• Journal of KIISE
• /
• v.44 no.12
• /
• pp.1282-1289
• /
• 2017

Since the study of deep convolutional neural network became prevalent, one of the important discoveries is that a feature map from a convolutional network can be extracted before going into the fully connected layer and can be used as a saliency map for object detection. Furthermore, the model can use features from each different layer for accurate object detection: the features from different layers can have different properties. As the model goes deeper, it has many latent skip connections and feature maps to elaborate object detection. Although there are many intermediate layers that we can use for semantic segmentation through skip connection, still the characteristics of each skip connection and the best skip connection for this task are uncertain. Therefore, in this study, we exhaustively research skip connections of state-of-the-art deep convolutional networks and investigate the characteristics of the features from each intermediate layer. In addition, this study would suggest how to use a recent deep neural network model for semantic segmentation and it would therefore become a cornerstone for later studies with the state-of-the-art network models.

• Journal of Digital Contents Society
• /
• v.18 no.1
• /
• pp.183-190
• /
• 2017

This paper proposes an audio event detection method using Deep Neural Networks (DNN). The proposed method applies Feed Forward Neural Network (FFNN) to generate output probabilities of twenty audio events for each frame. Mel scale filter bank (FBANK) features are extracted from each frame, and its five consecutive frames are combined as one vector which is the input feature of the FFNN. The output layer of FFNN produces audio event probabilities for each input feature vector. More than five consecutive frames of which event probability exceeds threshold are detected as an audio event. An audio event continues until the event is detected within one second. The proposed method achieves as 71.8% accuracy for 20 classes of the UrbanSound8K and the BBC Sound FX dataset.

• The Optical Journal
• /
• v.13 no.2 s.72
• /
• pp.49-55
• /
• 2001

레이저 용접은 매우 작은 점으로 집속된 높은 밀도의 에너지로 재료를 용융시키는 용접방법으로, 좁고 깊은 용접부를 얻을 수 있으며 출력만 충분하면 단 1회의 용접으로도 상당히 깊은 용접부를 쉽게 얻는다는 장점이 있다. 레이저 용접의 원리와 특징, 레이저 용접 공정 변수, 레이저 용접공정의 관리, 산업현장에서의 레이저 용접기술 적용 현황을 알아본다.

• Journal of Internet Computing and Services
• /
• v.20 no.4
• /
• pp.91-102
• /
• 2019

In autonomous driving systems, the ability to classify pedestrians in images captured by cameras is very important for pedestrian safety. In the past, after extracting features of pedestrians with HOG(Histogram of Oriented Gradients) or SIFT(Scale-Invariant Feature Transform), people classified them using SVM(Support Vector Machine). However, extracting pedestrian characteristics in such a handcrafted manner has many limitations. Therefore, this paper proposes a method to classify pedestrians reliably and effectively using CNN's(Convolutional Neural Network) deep features and transfer learning. We have experimented with both the fixed feature extractor and the fine-tuning methods, which are two representative transfer learning techniques. Particularly, in the fine-tuning method, we have added a new scheme, called M-Fine(Modified Fine-tuning), which divideslayers into transferred parts and non-transferred parts in three different sizes, and adjusts weights only for layers belonging to non-transferred parts. Experiments on INRIA Person data set with five CNN models(VGGNet, DenseNet, Inception V3, Xception, and MobileNet) showed that CNN's deep features perform better than handcrafted features such as HOG and SIFT, and that the accuracy of Xception (threshold = 0.5) isthe highest at 99.61%. MobileNet, which achieved similar performance to Xception and learned 80% fewer parameters, was the best in terms of efficiency. Among the three transfer learning schemes tested above, the performance of the fine-tuning method was the best. The performance of the M-Fine method was comparable to or slightly lower than that of the fine-tuningmethod, but higher than that of the fixed feature extractor method.

• 전기의세계
• /
• v.41 no.4
• /
• pp.7-12
• /
• 1992

전기 절연에 관한 학문과 기술은 전기 공학 분야가 태동한 시기부터 타분야와 더불어 진보되어온 깊은 역사를 갖고 있다. 그러나 그 내용면에서 다른 기술 분야와는 다른 독특한 특징을 갖고 있다. 이들 특징에 대한 정확한 이해와 아울러 산적한 문제들의 올바른 인식이 현재 당면한 문제 및 앞으로 다가올 문제를 해결하기 위해서 필요하다고 생각된다.

