• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.40 no.5
• /
• pp.381-391
• /
• 2022

This paper is intended to find one of the prevailing deep learning models that are a type of AI (Artificial Intelligence) that helps rapidly detect damaged buildings where disasters occur. The models selected are SSD-512, RetinaNet, and YOLOv3 which are widely used in object detection in recent years. These models are based on one-stage detector networks that are suitable for rapid object detection. These are often used for object detection due to their advantages in structure and high speed but not for damaged building detection in disaster management. In this study, we first trained each of the algorithms on xBD dataset that provides the post-disaster imagery with damage classification labels. Next, the three models are quantitatively evaluated with the mAP(mean Average Precision) and the FPS (Frames Per Second). The mAP of YOLOv3 is recorded at 34.39%, and the FPS reached 46. The mAP of RetinaNet recorded 36.06%, which is 1.67% higher than YOLOv3, but the FPS is one-third of YOLOv3. SSD-512 received significantly lower values than the results of YOLOv3 on two quantitative indicators. In a disaster situation, a rapid and precise investigation of damaged buildings is essential for effective disaster response. Accordingly, it is expected that the results obtained through this study can be effectively used for the rapid response in disaster management.

• Korean Journal of Remote Sensing
• /
• v.38 no.5_1
• /
• pp.461-469
• /
• 2022

This paper presents a novel training data extraction approach using semi-supervised learning (SSL)-based classification without the analyst intervention for time-series land-cover mapping. The SSL-based approach first performs initial classification using initial training data obtained from past images including land-cover characteristics similar to the image to be classified. Reliable training data from the initial classification result are then extracted from SSL-based iterative classification using classification uncertainty information and class labels of neighboring pixels as constraints. The potential of the SSL-based training data extraction approach was evaluated from a classification experiment using unmanned aerial vehicle images in croplands. The use of new training data automatically extracted by the proposed SSL approach could significantly alleviate the misclassification in the initial classification result. In particular, isolated pixels were substantially reduced by considering spatial contextual information from adjacent pixels. Consequently, the classification accuracy of the proposed approach was similar to that of classification using manually extracted training data. These results indicate that the SSL-based iterative classification presented in this study could be effectively applied to automatically extract reliable training data for time-series land-cover mapping.

• 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.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.3
• /
• pp.1100-1118
• /
• 2021

In a wide range of ML applications, the training data contains privacy-sensitive information that should be kept secure. Training the ML systems by privacy-sensitive data makes the ML model inherent to the data. As the structure of the model has been fine-tuned by training data, the model can be abused for accessing the data by the estimation in a reverse process called model inversion attack (MIA). Although, MIA has been applied to shallow neural network models of recognizers in literature and its threat in privacy violation has been approved, in the case of a deep learning (DL) model, its efficiency was under question. It was due to the complexity of a DL model structure, big number of DL model parameters, the huge size of training data, big number of registered users to a DL model and thereof big number of class labels. This research work first analyses the possibility of MIA on a deep learning model of a recognition system, namely a face recognizer. Second, despite the conventional MIA under the white box scenario of having partial access to the users' non-sensitive information in addition to the model structure, the MIA is implemented on a deep face recognition system by just having the model structure and parameters but not any user information. In this aspect, it is under a semi-white box scenario or in other words a gray-box scenario. The experimental results in targeting five registered users of a CNN-based face recognition system approve the possibility of regeneration of users' face images even for a deep model by MIA under a gray box scenario. Although, for some images the evaluation recognition score is low and the generated images are not easily recognizable, but for some other images the score is high and facial features of the targeted identities are observable. The objective and subjective evaluations demonstrate that privacy cyber-attack by MIA on a deep recognition system not only is feasible but also is a serious threat with increasing alert state in the future as there is considerable potential for integration more advanced ML techniques to MIA.

• Journal of Korean Home Economics Education Association
• /
• v.33 no.2
• /
• pp.45-56
• /
• 2021

This study evaluated the dietary habits and health behaviors of Korean adolescents according to their nutrition label utilization. This study was conducted on the adolescents aged between 12 and 18 who participated in the 2016-2018 National Health and Nutrition Examination Survey. Subjects were classified into nutrition label utilizing group and non-utilizing group. Then the general information, diet habit and nutrient intakes, health status and obesity and mental health of the two groups were analyzed. There was no significant difference in skipping breakfast, frequency of eating out, smoking status and alcohol drinking status according to the utilization of nutrition labels. Nutrition label utilizing group had higher % KDRI of protein(p<0.05), calcium (p<0.01), phosphorus(p<0.01) and potassium intakes(p<0.01) than non-utilizing group. Nutrition label utilizing group had higher EQ-5D score(0.97) than non-utilizing group(p<0.001). Practical nutrition education using nutrition label is needed. So these results can be useful for supporting dietary education regarding use of nutrition label for adolescents.

• Journal of the Korea Computer Graphics Society
• /
• v.28 no.3
• /
• pp.55-65
• /
• 2022

Recently, deep learning-based automated systems for identifying and detecting landmarks have been proposed. In order to train such a deep learning-based model without overfitting, a large amount of image and labeling data is required. Conventionally, an experienced reader manually identifies and labels landmarks in a patient's image. However, such measurement is not only expensive, but also has poor reproducibility, so the need for an automated labeling method has been raised. In addition, in the X-ray image, since various human tissues on the path through which the photons pass are displayed, it is difficult to identify the landmark compared to a general natural image or a 3D image modality image. In this study, we propose a geometric data augmentation technique that enables the generation of a large amount of labeling data in X-ray images. In addition, the optimal attention mechanism for landmark detection was presented through the implementation and application of various attention techniques to improve the detection performance of 16 major landmarks in the skull. Finally, among the major cranial landmarks, markers that ensure stable detection are derived, and these markers are expected to have high clinical application potential.

