Deteksi Garis Lapangan Sepakbola Robot dengan Metode Hough Transform dan SSD MobileneT V2

Abstract

When playing a wheeled soccer robot match properly, a robot must at least have some basic skills such as following the movement of the ball, dribbling, and kicking the ball towards the opponent's goal. To achieve this, the robot must be able to detect the four basic components in a soccer game, one of the most important things is detecting the field lines that are the boundaries. For this reason, a field line detection study was carried out in the image of a RoboCup Soccer SPL soccer game by implementing the Hough Transform method. The dataset used in this research is the Robocup SPL Instance Segmentation Dataset with 987 test images. The results of this study indicate that the Hough Transform method can perform line detection in RoboCup Soccer SPL images with implementation on the system. Line detection in the image of the RoboCup Soccer SPL soccer game shows a line detection accuracy value of 96%.