메뉴 건너뛰기
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)
고객센터 ENG
주제분류

논문 기본 정보

저자정보
(한국과학기술원) (한국과학기술원) (한국과학기술원) (국립생태원) (국립생태원) (한국과학기술원)
저널정보
대한전자공학회 대한전자공학회 학술대회 2024년도 대한전자공학회 하계학술대회 논문집
오류 신고하기

피인용 0

검색

    초록·키워드

    Species classification is crucial for wildlife monitoring to develop effective conservation strategies. Traditional methods rely on manual review of camera trap images by experts, which is expensive and time-consuming. Consequently, there has been recent advancement in deep learning-based approaches for wildlife species classification. However, wildlife image datasets obtained from nature often exhibit a long-tailed distribution. This class imbalance severely degrades the performance of classes with fewer samples. In this study, we proposed the Loss-aware Logit Adjustment (LLA) method, which utilizes class-specific loss values from the previous epoch to enhance the accuracy across all classes. By identifying the difficulties faced by each class and adjusting the logit values, the model could enhance the learning stability and mitigate the performance disparities among classes. We expect this study to contribute to the automation and efficiency of wildlife monitoring.

    최근 본 자료 전체보기