http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
AI 기반 실시간 온라인 랜덤 플레이 댄스 플랫폼 개발
김수빈(Soobin Kim),박지혜(Jihye Park),이민영(Minyoung Lee),문서영(Seoyeon Mun),이경미(Kyoung-Mi Lee) 한국디지털콘텐츠학회 2024 한국디지털콘텐츠학회논문지 Vol.25 No.3
With the recent increase in short-form videos, platforms that allow a user to upload his/her dance videos or watch others videos have become increasingly popular; however, direct participation is still limited. Hence, this study proposes an online random-play dance platform that allows participation from anywhere, anytime. A user accessing the proposed dance platform can stand in front of a mobile camera and dance to randomly selected music, whereby the platform detects the user’s body joints in real time and analyzes the dance motions to calculate their similarity with the reference dance motions of the randomly selected music. In real-time play, the front end sends the extracted joints to the server, and the platform offloads the front end by allowing the server to generate motion vectors to calculate the similarity. In addition, data caching was performed to enable the fast processing of rapidly changing data. As a result of fine-tuning a predefined gesture-similarity neural network, applying min–max normalization, and considering movements of only arms and legs (excluding the torso), motion-similarity error was reduced, thereby achieving better accuracy in computing dance-motion similarity.