In this study, we have developed a real-time safety advice system on driving simulator using vision-based head pose estimation. Since our prototype real-time safety advice system uses wearable sensors to measure driver’s scanning behavior with head motion, it is necessary for driver to wear body-mounted devices when he/she use the system. In this paper, we propose a new evaluation system of driver’s skill using vision-based head pose estimation. By measuring driver’s head motion and estimating driver’s safety scanning behavior with camera, we can evaluate his/her driving skill and give safety driving advice more easily without any devices such as wearable sensors.
This paper proposes a method for detecting similar scenes in soccer videos. In soccer games, the pass and players' formations are the basic elements of soccer tactics analyses. Therefore, in this paper, we propose a detection method of similar scenes by utilizing the pass regions and players' formations in soccer videos.
The purpose of this study is to contribute to improve badminton player’s performance for analyzing data in features playing badminton games and moreover estimating trend using it. Player’s feature quantity consists of player position on the court and strokes using motion analysis with multi view camera. Player’s feature quantity dataset in chronological order is generated by the combination of player’s feature quantity. Player’s feature quantity dataset of amateur player are compared with dataset of professional player in men's badminton singles games using mining sequential patterns. We confirmed the difference between amateur player and professional player.