http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
- 中文 을 입력하시려면 zhongwen을 입력하시고 space를누르시면됩니다.
- 北京 을 입력하시려면 beijing을 입력하시고 space를 누르시면 됩니다.
RISS 인기검색어
- 검색키워드
- 저자 : 김성빈(Cheng-Bin Jin) (검색결과 4 건)
- 무료
- 기관 내 무료
- 유료
-
Cheng-Bin Jin(김성빈),Trung Dung Do,Mingjie Liu,Hakil Kim(김학일) 제어로봇시스템학회 2018 제어·로봇·시스템학회 논문지 Vol.24 No.3
When we say a person is texting, can you tell the person is walking or sitting? Emphatically, no. In order to solve this incomplete representation problem, this paper presents a sub-action descriptor for detailed action detection. The sub-action descriptor consists of three levels: posture, locomotion, and gestures. The three levels provide three sub-action categories for a single action in order to address the representation problem. The proposed action detection model simultaneously localizes and recognizes the actions of multiple individuals in video surveillance using appearance-based temporal features with multi-convolutional neural networks. The proposed approach achieved a mean average precision of 76.6% for frame-based measurement and 83.5% for video-based measurement of the ICVL video surveillance dataset. Extensive experiments on the benchmark KTH dataset demonstrate that the proposed approach achieved better performance, which in turn improves action recognition performance in comparison to the stateof-the-art methods. The action detection model can run at around 25 fps with the ICVL dataset and at more than 80 fps with the KTH dataset, which is suitable for real-time surveillance applications.
-
CT-based MR Image Approximation using Cycle-Consistent Adversarial Networks
Cheng-Bin Jin(김성빈),Hakil Kim(김학일),Seongsu Joo(주성수),Eunsik Park(박은식),Young Saem Ahn(안영샘),In Ho Han(한인호),Jae Il Lee(이재일),Xuenan Cui(최학남) 대한전기학회 2019 대한전기학회 학술대회 논문집 Vol.2019 No.2
Computed tomography (CT) is widely used in various clinical applications. Magnetic resonance imaging (MRI) provides more anatomical details than CT for diagnostic purposes. However, the price of an MRI puts a heavy burden on low-income patients. This leads patients to undergo low-cost CT scans instead of MRIs, and this causes them to miss the opportunity for early diagnosis. To generate additional information and to increase the diagnostic value of CT, this paper proposes a method to approximate an MR image using a CT scan with the adversarial cycle-consistent networks. A novel objective function is introduced, consisting of adversarial loss, cycle-consistent loss, voxel-wise loss, gradient difference loss, and perceptual loss. Experimental results show that the proposed method significantly outperforms all baseline methods in all measurements, achieving the lowest mean absolute error and root mean square error and the highest peak-signal-to-noise-ratio, structural similarity, and Pearson correlation coefficient. This study can help the low-income patients, who cannot undergo MRI in clinical diagnosis, and patients in the developing countries where CT is the only diagnostic device.
-
위조지문 판별률 향상을 위한 학습데이터 혼합 증강 방법
김원진(Weonjin Kim),김성빈(Cheng-Bin Jin),유경송(Jinsong Liu),김학일(Hakil Kim) 한국정보보호학회 2017 정보보호학회논문지 Vol.27 No.2
최근 모바일 및 핀테크(fin-tech) 분야의 최신 트렌드로 지문인식, 홍채인식과 같은 생체인식을 통한 사용자 본인 인증이 주목 받고 있다. 특히 지문인식을 이용한 인증 방식은 전통적인 생체인식 방식으로써 사용자들이 사용하는데 발생하는 거부감이 다른 생체인식에 비해 현저히 낮아 현재 가장 보편적으로 이용되는 방식이다. 이와 동시에 지문을 이용한 인증 시 보안에 대한 중요성이 부각되어 지문의 위조 여부 판별의 중요성 또한 증가하고 있다. 본 논문에서는 CNN(Convolutional Neural Networks) 특징을 이용한 위조 여부 판별 방법에 있어 판별률을 향상시키기 위한 새로운 방법을 제시한다. 학습데이터에 영향을 많이 받는 CNN 특성 상 기존에는 판별률을 향상시키기 위해 아핀 변환(affine transformation) 또는 수평 반전(horizontal reflection)을 사용하여 학습데이터의 양을 증가 시키는 것이 일반적인 방법이었으나 본 논문에서는 위조지문 판별 난이도를 기반으로 한 효과적인 학습데이터 증강(data augmentation) 방법을 제시하며 실험을 통해 제안하는 방법의 타당성을 확인하였다. Recently, user authentication through biometric traits such as fingerprint and iris raise more and more attention especially in mobile commerce and fin-tech fields. In particular, commercialized authentication methods using fingerprint recognition are widely utilized mainly because customers are more adopted and used to fingerprint recognition applications. In the meantime, the security issues caused by fingerprint falsification bring lots of attention. In this paper, we propose a new method to improve the performance of fake fingerprint detection using CNN(Convolutional Neural Network). It is common practice to increase the amount of learning data by using affine transformation or horizontal reflection to improve the detection rate in CNN characteristics that are influenced by learning data. However, in this paper we propose an effective data augmentation method based on the database difficulty level. The experimental results confirm the validity of proposed method.
-
A Robust Direction-selectable Edge-based Panel Cutting-Line Detection Algorithm
Wei Li(이위),Hakil Kim(김학일),Cheng-Bin Jin(김성빈),Mingjie Ma(마명걸),Qiongxiu Li(이경수),Jong-Hee Kim(김종희),Xuenan Cui(최학남) 대한전자공학회 2016 대한전자공학회 학술대회 Vol.2016 No.6
This paper proposes three robust detection algorithms for locating the cutting line in an image captured by a panel-cutting system. All of the proposed methods contain two stages: edge detection and line fitting. A designed step-edge operator is applied to detect edges in the search direction of interest depending on the intensity concentration. Meanwhile, the proposed line-fitting algorithm is able to precisely fit a line by minimizing the summation of L1 distance from each detected edge point to the fitted line. As the result, all of the proposed methods achieve accuracy of more than 85%. Going one step further, full-scale edge detection (FSED) obtains the best performance at 99.05%, which is evaluated by using a variety of real-world images.
ScienceON
DBpia연관 검색어 추천
이 검색어로 많이 본 자료
활용도 높은 자료
