http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
김지예,서승희,Kim, Jiye,Suh, Seunhee 한국패션비즈니스학회 2017 패션 비즈니스 Vol.21 No.4
Fashion film has become a significant communication medium in the $21^{st}$ century. Fashion film, that tells unique quality stories, is a means of communicating brand value emotionally. To examine characteristics of storytelling according to types of fashion film, this study categorizes fashion film and investigates storytelling elements in terms of content, form, and communication. For methodology, a literature review was conducted to examine the concept of storytelling and types of fashion film storytelling. Empirical research was conducted on 32 fashion films from 2007 to recent years of 2017. Results are as follow. First, theatrical fashion film, based on linear narrative structure and closed-ending, is characterized content-based storytelling. Second, avant-garde fashion film, based on unconventional and experimental cinematic techniques, is characterized form-based storytelling. Finally, documentary fashion film that minimizes distortion and reproduces reality of designers' originality, refers to interactive communication-based storytelling which using digital technologies.
딥러닝 기반 실시간 다중 물체 인식 모바일 어플리케이션
김지예(Jiye Kim) 한국정보기술학회 2020 Proceedings of KIIT Conference Vol.2020 No.10
최근 장면 분할 알고리즘 분야는 물체 분할에 대한 정확도 달성을 목적으로 진행되었고 실제로 정확도는 일정 수준 확보하였으나 일상 생활에 적용하여 사용자에게 편의를 제공하기엔 장비의 한계와 속도 등의 문제점이 여전히 존재한다. 본 연구에서는 모바일에서 26개의 기계 부품을 인식하여 사용자에게 픽셀 단위로 기계부품의 정확한 정보를 전달해주는 것이 목적이며 ShuffleNet V2과 TensorFlow Lite 기반으로 다중 물체 분할 결과를 9fps (Frame Per Second)으로 실시간으로 인식이 가능함을 확인하였다. In recent years, the field of semantic segmentation algorithm has been progressed for the purpose of achieving accuracy, and the accuracy is actually secured to a certain level. However, there are still problems such as limitations of equipment and speed to provide convenience to users by applying them to everyday life. In this study, the purpose of this study is to recognize 26 mechanical parts on a mobile and deliver mechanical parts information to the user. Based on ShuffleNet V2 and TensorFlow Lite, the multi-object segmentation result is recognized in real time at 9fps (Frame Per Second).