摘要
足球是世界上最受欢迎的运动之一,大量的人通过电视或者互联网来观看足球比赛。一场足球比赛通常需要90分钟,但是一些观众可能只想观看这场比赛中的精彩场面来节省时间。为了解决这个问题,设计了一个基于深度学习的足球赛事视频动作识别系统,该系统可以识别足球比赛中红黄牌和射门等关键动作,这些动作可以作为足球比赛中的精彩片段。该系统利用了目标检测、人体姿态估计以及动作识别的方法来识别关键动作,以生成足球赛事视频的精彩片段,节省用户观看的时间。
A football game usually takes 90 minutes,but some viewers may just want to watch the highlights of the game to save time.In order to solve this problem,this paper designs a football match video action recognition system based on deep learning.The system can recognize key actions such as red/yellow cards and shooting in football matches.These actions can be used as highlights in football matches.The system uses object detection,human pose estimation and action recognition methods to recognize key actions,so as to generate the wonderful clips of football match videos and save the time of viewers to watch.
出处
《工业控制计算机》
2021年第7期99-100,103,共3页
Industrial Control Computer
关键词
足球赛事视频
深度学习
目标检测
姿态估计
动作识别
football match video
deep learning
object detection
pose estimation
action recognition
