메뉴 건너뛰기
.. 내서재 .. 알림
소속 기관/학교 인증
인증하면 논문, 학술자료 등을  무료로 열람할 수 있어요.
한국대학교, 누리자동차, 시립도서관 등 나의 기관을 확인해보세요
(국내 대학 90% 이상 구독 중)
로그인 회원가입 고객센터 ENG
주제분류

추천
검색
질문

논문 기본 정보

자료유형
학술대회자료
저자정보
신승미 (핸디소프트) 황종윤 (핸디소프트)
저널정보
대한전자공학회 대한전자공학회 학술대회 2021년도 대한전자공학회 추계학술대회 논문집
발행연도
2021.11
수록면
648 - 651 (4page)

이용수

표지
📌
연구주제
📖
연구배경
🔬
연구방법
🏆
연구결과
AI에게 요청하기
추천
검색
질문

초록· 키워드

오류제보하기
With the Safety e-Report system, we can mitigate everyday risks. As of September 2021, the cumulative number of reported incidents via the system has reached 7 million, which has been enough to be considered big data. Studies thus far have been majorly focusing on text contents rather than the images attached in these reports. In this paper, we propose a study that will categorize the attached images by the type of the reported risks. Also, we would like to prove a possible application of a deep learning model, Vision Transformer. In this experiment, we have collected images from major incidents that had been reported via the Safety e-Report as well as safety field images from AIHub. Then, we have classified images into different categories such as road danger, pedestrian danger, and facility damage risk. A vision Transformer model, which had been pre-trained with the ImageNet, has been applied to this classification process. The experiment proceeds to show the difference in classification accuracy based on the amount of transfer learning data and the performance by comparing the Vision Transformer with CNN models. As a result, we could identify a positive correlation between the amount of trained data and the classification accuracy of the Vision Transformer. In comparison with CNN models, the Vision Transformer achieves 98% accuracy, which is higher than the CNN models, with a relatively low training cost. We expect that the method of image classification by risk types with the Vision Transformer, which is proposed in this study, can be applied to other similar fields that are subject to public safety.

목차

Abstract
Ⅰ. 서론
Ⅱ. 본론
Ⅲ. 구현
Ⅳ. 결론 및 향후 연구 방향
참고문헌

참고문헌 (0)

참고문헌 신청

함께 읽어보면 좋을 논문

논문 유사도에 따라 DBpia 가 추천하는 논문입니다. 함께 보면 좋을 연관 논문을 확인해보세요!

이 논문의 저자 정보

이 논문과 함께 이용한 논문

최근 본 자료

전체보기

댓글(0)

0