Some feature vectors with high representative power are selected in set of patch features extracted as intermediate-level from lesion-free images. (3) Measure the difference from the feature vectors of lesion-free data selected based on the nearest neighbor search algorithm. The proposed system can simultaneously perform anomaly classification and localization for each image. In this paper, the anomaly detection performance of the proposed system for chest X-RAY images of PA projection is measured and presented by detailed conditions. We demonstrate effect of anomaly detection for medical images by showing 0.705 classification AUROC for random subset extracted from the PadChest dataset. The proposed system can be usefully used to improve the clinical diagnosis workflow of medical institutions, and can effectively support early diagnosis in medically poor area.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.24 no.2
• /
• pp.217-230
• /
• 2022

The adequate control of TBM face pressure is of vital importance to maintain face stability by preventing face collapse and surface settlement. An EPB shield TBM excavates the ground by applying face pressure with the excavated soil in the pressure chamber. One of the challenges during the EPB shield TBM operation is the control of face pressure due to difficulty in managing the excavated soil. In this study, the face pressure of an EPB shield TBM was predicted using the geological and operational data acquired from a domestic TBM tunnel site. Four machine learning algorithms: KNN (K-Nearest Neighbors), SVM (Support Vector Machine), RF (Random Forest), and XGB (eXtreme Gradient Boosting) were applied to predict the face pressure. The model comparison results showed that the RF model yielded the lowest RMSE (Root Mean Square Error) value of 7.35 kPa. Therefore, the RF model was selected as the optimal machine learning algorithm. In addition, the feature importance of the RF model was analyzed to evaluate appropriately the influence of each feature on the face pressure. The water pressure indicated the highest influence, and the importance of the geological conditions was higher in general than that of the operation features in the considered site.

• Composites Research
• /
• v.35 no.5
• /
• pp.317-321
• /
• 2022

In this study, the PIC design method with machine learning that automatically assigning different stacking sequences according to loading types was applied bumper beam. The input value and labels of the training data for applying machine learning were defined as coordinates and loading types of reference elements that are part of the total elements, respectively. In order to compare the 2D and 3D implementation method, which are methods of representing coordinate value, training data were generated, and machine learning models were trained with each method. The 2D implementation method is divided FE model into each face and generating learning data and training machine learning models accordingly. The 3D implementation method is training one machine learning model by generating training data from the entire finite element model. The hyperparameter were tuned to optimal values through the Bayesian algorithm, and the k-NN classification method showed the highest prediction rate and AUC-ROC among the tuned models. The 3D implementation method revealed higher performance than the 2D implementation method. The loading type data predicted through the machine learning model were mapped to the finite element model and comparatively verified through FE analysis. It was found that 3D implementation PIC bumper beam was superior to 2D implementation and uni-stacking sequence composite bumper.

• Korean Journal of Remote Sensing
• /
• v.38 no.5_2
• /
• pp.793-810
• /
• 2022

Synthetic Aperture Radar (SAR) is considered to be suitable for near real-time inundation monitoring. The distinctly different intensity between water and land makes it adequate for waterbody detection, but the intrinsic speckle noise and variable intensity of SAR images decrease the accuracy of waterbody detection. In this study, we suggest two modules, named 'morphology module' and 'edge-enhanced module', which are the combinations of pooling layers and convolutional layers, improving the accuracy of waterbody detection. The morphology module is composed of min-pooling layers and max-pooling layers, which shows the effect of morphological transformation. The edge-enhanced module is composed of convolution layers, which has the fixed weights of the traditional edge detection algorithm. After comparing the accuracy of various versions of each module for U-Net, we found that the optimal combination is the case that the morphology module of min-pooling and successive layers of min-pooling and max-pooling, and the edge-enhanced module of Scharr filter were the inputs of conv9. This morphologic and edge-enhanced U-Net improved the F1-score by 9.81% than the original U-Net. Qualitative inspection showed that our model has capability of detecting small-sized waterbody and detailed edge of water, which are the distinct advancement of the model presented in this research, compared to the original U-Net.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.15 no.6
• /
• pp.463-470
• /
• 2022

Intravenous injection is widely used for patient treatment, including injection drugs, fluids, parenteral nutrition, and blood products, and is the most frequently performed invasive treatment for inpatients, including blood collection, peripheral catheter insertion, and other IV therapy, and more than 1 billion cases per year. Intravenous injection is one of the difficult procedures performed only by experienced nurses who have been trained in intravenous injection, and failure can lead to thrombosis and hematoma or nerve damage to the vein. Nurses who frequently perform intravenous injections may also make mistakes because it is not easy to detect veins due to factors such as obesity, skin color, and age. Accordingly, studies on auxiliary equipment capable of visualizing the venous structure of the back of the hand or arm have been published to reduce mistakes during intravenous injection. This paper is about the development of venous detection equipment that visualizes venous structure during intravenous injection, and the optimal combination was selected by comparing the brightness of acquired images according to the combination of near-infrared (NIR) LED and Filter with different wavelength bands. In addition, an image processing algorithm was derived to threshehold and making blood vessel part to green through grayscale conversion, histogram equilzation, and sharpening filters for clarity of vein images obtained through the implemented venous detection experimental module.

