Improved YOLOX-S-based video target detection method for football matches
In order to improve the level of football matches,give birth to new football tactics,identify the position of football superstar Lionel Messi and football,and provide a good foundation for further tracking,an improved YOLOX-S target detection method for football matches is proposed.The Pseudo-IoU metric is used to improve the preliminary screening mechanism of positive samples in YOLOX-S,and more standardized and accurate allocation rules are introduced into the YOLOX-S anchor free detection framework.In addition,Focal Loss is used in the loss function to balance the difficulty and easy samples.The experimental results show that the proposed model has better overall performance compared with the YOLOX-S model,with an average accuracy of 79.8％for soccer category,72.6％for Messi category,and 76.2％for the mean accuracy.
target detectionYOLOX-Sfootball eventsPseudo-IouFocal Loss
万方数据
维普
