ゴール型球技における人流解析AIを用いた選手移動軌跡の取得―バーチャル環境への実装に向けたゲーム開発統合ソフトウェアの利用―

抄録

This study aims to acquire the movement trajectories of players by using people flow analysis AI, taking futsal, a goal-type ball game, as an example. Furthermore, in order to implement it in a virtual environment, we reproduced a futsal match by importing the data on the players’ movement trajectories into game development integration software. The video footage from a futsal practice session was analyzed, and a framework was established to track the movement trajectories of ten players. In cases where player detection was interrupted, newly detected numbers were linked to the corresponding players to maintain continuous tracking. As a result, we were able to assign the same number to the same player for about 90% of the detected numbers. Next, we used perspective transformation technology to reproduce a futsal match in a 3D virtual environment. As a result, the sense of distance and positional relationships between the players were generally consistent.