• Composites Research
• /
• v.36 no.1
• /
• pp.53-58
• /
• 2023

The PIC design method is assigning different stacking sequences for each shell element through the preliminary FE analysis. In previous study, machine learning was applied to the PIC design method in order to assign the region efficiently, and the training data is labeled by dividing each region into tension, compression, and shear through the preliminary FE analysis results value. However, since buckling is not considered, when buckling occurs, it can't be divided into appropriate loading type. In the present study, it was proposed PIC-NTL (PIC design using novel technique for analyzing load type) which is method for applying a novel technique for analyzing load type considering buckling to the conventional PIC design. The stress triaxiality for each ply were analyzed for buckling analysis, and the representative loading type was designated through the determined loading type within decision area divided into two regions of the same size in the thickness direction of the elements. The input value of the training data and label consisted in coordination of element and representative loading type of each decision area, respectively. A machine learning model was trained through the training data, and the hyperparameters that affect the performance of the machine learning model were tuned to optimal values through Bayesian algorithm. Among the tuned machine learning models, the SVM model showed the highest performance. Most effective stacking sequence were mapped into PIC tube based on trained SVM model. FE analysis results show the design method proposed in this study has superior external loading resistance and energy absorption compared to previous study.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.26 no.3
• /
• pp.85-96
• /
• 2023

This study developed and evaluated a safe route guidance service for electric personal assistive mobility device used mainly by the mobility impaired people to improve their mobility. Thirteen underlying factors affecting the mobility of electric personal assistive mobility device have been derived through a survey with the mobility impaired people and employees in related organizations in Busan Metropolitan City. After assigning safety scores to individual factors and identifying the relevant factors along routes of interest with an object detection AI model, the safe route for electric personal assistive mobility device was provided through an optimal path-finding algorithm. As a result of comparing the general route of T-map and the recommended route of this study for the identical routes, the latter had relatively fewer obstacles and the gentler slope than the former, implicating that the recommended route is safer than the general one. As future works, it is necessary to enhance the function of a route guidance service based on the real-time location of users and to conduct spot investigations to evaluate and verify its social acceptability.

• Journal of the Korea Society for Simulation
• /
• v.19 no.1
• /
• pp.143-151
• /
• 2010

Fingerprint classification is a step to increase the efficiency of an 1:N fingerprint recognition system and plays a role to reduce the matching time of fingerprint and to increase accuracy of recognition. It is difficult to classify fingerprints, because the ridge pattern of each fingerprint class has an overlapping characteristic with more than one class, fingerprint images may include a lot of noise and an input condition is an exceptional case. In this paper, we propose a novel approach to design a stochastic model and to accomplish fingerprint classification using a directional characteristic of fingerprints for an effective classification of various qualities. We compute the directional value by searching a fingerprint ridge pixel by pixel and extract a directional characteristic by merging a computed directional value by fixed pixels unit. The modified Markov model of each fingerprint class is generated using Markov model which is a stochastic information extraction and a recognition method by extracted directional characteristic. The weight list of classification model of each class is decided by analyzing the state transition matrixes of the generated Markov model of each class and the optimized value which improves the performance of fingerprint classification using GA (Genetic Algorithm) is estimated. The performance of the optimized classification model by GA is superior to the model before the optimization by the experiment result of applying the fingerprint database of various qualities to the optimized model by GA. And the proposed method effectively achieved fingerprint classification to exceptional input conditions because this approach is independent of the existence and nonexistence of singular points by the result of analyzing the fingerprint database which is used to the experiments.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.317-321
• /
• 2022

In response to Incheon tap water accident in 2019, the Ministry of Environment has created the "Comprehensive Measures for Water Safety Management" to improve water operation management, provide systematic technical support, and respond to accidents. Accordingly, K-water is making a smart water supply management system for the entire process of tap water. In order to advance the response to water accidents, it is essential to secure the reliability of real-time water operation data such as flow rate, pressure, and water level, and to develop and apply a warning algorithm in advance using big data analysis techniques. In this paper, various statistical techniques are applied using water supply operation data (flow, pressure, water level, etc) to prepare the foundation for the selection of the optimal operating range and advancement of the monitoring and alarm system. In addition, the arrival time is analyzed through cross-correlation analysis of changes in raw water turbidity between the water intake and water treatment plants. The purpose of this paper is to study the model that predicts the raw water turbidity of a water treatment plant by applying raw water turbidity data considering the time delay according to the flow rate change.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.3
• /
• pp.166-172
• /
• 2022

In this study, we photograph driving obstacle objects such as crosswalks, side spheres, manholes, braille blocks, partial ramps, temporary safety barriers, stairs, and inclined curb that hinder or cause inconvenience to the movement of the vulnerable using electric mobility aids. We develop an optimal AI model that classifies photographed objects and automatically recognizes them, and implement an algorithm that can efficiently determine obstacles in front of electric mobility aids. In order to enable object detection to be AI learning with high probability, the labeling form is labeled as a polygon form when building a dataset. It was developed using a Mask R-CNN model in Detectron2 framework that can detect objects labeled in the form of polygons. Image acquisition was conducted by dividing it into two groups: the general public and the transportation weak, and image information obtained in two areas of the test bed was secured. As for the parameter setting of the Mask R-CNN learning result, it was confirmed that the model learned with IMAGES_PER_BATCH: 2, BASE_LEARNING_RATE 0.001, MAX_ITERATION: 10,000 showed the highest performance at 68.532, so that the user can quickly and accurately recognize driving risks and obstacles.