지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학위논문
- 저자정보
- 지도교수
- 박혜영
- 발행연도
- 2018
- 저작권
- 경북대학교 논문은 저작권에 의해 보호받습니다.
이용수1
이 논문의 연구 히스토리 (2)
- 이 논문의 후속연구가 궁금하신가요?
- 연관 학술논문 또는 학술발표를 통해 보다 발전된 연구결과를 확인하실 수 있습니다.
- 이 논문의 연구 히스토리 확인하기
초록· 키워드
오류제보하기
최근 합성곱 신경망을 비롯한 심층 학습 기술의 발전으로 영상에서의 객체 인식의 성능이 월등히 향상되었다. 하지만 객체 인식은 영상에 포함된 다양한 변형과 인식 대상이 되는 객체의 다양성 등으로 여전히 정복하기 어려운 문제들이 남아있다. 특히 저해상도 영상에서의 객체 인식에 관한 연구는 아직 초기 단계로 만족할 만한 성능을 보이지 못하고 있다. 본 논문에서는 저해상도 영상에서의 객체 인식 성능을 향상시키기 위한 영상 개선 신경망을 제안하고 이로부터 획득한 영상을 합성곱 신경망 기반의 객체인식 모델의 학습 및 인식에 추가적으로 활용함으로써 해상도 변화에 강건한 객체 인식 방법을 제안한다. 제안하는방법의 효율성을 확인하기 위해 CIFAR-10 데이터베이스와 CIFAR-100 데이터베이스를 사용하여 저해상도 환경에서의 객체 인식 성능을 측정하였고, 제안하는 방법이 저해상도 객체 인식 성능을 향상시킴과 동시에 고해상도 객체 인식 성능도 안정적으로 유지하는 것을 확인하였다.
목차
1. 서론 ······························································································ 12. 관련 연구 ······················································································ 52.1 객체 인식 ···················································································· 52.2 초고해상도 기법 ·········································································· 92.3 저해상도 객체 인식을 위한 심층 신경망 ······································· 163. 제안하는 방법 ············································································· 183.1 제안하는 방법의 전체 구조 ························································· 183.2 영상 개선 신경망 ······································································· 203.3 객체 인식 신경망의 학습과 인식 ·················································· 254. 객체 인식 실험 결과 ····································································· 284.1 실험데이터 및 환경 구성 ···························································· 284.2 CIFAR-10 데이터베이스를 이용한 실험 결과 ······························· 324.3 CIFAR-100 데이터베이스를 이용한 실험 결과 ····························· 345. 결론 ··························································································· 36참고 문헌 ······················································································· 37영문 초록 ······················································································· 40
최근 본 자료
댓글(0)
0