Attention-aware Human-Machine Interface to Support Video Surveillance Task

摘要

There is a growing and pressing need to monitor dense, unconstrained pedestrian crowds and to detectsuspicious individuals, e.g., suicide bombers. However, this is a perceptually and cognitively challengingtask for human operators, because they need to track and inspect a constant stream of moving people, tomake real-time decisions, and also maintain a high level of vigilance. An attention-aware human-machineinterface (HMI) is proposed and prototyped, which monitors human operators’ attention allocation andsupports eye gaze based interaction to support operator surveillance task. The goals are to reduce theoperators’ cognitive load and fatigue, and speed up visual search, thus enabling the surveillance of densepedestrian streams with increased accuracy. A preliminary user study showed promising results forenhancing operators’ detection performance using the attention-aware HMI, especially while searching atthe high density crowds.