상세정보
학술대회
Detecting In-Game Play Event in Live Esports Stream
- 발행일
- 202210
- 출처
- International Conference on Information and Communication Technology Convergence (ICTC) 2022, pp.1929-1931
- 협약과제
- 22IH1600, Game Now : e-스포츠 서비스를 위한 인공지능 기반 실시간 게임 분석 기술 개발, 이상광
- 초록
- Recently, the multiplayer online battle arena (MOBA) genre has been growing in popularity worldwide. Especially League of Legends (LoL) successfully hosts national leagues and world championships yearly with many worldwide fans. Following this popularity, there are attempts to make small-scale esports broadcasts to the public or amateurs in regional and internet broadcasts. However, there are many hurdles to providing sufficient broadcast services to the audience. Specifically, if the game does not provide application programming interfaces (APIs), it is hard to get instant information about in-game situations. In this paper, we propose an automatic in-game play event detection system that can get instant information without game APIs. The proposed system uses the machine learning classifier and analyzes the broadcast screens to detect in-game play events such as killing players and destroying buildings. Our experimental results show that the proposed system can automatically detect in-game play events in the live game streams.
- KSP 제안 키워드
- Application programming interface, Event detection, Intrusion detection system(IDS), League of legends, Machine learning classifiers, Multiplayer online battle arena, Small-scale, game play