• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.10
• /
• pp.3608-3626
• /
• 2021

The detection of bolts is an important task in high-speed train inspection systems, and it is frequently performed to ensure the safety of trains. The difficulty of the vision-based bolt inspection system lies in small sample defect detection, which makes the end-to-end network ineffective. In this paper, the problem is resolved in two stages, which includes the detection network and cascaded classification networks. For small bolt detection, all bolts including defective bolts and normal bolts are put together for conducting annotation training, a new loss function and a new boundingbox selection based on the smallest axis-aligned convex set are proposed. These allow YOLOv3 network to obtain the accurate position and bounding box of the various bolts. The average precision has been greatly improved on PASCAL VOC, MS COCO and actual data set. After that, the Siamese network is employed for estimating the status of the bolts. Using the convolutional Siamese network, we are able to get strong results on few-shot classification. Extensive experiments and comparisons on actual data set show that the system outperforms state-of-the-art algorithms in bolt inspection.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.11
• /
• pp.4426-4442
• /
• 2020

Since scene segmentation is becoming a hot topic in the field of autonomous driving and medical image analysis, researchers are actively trying new methods to improve segmentation accuracy. At present, the main issues in image semantic segmentation are intra-class inconsistency and inter-class indistinction. From our analysis, the lack of global information as well as macroscopic discrimination on the object are the two main reasons. In this paper, we propose a Densely connected residual Attention network (DA-Res2Net) which consists of a dense residual network and channel attention guidance module to deal with these problems and improve the accuracy of image segmentation. Specifically, in order to make the extracted features equipped with stronger multi-scale characteristics, a densely connected residual network is proposed as a feature extractor. Furthermore, to improve the representativeness of each channel feature, we design a Channel-Attention-Guide module to make the model focusing on the high-level semantic features and low-level location features simultaneously. Experimental results show that the method achieves significant performance on various datasets. Compared to other state-of-the-art methods, the proposed method reaches the mean IOU accuracy of 83.2% on PASCAL VOC 2012 and 79.7% on Cityscapes dataset, respectively.

• Journal of English Language & Literature
• /
• v.64 no.3
• /
• pp.469-481
• /
• 2018

When literatures and cultures encounter their counterparts in terms of the big data or statistics of a new reconfiguration in the cognitive map, the tangential points of the borderland will be reduced to what Mitchell calls "a mere abstraction on a map," which nevertheless will provide a vast interstitial zone of "intersections, competition, and exclusions." This zone will be the dynamic vortex for the aesthetics, politics, and ethics of cultural translation. The translated discourse will engage in carrying across the disturbing region of untranslatability and demonstrate how the literary texts of world literature reveal enriching but threatening human experience. This dynamic border of vortex will construct the translational space of world literature, transcending the fragmentary untranslatable nature of the hybrid convergence of the ethnic, racial, cultural and national intermixtures and constructing what Pascal Casanova terms "The World Republic of Letters." In this paper, I will demonstrate how the very concept of scale is related to literary space as well as how distance creates a poetics of literary landscapes which looks ahead of world literature. Also, I will attempt to find the possibility to relate the "micro-scale" with the "macro-scale," and to construct the scale politics of representation. "Glocalization" is a convenient theoretical tool for the double movement of the up-scale and down-scale.

• Structural Engineering and Mechanics
• /
• v.77 no.6
• /
• pp.819-830
• /
• 2021

To evaluate the performance of concrete load bearing walls in a structure under horizontal loads after being exposed to real fire, two steps were followed. In the first step, an experimental study was performed on the thermo-mechanical properties of concrete after heating to temperatures of 200-1000℃ with the purpose of determining the residual mechanical properties after cooling. The temperature was increased in line with natural fire curve in an electric furnace. The peak temperature was maintained for a period of 1.5 hour and then allowed to cool gradually in air at room temperature. All specimens were made from calcareous aggregate to be used for determining the residual properties: compressive strength, static and dynamic elasticity modulus by means of UPV test, including the mass loss. The concrete residual compressive strength and elastic modulus values were compared with those calculated from Eurocode and other analytical models from other studies, and were found to be satisfactory. In the second step, experimental analysis results were then implemented into structural numerical analysis to predict the post-fire load-bearing capacity response of the walls under vertical and horizontal loads. The parameters considered in this analysis were the effective height, the thickness of the wall, various support conditions and the residual strength of concrete. The results indicate that fire damage does not significantly affect the lateral capacity and stiffness of reinforced walls for temperature fires up to 400℃.

• Korean Journal of Remote Sensing
• /
• v.37 no.4
• /
• pp.747-761
• /
• 2021

With the increasing application of UAVs in intelligent transportation systems, vehicle detection for aerial images has become an essential engineering technology and has academic research significance. In this paper, a vehicle detection method for aerial images based on the YOLOv4 deep learning algorithm is presented. At present, the most known datasets are VOC (The PASCAL Visual Object Classes Challenge), ImageNet, and COCO (Microsoft Common Objects in Context), which comply with the vehicle detection from UAV. An integrated dataset not only reflects its quantity and photo quality but also its diversity which affects the detection accuracy. The method integrates three public aerial image datasets VAID, UAVD, DOTA suitable for YOLOv4. The training model presents good test results especially for small objects, rotating objects, as well as compact and dense objects, and meets the real-time detection requirements. For future work, we will integrate one more aerial image dataset acquired by our lab to increase the number and diversity of training samples, at the same time, while meeting the real-time requirements.

