닫기
Loading..

Please wait....

학술대회 프로시딩

홈 홈 > 연구문헌 > 학술대회 프로시딩 > 한국통신학회 학술대회 > 한국통신학회 2019년도 하계종합학술발표회

한국통신학회 2019년도 하계종합학술발표회

Current Result Document : 46 / 758 이전건 이전건   다음건 다음건

한글제목(Korean Title) KCR-MobileNet: Korean Character Recognition with Depth Wise Separable Convolutions
영문제목(English Title) KCR-MobileNet: Korean Character Recognition with Depth Wise Separable Convolutions
저자(Author) Ik Hwan Jeon   Seung Heon Han   Soo Young Shin  
원문수록처(Citation) VOL 69 NO. 01 PP. 0092 ~ 0093 (2019. 06)
한글내용
(Korean Abstract)
영문내용
(English Abstract)
This paper presents an efficient deep network for Korean characters(Hangul) recognition based on depth wise separable convolutions, namely KCR-MobileNet. The architecture mainly consists of depth wise separable convolutions for efficiency in computation and the number of parameters. In the experiments, two Korean character data sets are used, each contain 2,350 typical Korean characters. We then show the strong performance of KCR-MobilNet compared to other popular models on Korean character recognition.
키워드(Keyword)   
원문 PDF 다운로드