http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
인공신경망 기반의 Decolorization 파라메터 자동 결정 방법
조용채(Yong Chae Cho),정우진(Woojin Jeong),한복규(Bok Gyu Han),양현석(Hyeon Seok Yang),심재준(Jae Jun Sim),남대현(Dae Hyun Nam),문영식(Young Shik Moon) 대한전자공학회 2018 대한전자공학회 학술대회 Vol.2018 No.6
In this paper, we propose a method to find optimal parameters for decolorization using convolutional neural networks. Decolorization is a technique for converting color images into gray images. The conventional decolorization methods have a problem that the parameter can not be determined automatically. The proposed method automatically finds optimal parameters with a deep neural network. In the experiment, the proposed method generates a visually pleasing gray image compared to the conventional method.
단계적 딥러닝 네트워크 학습 방법을 통한 3차원 관절 좌표 추정
조용채(Yong Chae Cho),한정훈(Jeong Hoon Han),이호경(Ho Gyeong Lee),문영식(Young Shik Moon) 대한전자공학회 2019 대한전자공학회 학술대회 Vol.2019 No.11
3D pose estimation is a study of estimating human 3D joints from a single image, and it is widely used in industrial fields and applications. The performance of 3D pose estimation has dramatically improved with the deep learning. However, the lack of 3D data has always been a constant problem. To solve this issue, we propose multi-stage learning method that uses both 2D and 3D datasets. We achieved 92.0% accuracy with Human3.6M dataset and obtained natural 3D pose results on outdoor images.
심재준(Jae Jun Sim),정우진(Woo Jin Jeong),양현석(Hyeon Seok Yang),한복규(Bok Gyu Han),조용채(Yong Chae Cho),이호경(Ho Gyeong Lee),문영식(Young Shik Moon) 대한전자공학회 2018 대한전자공학회 학술대회 Vol.2018 No.11
We propose an indoor haze removal method using MSCNN and cGAN. The structure of the network consists of multi-scale CNN and cGAN for photo realistic result. Our method outputs the haze removal image immediately, unlike the existing methods of estimating the depth map. Our method has a quantitative evaluation of 22.6879 in PSNR and 0.8872 in SSIM, which is higher than state of the art by 1.342 in PSNR and 0.0116 in SSIM. It also has good results in qualitative evaluation.
심재준(Jae Jun Sim),정우진(Woo Jin Jung),양현석(Hyeon Seok Yang),한복규(Bok Gyu Han),조용채(Yong Chae Cho),문영식(Young Shik Moon) 대한전자공학회 2018 전자공학회논문지 Vol.55 No.9
최근 신경망이 활발히 연구되어 다양한 분야에 적용되고 있으며, 영상처리의 다양한 분야(초해상도 복원, 영상 분류, 영상분할 등등)에서도 신경망을 도입하여 이전보다 나은 성과를 내고 있다. 본 논문에서는 의료영상에 깊은 신경망을 활용하여 세포핵 영역을 분할하는 기법을 제안한다. 본 논문에서 제안하는 네트워크 구조는 수용영역이 서로 다른 세 개의 네트워크를 병렬 처리하는 병렬 네트워크와 분류 네트워크로 이루어져 있다. 네트워크의 입력은 원본 영상을 전처리한 영상과 가이드 영상을 사용한다. 제안하는 방법은 풀링을 제거한 Deeplab-v1보다 mIOU가 4.61% 높고, 1024×1024 크기 영상에서 1.92배 빠르다. Recently, neural networks have been actively studied and applied in various fields. In the various fields of image processing (super resolution restoration, image classification, image segmentation, etc.), neural networks have been introduced to achieve better results than before. In this paper, we propose a technique to segment the nuclei region using deep neural network for medical images. The network structure used in this paper consists of three networks with different receptive field and a classification network. The input of the network is the pre - processed image and the guide image of the original image. The proposed method is 4.61% higher in mIOU than Deeplab-v1 with pooling removed and 1.92 times faster in 1024×1024 size image.
7차 슬기로운 생활 교과의 교육과정 운영실태와 개선방향
남경희,남호엽,조용채 서울교육대학교 2003 한국초등교육 Vol.14 No.2
Overall objectives of 'Life for Wisdom' are to help pupils have an interest in the interaction between oneself, and society around them and nature through concrete activities and personal experiences, think about oneself and one's own life. The purpose of this paper is to find a reform measure of the management of 'Life for Wisdom' in 7th National Curriculum, and the method of it is to inquiry the management through questionnaire. As findings on inquiry, Consideration should be given to the following items: 1. To intensify the principal of integrations and to raise the standard of integrations 2. To raise the quality of activity in the contents of a textbook. 3. To expand experiences and to select the optimal material in the lessons. 4. To value much of the process in the evaluations.