GAZE-BASED INTERACTION FOR REAL-TIME VIDEO SURVEILLANCE SYSTEMS

摘要

Users of real-time video surveillance systems have to deal with visually rich and dynamic input. Due to this, human computer interaction comprises specific interaction tasks like moving object selection in videos or the navigation through videos. Being basically a visual task, video analysis is performed best if the visual focus of interest can stay continuously on the video output. Therefore, we propose the use of gaze-based interaction techniques that utilize eye pointing for object selection. In this contribution, we focus on the potential of the gaze + key press interaction technique. Implementation of gaze + key press into the real-time video surveillance system NEST is reported. Three pilot studies on moving object selection using gaze + key press are shortly summarized. Evaluation results show that gaze + key press might be a fast and comfortable interaction alternative to mouse interaction.