• Journal of Korean Tunnelling and Underground Space Association
• /
• v.23 no.6
• /
• pp.549-558
• /
• 2021

In order to keep infrastructures such as tunnels and underground facilities safe, cracks of concrete lining in tunnel should be detected by regular inspections. Since regular inspections are accomplished through manual efforts using maintenance lift vehicles, it brings about traffic jam, exposes works to dangerous circumstances, and deteriorates consistency of crack inspection data. This study aims to provide methodology to automatically extract cracks from tunnel concrete lining images generated by the existing tunnel image acquisition system. Specifically, we train a deep learning based semantic segmentation model with open dataset, and evaluate its performance with the dataset from the existing tunnel image acquisition system. In particular, we compare the model performance in case of using all of a public dataset, subset of the public dataset which are related to tunnel surfaces, and the tunnel-related subset with negative examples. As a result, the model trained using the tunnel-related subset with negative examples reached the best performance. In the future, we expect that this research can be used for planning efficient model training strategy for crack detection.

• Smart Structures and Systems
• /
• v.29 no.1
• /
• pp.195-206
• /
• 2022

Crack detection is essential for inspection of existing structures and crack segmentation based on deep learning is a significant solution. However, datasets are usually one of the key issues. When building a new dataset for deep learning, laborious and time-consuming annotation of a large number of crack images is an obstacle. The aim of this study is to develop an approach that can automatically select a small portion of the most informative crack images from a large pool in order to annotate them, not to label all crack images. An active learning method with difficulty learning mechanism for crack segmentation tasks is proposed. Experiments are carried out on a crack image dataset of a steel box girder, which contains 500 images of 320×320 size for training, 100 for validation, and 190 for testing. In active learning experiments, the 500 images for training are acted as unlabeled image. The acquisition function in our method is compared with traditional acquisition functions, i.e., Query-By-Committee (QBC), Entropy, and Core-set. Further, comparisons are made on four common segmentation networks: U-Net, DeepLabV3, Feature Pyramid Network (FPN), and PSPNet. The results show that when training occurs with 200 (40%) of the most informative crack images that are selected by our method, the four segmentation networks can achieve 92%-95% of the obtained performance when training takes place with 500 (100%) crack images. The acquisition function in our method shows more accurate measurements of informativeness for unlabeled crack images compared to the four traditional acquisition functions at most active learning stages. Our method can select the most informative images for annotation from many unlabeled crack images automatically and accurately. Additionally, the dataset built after selecting 40% of all crack images can support crack segmentation networks that perform more than 92% when all the images are used.

• Journal of KIISE:Software and Applications
• /
• v.31 no.5
• /
• pp.595-604
• /
• 2004

Automatic analysis of concepts or semantic relations from text documents enables not only an efficient acquisition of relevant information, but also a comparison of documents in the concept level. We present a multiple cause model-based approach to text analysis, where latent topics are automatically extracted from document sets and similarity between documents is measured by semantic kernels constructed from the extracted topics. In our approach, a document is assumed to be generated by various combinations of underlying topics. A topic is defined by a set of words that are related to the same topic or cooccur frequently within a document. In a network representing a multiple-cause model, each topic is identified by a group of words having high connection weights from a latent node. In order to facilitate teaming and inferences in multiple-cause models, some approximation methods are required and we utilize an approximation by Helmholtz machines. In an experiment on TDT-2 data set, we extract sets of meaningful words where each set contains some theme-specific terms. Using semantic kernels constructed from latent topics extracted by multiple cause models, we also achieve significant improvements over the basic vector space model in terms of retrieval effectiveness.

• The Journal of Korea Robotics Society
• /
• v.17 no.3
• /
• pp.245-254
• /
• 2022

Sensor dataset for autonomous driving is one of the essential components as the deep learning approaches are widely used. However, most driving datasets are focused on typical environments such as sunny or cloudy. In addition, most datasets deal with color images and lidar. In this paper, we propose a driving dataset with multi-spectral images and lidar in adverse weather conditions such as snowy, rainy, smoky, and dusty. The proposed data acquisition system has 4 types of cameras (color, near-infrared, shortwave, thermal), 1 lidar, 2 radars, and a navigation sensor. Our dataset is the first dataset that handles multi-spectral cameras in adverse weather conditions. The Proposed dataset is annotated as 2D semantic labels, 3D semantic labels, and 2D/3D bounding boxes. Many tasks are available on our dataset, for example, object detection and driveable region detection. We also present some experimental results on the adverse weather dataset.

• Human Ecology Research
• /
• v.58 no.4
• /
• pp.613-633
• /
• 2020

This study identifies consumers' perceived benefits and costs when using Samsung Health (a healthcare app) based on consumer reviews from Google Play Store's app and social media discourse. We examine the differences in the benefits and the costs of Samsung Health using these two sources of data. We conducted text frequency analysis, clustering analysis, and semantic network analysis using R programming. The major findings are as follows. First, consumers experience benefits and costs on several functions of the app, such as step counting, device interlocking, information acquisition, and competition with global consumers. Second, the results of semantic network analysis showed that there were eight benefit factors and three cost factors. We also found that the three costs correspond to the benefits, indicating that some consumers gained benefits from certain functions while others gained costs from the same functions. Third, the comparison between consumer app review and social media discourse showed that the former is appropriate to assess the performance of app functions, while the latter is appropriate to examine how the app is used in daily life and how consumers feel about it. The current study suggests managerial implications to healthcare app service providers regarding what they should strengthen and improve to enhance consumers' satisfaction. It also suggests some implications from the two media, which can be mutually complementary, for researchers who study consumer opinions.

