Towards the detection of driver–pedestrian eye contact

Non-verbal communication, such as eye contact between drivers and pedestrians, has been regarded one way to reduce accident risk. So far, studies have assumed rather than objectively measured the occurrence of contact. We address this research gap by developing an detection method testing it in indoor experiment with scripted driver–pedestrian interactions at a pedestrian crossing. Thirty participants acted either standing on imaginary curb or crossing one-lane road front stationary vehicle experimenter driver’s seat. In half trials, pedestrians were instructed make driver; other half, they prohibited from doing so. Both parties’ gaze was recorded using trackers. An in-vehicle stereo camera car’s point view, head-mounted pedestrian’s location eyes estimated image recognition. demonstrate that can be detected measuring angles vector joining eyes, instantaneous directions, respectively, identifying whether these fall below threshold 4°. achieved 100% correct classification trials involving those without contact, based duration. The proposed may useful for future into