基于深度学习的竞技体育运动表现视频分析研究进展

Review of Research Progress in Video Analysis of Competitive Sports Performance Based on Deep Learning

随着人工智能技术的发展,深度学习在体育领域的研究与应用也在逐渐增多。将人工智能技术应用于竞技体育运动视频分析,既是国内外理论研究关注的热点,也是促进竞技体育高质量发展的有效手段。通过Google Scholar、中国知网、Web of Science及SPORTDiscus数据库检索文献,并进行系统的整理与分析,回顾近10年来国内外基于深度学习的运动表现视频分析技术研究,包括深度学习算法、计算机视觉等在体育分析领域的应用现状,揭示深度学习在运动分析领域的潜在应用,探索深度学习作为分析体育运动的工具的潜力。目前在运动员及球跟踪、动作行为分析和表现评价方面,基于深度学习的分析方法已经表现出了良好的性能,未来拥有着巨大的发展潜力。

With the development of artificial intelligence technology(Al),the research and application of deep learning in the field of sports are also increasingly implemented.The AI application to the video analysis of competitive sports becomes not only a hot spot of theoretical research at home and abroad,but also an effective means to promote the high-quality development of competitive sports.In this paper,literature was retrieved through Google Scholar,CNKI,Web of Science and SPORTDiscus databases,and systematic collation and analysis were carried out to review the research on sports performance video analysis technology based on deep learning at home and abroad in recent years.It includes the application status of deep learning algorithms and computer vision in the field of sports analysis,reveals the potential application of deep learning in the field of sports analysis,explores the potential of deep learning as a tool for analyzing sports.At present,in the aspects of player and ball tracking,action behavior analysis and performance evaluation,the method based on deep learning has shown good performance,and has great potential for development in the future.