• Computers and Concrete
• /
• v.29 no.5
• /
• pp.347-360
• /
• 2022

Fiber-reinforced cementitious composites (FRCC) have emerged as a response to the calls for strong, ductile and sustainable concrete mixes. FRCC has shown outstanding mechanical properties and ductility where special fibres are used in the mixes to give it the strength and the ability to exhibit strain hardening. With the possibility of designing the FRCC mixes to include sustainable constituents and by-products materials such as fly ash, FRCC started to emerge as a green alternative as well. To be able to design mixes that achieve these conflicting properties in concrete, there is a need to understand the composition effect on FRCC and optimize these compositions. Therefore, this paper aims to investigate the influence of FRCC compositions on the properties of fresh and hardened of FRCC and then to optimize these mix compositions using factorial design approach. Three factors, water-to-binder ratio (w/b), mineral admixtures (total of fly ash and metakaolin by cement content (MAR)), and metakaolin content (MK), were investigated to determine their effects on the properties of fresh and hardened FRCC. The results show the importance of combining both FA and MK in obtaining a satisfactory fresh and mechanical properties of FRCC. Models were suggested to elucidate the role of the studied factors and a method for optimization was proposed.

• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.2
• /
• pp.42.1-42.1
• /
• 2021

Galaxy/Halo finding based on the friends-of-friend (FoF) algorithm has been widely adopted for its simplicity and expandability to the phase-space. However, cosmological simulations have been progressively bigger in size and more accurate in resolutions, resulting in that galaxy/halo finding gets computationally expensive more and more. In fact, we confirm this issue through our exercise of applying the 6-dimensional (6D) FoF galaxy finder code, VELOCIraptor (Elahi et al.2019) on the NewHorizon simulation (Dubois et al. 2021), in which typical galaxies with about 1e11 M_sun (10⁷ particles) are identified with very low speed (longer than a day). We have applied several improvements to the original VELOCIraptor code that solve the low-performance problem of galaxy finding on a simulation with high resolutions. Our modifications find the exact same FoF group and can be readily applied to any tree-based FoF code, achieving a 2700 (12) times speedup in the 3D (6D) FoF search compared to the original execution. We applied the updated version of VELOCIraptor on the entire NewHorizon simulation (834 snapshots) and identified its galaxies and halos. We present several quick comparisons of galaxy properties with those with GALAXYMaker data.

• Journal of Nuclear Fuel Cycle and Waste Technology(JNFCWT)
• /
• v.21 no.4
• /
• pp.571-576
• /
• 2023

The initial development plans for the six reactor designs, soon after the release of Generation IV International Forum (GIF) TRM in 2002, were characterized by high ambition [1]. Specifically, the sodium-cooled fast reactor (SFR) and very-high temperature reactor (VHTR) gained significant attention and were expected to reach the validation stage by the 2020s, with commercial viability projected for the 2030s. However, these projections have been unrealized because of various factors. The development of reactor designs by the GIF was supposed to be influenced by events such as the 2008 global financial crisis, 2011 Fukushima accident [2, 3], discovery of extensive shale oil reserves in the United States, and overly ambitious technological targets. Consequently, the momentum for VHTR development reduced significantly. In this context, the aims of this study were to compare and analyze the development progress of the six Gen IV reactor designs over the past 20 years, based on the GIF roadmaps published in 2002 and 2014. The primary focus was to examine the prospects for the reactor designs in relation to spent nuclear fuel burning in conjunction with small modular reactor (SMR), including molten salt reactor (MSR), which is expected to have spent nuclear fuel management potential.

• Journal of Nuclear Fuel Cycle and Waste Technology(JNFCWT)
• /
• v.21 no.4
• /
• pp.489-502
• /
• 2023

This article presents the crucial role played by the French underground research laboratory (URL) in initiating the deep geological repository project Cigéo. In January 2023, Andra finalized the license application for the initial construction of Cigéo. Depending on Government's decision, the construction of Cigéo may be authorized around 2027. Cigéo is the result of a National program, launched in 1991, aiming to safely manage high-level and intermediate level long-lived radioactive wastes. This National program is based on four principles: 1) excellent science and technical knowledge, 2) safety and security as primary goals for waste management, 3) high requirements for environment protection, 4) transparent and open-public exchanges preceding the democratic decisions and orientations by the Parliament. The research and development (R&D) activities carried out in the URL supported the design and the safety demonstration of the Cigéo project. Moreover, running the URL has provided an opportunity to gain practical experience with regard to the security of underground operations, assessment of environmental impacts, and involvement of the public in the preparation of decisions. The practices implemented have helped gradually build confidence in the Cigéo project.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.81-83
• /
• 2024

본 논문에서는 mask R-CNN의 이미지 세그먼테이션(Image Segmentation) 기법을 이용하여 철도의 선로를 식별하고 분류하는 방법을 제안한다. mask R-CNN의 이미지 세그먼테이션은 바운딩 박스(Bounding Box)를 통해 이미지에서 객체를 식별하는 R-CNN 알고리즘과는 달리 픽셀 단위로 관심 있는 객체를 검출하고 분류하는 기법으로서 오브젝트 디텍션(Object Detection)보다 더욱 정교한 객체 식별이 가능하다. 본 연구에서는 Pascal VOC 형태의 고속철도 데이터 24,205셋의 데이터를 전처리하고 MS COCO 데이터셋으로 변환하여, MMDetection의 mask R-CNN을 통해 픽셀 단위로 철도선로를 식별하고 정상/불량 상태를 분류하는 연구를 수행하였다. 선행연구에서는 YOLO를 활용하여 Polygon형태의 좌표를 바운딩 박스로 분류하였는데, 본 연구에서는 mask R-CNN을 활용함으로써 철도 선로를 더욱 정교하게 식별하였으며 정상/불량의 상태 분류는 YOLO와 유사한 성능을 보였다.