Smart Badminton Detection System Based on Scaled-YOLOv4

In recent years, artificial intelligence applied to computer vision is growing popularly. Also, the smart sport system is now a flourishing research topic that more and more researchers are studying in this area. In this paper, we used scaled-YOLOv4 [1], which is now the state-of-the-art deep learning model on object detection, as our deep learning network to build a smart badminton detection system. We proposed the modified scaled-YOLOv4 by adding residual attention modules to detect the positions of badminton players and the shuttlecock and an image projection method to calculate the moving distance of the players.