With the establishment of the Japan Sports Agency, expectations are increasing for improving sports performance using ICT. In soccer in particular, there has been a noticeable shift from traditional qualitative coaching methods to quantitative coaching based on match data. However, the focus was primarily on player positioning, overlooking important aspects of ball detection, which are necessary for pass accuracy and goal prediction.
Therefore, in this research, we develop a ball detection technology using image processing and deep learning specialized for wide-angle images taken with a single video camera. Ultimately, we aim to realize automatic soccer ball tracking.
