http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
최근 들어, 딥러닝 알고리즘을 사람의 얼굴의 감정인식에 사용하는 연구가 활발하게 진행 중이다. 딥러닝 알고리즘을 적용한 기존의 방법들은 감정인식 데이터를 합성곱 인공신경망과 완전히 연결된 레이어를 통해서 이미지의 특징을 추출하고 1차원 형태로 감정을 인식한다. 앞에서 언급한 연구들에서는 가려지지 않은 얼굴의 감정인식을 진행하였는데, 본 논문에서는 어텐션 모듈과 합성곱 인공신경망과 완전히 연결된 레이어를 사용하여 네트워크를 설계하여, 가려지지 않은 얼굴이나, 가려진 얼굴에서도 감정인식이 가능한 방법론을 제안한다. Recently, studies using deep learning algorithms for emotion recognition of human faces are actively underway. Existing methods using deep learning algorithms recognize emotions by extracting the feature of the image and transforming the emotion recognition data into a one-dimensional form through a layer completely connected to the artificial neural network. In the previous studies, emotion recognition of uncovered faces was performed. In this paper, the network is designed using the layers completely connected to the attention module and the artificial neural network, so that emotion recognition can be performed on uncovered or hidden faces. I suggest this possible methodology.
나후승 Graduate School, Yonsei University 2023 국내박사
Recently, fine dust was classified as a Group 1 carcinogen, so global interest in the danger it poses is increasing. Fine dust is more dangerous to the health of children whose respiratory systems are still developing than to healthy adults, so national efforts are being made to reduce fine dust levels in schools. The Ministry of Education implemented air quality management standards in the School Health Act of 2006, suggested countermeasures against high concentrations of fine dust in schools, and provided air purifiers for each classroom nationwide. Recently, academics and parents are increasingly calling for improvements to the overall air quality in schools, including not only fine dust but also carbon dioxide. In this study, I present a mechanical system that can improve schools’ air quality. This dissertation makes three major contributions. The first contribution is that it presents the efficacy of using air curtains to block the inflow of fine dust into schools both in the laboratory and in three schools over a long period of time. The second contribution is that it discusses air quality improvement methods for different places in schools. First, the airtightness of classrooms in schools was analyzed. Air quality improved in both general classrooms and special rooms. With regard to special classrooms, science classrooms’ air quality was affected by the use of chemicals, technology and housework management classrooms’ air quality was affected by fine dust generated during cooking, and food service classrooms’ air quality was affected by fine dust inflows due to doors to the outside always being open. The third contribution is that it presents a plan for how classroom equipment can be operated. It discusses why ventilation should be prioritized over using air purifiers in existing schools and how to improve air quality without excessive noise, which has become an issue recently. The effectiveness of this plan was tested in schools. This dissertation presents a plan for how to mechanically improve school air quality. Its key contents are how to block the inflow of fine dust, how to improve the air quality of different types of rooms in schools, and a plan for operating mechanical air quality improvement systems. I hope that this thesis will be used by schools across the country to improve their air quality for their students’ benefit.