• Journal of Intelligence and Information Systems
• /
• v.12 no.3
• /
• pp.95-110
• /
• 2006

To enhance the rule-based reasoning capability of Semantic Web, the XRML (eXtensible Rule Markup Language) approach embraces the meta-information necessary for the extraction of explicit rules from Web pages and its maintenance. To effectuate the automatic identification of rules from unstructured texts, this research develops a framework of using rule ontology. The ontology can be acquired from a similar site first, and then can be used for multiple sites in the same domain. The procedure of ontology-based rule identification is regarded as a graph search problem with incomplete nodes, and an A* algorithm is devised to solve the problem. The procedure is demonstrated with the domain of shipping rates and return policy comparison portal, which needs rule based reasoning capability to answer the customer's inquiries. An example ontology is created from Amazon.com, and is applied to the many online retailers in the same domain. The experimental result shows a high performance of this approach.

• Asian review of World Histories
• /
• v.4 no.1
• /
• pp.125-132
• /
• 2016

This paper, as a conlcusion to this special issue, presents the future work that is being carried out at NTU Singapore in collaboration with Microsoft Research and Microsoft Azure for Research. For our research team the real frontier research in world histories starts when we want to use computers to structure historical information, model historical narratives, simulate theoretical large scale hypotheses, and incent world historians to use virtual assistants and/or engage them in teamwork using social media and/or seduce them with immersive spaces to provide new learning and sharing environments, in which new things can emerge and happen: "You do not know which will be the next idea. Just repeating the same things is not enough" (Carlo Rubbia, 1984 Nobel Price in Physics, at Nanyang Technological University on January 19, 2016).

• Journal of Korean language education
• /
• v.23 no.1
• /
• pp.161-182
• /
• 2012

The purpose of this study is to examine level of individual foreign learners' quantitative knowledge of korean vocabulary through the Foreign Learners' Korean Vocabulary Knowledge Assessment and to propose a vocabulary education plan reflecting the result. This assessment was written in focus of semantic relation to identify category, integrated and analytical knowledge of Korean verbs acquired by foreign learners and as a result, the following educational implications were made. First, the learners' knowledge about the vocabulary combinational relationship was very limited. Second, learners are not interested in vocabulary's synonymous relation and antonymous relations. Lastly, there needs to be a concern about postpositional particle-use education. The results of these kinds of research, observes the learners vocabulary acquisition process to determine Korean textbook and present vocabulary level in appropriate in terms of learners' perspective, and provide an opportunity to check if current professors' method is adequate and valid.

• Journal of the Korean earth science society
• /
• v.31 no.3
• /
• pp.288-299
• /
• 2010

Recently, scientific literacy means not only the acquisition of scientific knowledge but also the linguistic ability to participate in a scientific discourse community. Keeping this in mind, this study investigated middle school students' writings about phase transitions of water in air. Sixty seven students at 9th grade (age 15) students participated in this study and wrote two individual short texts. The result of text analysis can be summarized as follows: (1) students had problems with familiar scientific terms such as 'water vapor' and 'steam' as well as unfamiliar ones like 'dew point'. (2) Students described right semantic relations and at the same time wrong ones more in the idea formed from everyday experience than those from school instruction. (3) While students showed action and process centered writing in text about everyday phenomenon, they showed more preference for technical words and nouns in text about school science. This study suggest that students could develop linguistic ability of science from both spontaneous process based on experience and formal and theoretical learning; the former in forming various semantic relations, the latter in technical and abstract aspect of scientific writing.

• Korean Journal of Remote Sensing
• /
• v.37 no.6_1
• /
• pp.1573-1587
• /
• 2021

A Drone image is an ultra-high-resolution image that is several or tens of times higher in spatial resolution than a satellite or aerial image. Therefore, drone image-based remote sensing is different from traditional remote sensing in terms of the level of object to be extracted from the image and the amount of data to be processed. In addition, the optimal scale and size of data used for model training is different depending on the characteristics of the applied deep learning model. However, moststudies do not consider the size of the object to be found in the image, the spatial resolution of the image that reflects the scale, and in many cases, the data specification used in the model is applied as it is before. In this study, the effect ofspatial resolution and image size of drone image on the accuracy and training time of the semantic segmentation deep learning model of six wintering vegetables was quantitatively analyzed through experiments. As a result of the experiment, it was found that the average accuracy of dividing six wintering vegetablesincreases asthe spatial resolution increases, but the increase rate and convergence section are different for each crop, and there is a big difference in accuracy and time depending on the size of the image at the same resolution. In particular, it wasfound that the optimal resolution and image size were different from each crop. The research results can be utilized as data for getting the efficiency of drone images acquisition and production of training data when developing a winter vegetable segmentation model using drone images.