Current Result Document :
| 한글제목(Korean Title) |
심층신경망 특징을 이용한 이미지 검색에서의 중복 제거 |
| 영문제목(English Title) |
Deduplication of Retrieved Image Data Using Deep Network Features |
| 저자(Author) |
양희성
허찬
현창훈
박혜영
Heesung Yang
Chan Hur
Changhun Hyun
Hyeyoung Park
|
| 원문수록처(Citation) |
VOL 48 NO. 02 PP. 1518 ~ 1520 (2021. 12) |
한글내용
(Korean Abstract) |
In this paper, we propose a method for duplicating retrieved image data for image recommendation system. From the candidate images that are retrieved for a given queries, the proposed system finds duplicate images with no cognitive difference and make them into one cluster. By choosing a representative image from each clusters it is possible to avoid useless memory consumption and give users more compact retrieval result. The proposed system is in the form of two modules connected. First, the feature extraction module represents candidate images as cognitive features obtained by using a pre-trained deep network model. Second, hierarchical clustering is applied to the feature vectors in order to find clusters with duplicated images. Through computational experiments, we confirm that the proposed method has competitive performance compared with a well-designed image processing module: using several handcrafted filters. |
영문내용
(English Abstract) |
|
| 키워드(Keyword) |
|
| 파일첨부 |
PDF 다운로드
|
