• Smart Structures and Systems
• /
• 제31권4호
• /
• pp.365-381
• /
• 2023

The existing vision-based techniques for inspection and condition assessment of civil infrastructure are mostly manual and consequently time-consuming, expensive, subjective, and risky. As a viable alternative, researchers in the past resorted to deep learning-based autonomous damage detection algorithms for expedited post-disaster reconnaissance of structures. Although a number of automatic damage detection algorithms have been proposed, the scarcity of labeled training data remains a major concern. To address this issue, this study proposed a semi-supervised learning (SSL) framework based on consistency regularization and cross-supervision. Image data from post-earthquake reconnaissance, that contains cracks, spalling, and exposed rebars are used to evaluate the proposed solution. Experiments are carried out under different data partition protocols, and it is shown that the proposed SSL method can make use of unlabeled images to enhance the segmentation performance when limited amount of ground truth labels are provided. This study also proposes DeepLab-AASPP and modified versions of U-Net++ based on channel-wise attention mechanism to better segment the components and damage areas from images of reinforced concrete buildings. The channel-wise attention mechanism can effectively improve the performance of the network by dynamically scaling the feature maps so that the networks can focus on more informative feature maps in the concatenation layer. The proposed DeepLab-AASPP achieves the best performance on component segmentation and damage state segmentation tasks with mIoU scores of 0.9850 and 0.7032, respectively. For crack, spalling, and rebar segmentation tasks, modified U-Net++ obtains the best performance with Igou scores (excluding the background pixels) of 0.5449, 0.9375, and 0.5018, respectively. The proposed architectures win the second place in IC-SHM2021 competition in all five tasks of Project 2.

• Smart Structures and Systems
• /
• 제29권1호
• /
• 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.

• 국제학술발표논문집
• /
• The 9th International Conference on Construction Engineering and Project Management
• /
• pp.543-550
• /
• 2022

The U.S. Department of Energy conducts the Solar Decathlon competition as a student-based achievement that encourages sustainable design with energy efficiency and solar energy technologies. In the 2020 competition, the University of Nevada, Las Vegas (UNLV) team designed, fabricated, and constructed a net-zero modular house that applies innovative and highly efficient building technologies. This paper focused on the lessons learned during the early phases of this ongoing modular project. The research methodology included obtaining feedback from key project participants using a well-structured questionnaire. The results showed that the major items/challenges in the project's planning phase included selecting the modular size, planning the construction system, planning the materials and procurement, estimating costs and duration, selecting a fabricator, collaboration and communication, safety, and planning module transportation. These findings will help modular practitioners and future Solar Decathlon competition participants better understand how and what factors they should consider most during the early phases through the lessons learned.

• 한국인공지능학회지
• /
• 제11권3호
• /
• pp.17-22
• /
• 2023

In this paper, we propose a black ice detection system using Convolutional Neural Networks (CNNs). Black ice poses a serious threat to road safety, particularly during winter conditions. To overcome this problem, we introduce a CNN-based architecture for real-time black ice detection with an encoder-decoder network, specifically designed for real-time black ice detection using thermal images. To train the network, we establish a specialized experimental platform to capture thermal images of various black ice formations on diverse road surfaces, including cement and asphalt. This enables us to curate a comprehensive dataset of thermal road black ice images for a training and evaluation purpose. Additionally, in order to enhance the accuracy of black ice detection, we propose a multi-scale dilation convolution feature fusion (MsDC-FF) technique. This proposed technique dynamically adjusts the dilation ratios based on the input image's resolution, improving the network's ability to capture fine-grained details. Experimental results demonstrate the superior performance of our proposed network model compared to conventional image segmentation models. Our model achieved an mIoU of 95.93%, while LinkNet achieved an mIoU of 95.39%. Therefore, it is concluded that the proposed model in this paper could offer a promising solution for real-time black ice detection, thereby enhancing road safety during winter conditions.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2001년도 가을 학술발표논문집 Vol.28 No.2 (3)
• /
• pp.85-87
• /
• 2001

하드웨어와 소프트웨어의 발전과 함께 컴퓨터에서 처리 해야 할 데이터량이 크게 증가하고 있다. 클러스터 내의 node들 사이에서 이런 대용량의 데이터들을 보다 빠르게 전송하기 위해서 Lightweight Messaging 기법이 등장하였으며, 대표적으로 AM, FM, U-Net, VIA등이 있다. 이 중에서 VIA는 커널 수준에서 구현된 TCP/IP를 대신해서 사용자 수준에서 커널을 거치지 않고 네트워크 장치와 직접적으로 통신을 할 수 있게 하여 다양한 분야에서 사용되고 있으며, 새로운 프로토콜의 표준으로 자리를 잡아가고 있다. 그러나 이러한 장점에도 불구하고 프로그래밍의 난이성 때문에 제대로 숙지하기 가지는 많은 시간의 투자가 필요한 것이 사실이다. 이에 이 논문에서는 EVIL(Easy-to-use Virtual Interface Library)이라는, 개발자들이 좀더 쉽게 접근할 수 있는 라이브러리를 제안하였다 그리고 EVIL. Native VIA TCP/IP로 각각 같은 역할을 하는 프로그램을 작성하여 기존의 프로토콜들과 성능을 비교하였다.