• Smart Structures and Systems
• /
• v.29 no.1
• /
• pp.1-16
• /
• 2022

Despite recent breakthroughs in deep learning and computer vision fields, the pixel-wise identification of tiny objects in high-resolution images with complex disturbances remains challenging. This study proposes a modified U-net for tiny crack segmentation in real-world steel-box-girder bridges. The modified U-net adopts the common U-net framework and a novel Self-Attention-Self-Adaption (SASA) neuron as the fundamental computing element. The Self-Attention module applies softmax and gate operations to obtain the attention vector. It enables the neuron to focus on the most significant receptive fields when processing large-scale feature maps. The Self-Adaption module consists of a multiplayer perceptron subnet and achieves deeper feature extraction inside a single neuron. For data augmentation, a grid-based crack random elastic deformation (CRED) algorithm is designed to enrich the diversities and irregular shapes of distributed cracks. Grid-based uniform control nodes are first set on both input images and binary labels, random offsets are then employed on these control nodes, and bilinear interpolation is performed for the rest pixels. The proposed SASA neuron and CRED algorithm are simultaneously deployed to train the modified U-net. 200 raw images with a high resolution of 4928 × 3264 are collected, 160 for training and the rest 40 for the test. 512 × 512 patches are generated from the original images by a sliding window with an overlap of 256 as inputs. Results show that the average IoU between the recognized and ground-truth cracks reaches 0.409, which is 29.8% higher than the regular U-net. A five-fold cross-validation study is performed to verify that the proposed method is robust to different training and test images. Ablation experiments further demonstrate the effectiveness of the proposed SASA neuron and CRED algorithm. Promotions of the average IoU individually utilizing the SASA and CRED module add up to the final promotion of the full model, indicating that the SASA and CRED modules contribute to the different stages of model and data in the training process.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.10
• /
• pp.11-17
• /
• 2022

In this paper, a model that can increase the efficiency of work in arranging interior furniture by applying augmented reality technology was studied. In the existing system to which augmented reality is currently applied, there is a problem in that information is limitedly provided depending on the size and nature of the company's product when outputting the image of furniture. To solve this problem, this paper presents an AR labeling algorithm. The AR labeling algorithm extracts feature points from the captured images and builds a database including indoor location information. A method of detecting and learning the location data of furniture in an indoor space was adopted using the CNN technique. Through the learned result, it is confirmed that the error between the indoor location and the location shown by learning can be significantly reduced. In addition, a study was conducted to allow users to easily place desired furniture through augmented reality by receiving detailed information about furniture along with accurate image extraction of furniture. As a result of the study, the accuracy and loss rate of the model were found to be 99% and 0.026, indicating the significance of this study by securing reliability. The results of this study are expected to satisfy consumers' satisfaction and purchase desires by accurately arranging desired furniture indoors through the design and implementation of AR labels.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.21 no.5
• /
• pp.274-286
• /
• 2022

Multi-object tracking has been studied for a long time under computer vision and plays a critical role in applications such as autonomous driving and driving assistance. Multi-object tracking techniques generally consist of a detector that detects objects and a tracker that tracks the detected objects. Various publicly available datasets allow us to train a detector model without much effort. However, there are relatively few publicly available datasets for training a tracker model, and configuring own tracker datasets takes a long time compared to configuring detector datasets. Hence, the detector is often developed separately with a tracker module. However, the separated tracker should be adjusted whenever the former detector model is changed. This study proposes a system that can train a model that performs detection and tracking simultaneously using only the detector training datasets. In particular, a Siam network with augmentation is used to compose the detector and tracker. Experiments are conducted on public datasets to verify that the proposed algorithm can formulate a real-time multi-object tracker comparable to the state-of-the-art tracker models.

• Journal of Nutrition and Health
• /
• v.56 no.2
• /
• pp.184-202
• /
• 2023

Purpose: This study investigates the necessary educational materials from the perspective of caregivers managing food allergies of preschoolers. Methods: An online survey enrolled 400 caregivers of preschoolers with food allergies. The study was conducted in July 2022, and data were analyzed using the IBM SPSS Statistics 25. Results: The caregivers of preschoolers with food allergies showed low food allergy knowledge. The highest percentage of incorrect answers were for items related to the cause (74.2%) and response (65.2%) to food allergies. Moreover, categorizing the levels of knowledge as high, medium, and low, we found that caregivers with younger children exhibited a better understanding of food allergies. Approximately 56.9% of subjects in the high class were caregivers with preschoolers under three years, whereas caregivers with preschoolers over four years comprised 43.2% of the high-class subjects. Regarding food allergy treatment, four questions measured how well the caregivers tended to the children. With higher levels of knowledge, caregivers were more likely to read nutrient information and check for food allergens. The caregivers of preschoolers who were clinically diagnosed with food allergies demonstrated a higher percentage of using separate ingredients and cooking utensils. Conclusion: This study shows how the caregivers' knowledge levels are critical in managing and treating food allergies among preschoolers. With caregivers' lower knowledge of food allergies and preschoolers younger than four years, self-diagnosis of food allergies and existing food allergies were not managed efficiently. Hence, customized education for caregivers is essential to improve their knowledge level regarding limiting foods that may cause allergies, reading food labels, and using ingredients and cooking utensils separately.