• Korean Journal of Ichthyology
• /
• v.25 no.1
• /
• pp.38-41
• /
• 2013

A deep-dwelling goby, Obliquogobius yamadai, was described as the first record from Korea based on three specimens (30.3~47.0 mm SL) collected from the southern sea off Jeju Island. The species is characterized by having I, 9 second dorsal fin rays, scaled lateral side of nape, smaller head with large eye, and asymmetrical caudal fin dorsoventrally as well as about nine yellowish bars on body side. A new Korean name, "Gip-eun-ba-da-no-ran-ddi-mang-dug", is proposed for the species.

• Journal of the Speleological Society of Korea
• /
• v.21 no.22
• /
• pp.57-66
• /
• 1990

동굴의 습도는 년중 거의가 일정한 것이 특징이다. 그리고 캄캄한 암흑의 세계라는 점도 특색의 하나이다. 대체로 동굴의 대기는 그 기류의 움직임이 매우 느린 관계로 동구부근에는 대기온도와 동벽의 온도가 크게 달리 나타나지만 차차 거의가 비슷하게 나타난다. 이 때문에 종유석속 깊은 지점의 온도는 석회암의 온도와 관계되고 있으며 그 온도는 대체로 지표의 년간 평균기온과 비슷하다.(중략)

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.27 no.4
• /
• pp.305-312
• /
• 2014

The current American Concrete Institute(ACI), Canadian Standard Associate(CSA) and CEB-FIP Model Code 2010 provisions on the shear strength of a simply supported deep beam suggest that deep beams should be designed using the strut-and-tie model. Although this is a useful methodology to design members in disturbed regions, the quality of the design is highly dependent on the truss model that designers create. However, Hong et al. derived the shear strength equations of reinforced concrete deep beams. This thesis investigates the validity of the current ACI, CSA and CEB-FIP code provisions on the shear strength of simply supported reinforced concrete deep beams by comparing them with the shear strength equations proposed by Hong et al. The comparison shows that all of these code provisions provide reasonable estimates on the shear strength of concrete deep beam members and the selection of an internal truss model plays an important role on the estimation of shear strength.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.2 no.4
• /
• pp.619-626
• /
• 1998

This paper presents the features for quantitative evaluation of medical images based on the technique of image processing. In consideration of advice from medical doctor, 1 derive three features seemed to be strongly correlated with the degree of disease's advance. From comparison between each feature and evaluation value by medical doctor, our research shows that three features can be a useful aids in quantitative evaluation of lung's disease on chest X-ray images.

• Proceedings of the Korea Contents Association Conference
• /
• 2017.05a
• /
• pp.443-444
• /
• 2017

본 연구는 윌리엄 포사이드(William Forsythe)의 "One Flat Thing Reproduced"를 바르테니에프 펀더멘탈(Bartenieff Fundamentals)을 통해 그의 무용수들이 가지는 특징적인 움직임 패턴을 분석하였다. 본 연구 목적은 윌리엄 포사이드의 대표적인 안무적 특징인 역동성, 강인함과 공간적 조화로움을 가능하게 하는 그의 무용수들이 가지는 특징적인 움직임 패턴을 알아보기 위함이다. 연구결과 윌리엄 포사이드는 호흡을 안무의 요소로 사용하였고, 상-하체분리패턴은 포사이드 움직임의 특징적인 요소로 관찰되었으며, 서포트(Support)가 있는 듀엣(Duet) 움직임에서는 신체반쪽연결패턴을 사용하는 것으로 나타났다. 이는 역동성, 강인함과 공간적 조화로움이 깊은 관련이 있을 것으로 보여 진다. 더 많은 사례를 통해 움직임 패턴을 연구한다면 에너지 표현에 직접 적용 가능한 패턴들을 구체적으로 밝혀내어 안무를 하는데 있어 기초자료로 활용될 수 있을 것으로 예상된다.