• International Journal of Naval Architecture and Ocean Engineering
• /
• 제4권4호
• /
• pp.412-422
• /
• 2012

Air lubrication techniques for frictional drag reduction on ships have been proposed by numerous researchers since the 19th century. However, these techniques have not been widely adopted as questions persist about their drag reduction performance beyond the laboratory, as well as energy and economic cost-benefit. This paper draws on data from the literature to consider the suitability of air lubrication for large ocean going and U.S. Great Lakes ships, by establishing the basic energy economic calculations and presenting results for a hypothetical air lubricated ship. All the assumptions made in the course of the analysis are clearly stated so that they can be refined when considering application of air lubrication to a specific ship. The analysis suggests that, if successfully implemented, both air layer and partial cavity drag reduction could lead to net energy savings of 10 to 20%, with corresponding reductions in emissions.

• 정보와 통신
• /
• 제30권11호
• /
• pp.75-83
• /
• 2013

국방 정보분야의 효율 효과적인 개발 및 운용을 위해 통합과 표준의 설정 등을 통한 상호운용성 확보가 요구된다. 상호운용성 표준화 기술은 전사적 아키텍처 기반 상호운용성 조정 통제, 플랫폼 중심의 상호운용성 확보를 위한 공통운용환경 및 데이터공유환경 구축에서 시작하여, 현재에는 네트워크 중심 환경에 적합한 SOA(Service Oriented Architecture) 기반 상호운용성 증진 전략으로 패러다임이 변화하고 있다. 미군의 경우 단일체계 관점에서 복합체계 관점으로 확대 적용을 위하여 기존의 공통운용환경 및 데이터 공유환경을 SOA 기반의 NCES(Net-Centric Enterprise Service)로 전환을 추진해왔다. 또한 이와 병행하여 미 육군 수준에서는 전쟁을 수행하는 각 객체를 컴퓨팅 환경으로 구분하여 육군 차원의 공통운용환경을 구축하고 있다. 이에 본 고에서는 국방 공통운용환경의 동향을 미군의 사례를 중심으로 살펴보고 시사점을 도출한다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권1호
• /
• pp.60-79
• /
• 2022

The U-Net architecture-based segmentation models attained remarkable performance in numerous medical image segmentation missions like skin lesion segmentation. Nevertheless, the resolution gradually decreases and the loss of spatial information increases with deeper network. The fusion of adjacent layers is not enough to make up for the lost spatial information, thus resulting in errors of segmentation boundary so as to decline the accuracy of segmentation. To tackle the issue, we propose a new deep learning-based segmentation model. In the decoding stage, the feature channels of each decoding unit are concatenated with all the feature channels of the upper coding unit. Which is done in order to ensure the segmentation effect by integrating spatial and semantic information, and promotes the robustness and generalization of our model by combining the atrous spatial pyramid pooling (ASPP) module and channel attention module (CAM). Extensive experiments on ISIC2016 and ISIC2017 common datasets proved that our model implements well and outperforms compared segmentation models for skin lesion segmentation.

• Nuclear Engineering and Technology
• /
• 제49권3호
• /
• pp.517-524
• /
• 2017

Cyber security is an important issue in the field of nuclear engineering because nuclear facilities use digital equipment and digital systems that can lead to serious hazards in the event of an accident. Regulatory agencies worldwide have announced guidelines for cyber security related to nuclear issues, including U.S. NRC Regulatory Guide 5.71. It is important to evaluate cyber security risk in accordance with these regulatory guides. In this study, we propose a cyber security risk evaluation model for nuclear instrumentation and control systems using a Bayesian network and event trees. As it is difficult to perform penetration tests on the systems, the evaluation model can inform research on cyber threats to cyber security systems for nuclear facilities through the use of prior and posterior information and backpropagation calculations. Furthermore, we suggest a methodology for the application of analytical results from the Bayesian network model to an event tree model, which is a probabilistic safety assessment method. The proposed method will provide insight into safety and cyber security risks.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2021년도 추계학술발표대회
• /
• pp.700-703
• /
• 2021

Infrared cameras are widely used in recent research for automatic monitoring the abnormal behaviors of the pig. However, when deployed in real pig farms, infrared cameras always get polluted due to the harsh environment of pig farms which negatively affects the performance of pig monitoring. In this paper, we propose a real-time noise-robust infrared camera-based pig automatic monitoring system to improve the robustness of pigs' automatic monitoring in real pig farms. The proposed system first uses a preprocessor with a U-Net architecture that was trained as a GAN generator to transform the noisy images into clean images, then uses a YOLOv5-based detector to detect pigs. The experimental results show that with adding the preprocessing step, the average pig detection precision improved greatly from 0.639 to 0.759.